resnet_1.ipynb

{"nbformat":4,"nbformat_minor":0,"metadata":{"colab":{"provenance":[],"mount_file_id":"1sVxEeX_6dCboNk-7MDukxWr5zWki6utD","authorship_tag":"ABX9TyNIbztOG5vbvOCWWhSYMyvg"},"kernelspec":{"name":"python3","display_name":"Python 3"},"language_info":{"name":"python"},"accelerator":"GPU","gpuClass":"standard"},"cells":[{"cell_type":"code","execution_count":null,"metadata":{"colab":{"base_uri":"https://localhost:8080/"},"id":"VGPF4Eoi4rhu","executionInfo":{"status":"ok","timestamp":1682434604821,"user_tz":-540,"elapsed":5,"user":{"displayName":"박민수","userId":"07044323519487942912"}},"outputId":"de2d6b1a-074d-4844-84ca-91661e614c9c"},"outputs":[{"output_type":"stream","name":"stdout","text":["/content/drive/MyDrive\n"]}],"source":["%cd drive/MyDrive"]},{"cell_type":"code","source":["!git clone https://github.com/sebastianbk/finetuned-resnet50-keras.git"],"metadata":{"colab":{"base_uri":"https://localhost:8080/"},"id":"ES8bFyac4uvj","executionInfo":{"status":"ok","timestamp":1682434614407,"user_tz":-540,"elapsed":1844,"user":{"displayName":"박민수","userId":"07044323519487942912"}},"outputId":"1a622491-05c3-44f3-cbc5-f27dd59f31ef"},"execution_count":null,"outputs":[{"output_type":"stream","name":"stdout","text":["Cloning into 'finetuned-resnet50-keras'...\n","remote: Enumerating objects: 12, done.\u001b[K\n","remote: Total 12 (delta 0), reused 0 (delta 0), pack-reused 12\u001b[K\n","Unpacking objects: 100% (12/12), 2.71 KiB | 16.00 KiB/s, done.\n"]}]},{"cell_type":"code","source":["%cd drive/MyDrive/finetuned-resnet50-keras/"],"metadata":{"id":"ZVRXLCYq6rzG","colab":{"base_uri":"https://localhost:8080/"},"executionInfo":{"status":"ok","timestamp":1682584927732,"user_tz":-540,"elapsed":1583,"user":{"displayName":"박민수","userId":"07044323519487942912"}},"outputId":"e6c66fc6-ce16-445b-8c53-6b1be74a5b89"},"execution_count":null,"outputs":[{"output_type":"stream","name":"stdout","text":["[Errno 2] No such file or directory: 'drive/MyDrive/finetuned-resnet50-keras/'\n","/content/drive/MyDrive/finetuned-resnet50-keras\n"]}]},{"cell_type":"code","source":["!ls"],"metadata":{"colab":{"base_uri":"https://localhost:8080/"},"id":"JDHa5CXV_wI2","executionInfo":{"status":"ok","timestamp":1682469529294,"user_tz":-540,"elapsed":475,"user":{"displayName":"박민수","userId":"07044323519487942912"}},"outputId":"f6759563-ea57-47f0-d2f1-e4d43b9cfbf6"},"execution_count":null,"outputs":[{"output_type":"stream","name":"stdout","text":["README.md  resnet50_predict.py\tresnet50_train.py\n"]}]},{"cell_type":"code","source":[],"metadata":{"id":"34vIEBt8b8OG"},"execution_count":null,"outputs":[]},{"cell_type":"code","source":["!python resnet50_predict.py resnet50_final.h5 ./data/test/winter.jpg"],"metadata":{"colab":{"base_uri":"https://localhost:8080/"},"id":"upQ9gZjAb8Sx","executionInfo":{"status":"ok","timestamp":1682481678197,"user_tz":-540,"elapsed":17739,"user":{"displayName":"박민수","userId":"07044323519487942912"}},"outputId":"dc3b9f46-ae4a-44dd-b1a3-2813ae8f39df"},"execution_count":null,"outputs":[{"output_type":"stream","name":"stdout","text":["2023-04-26 04:01:01.399835: I tensorflow/core/platform/cpu_feature_guard.cc:182] This TensorFlow binary is optimized to use available CPU instructions in performance-critical operations.\n","To enable the following instructions: AVX2 FMA, in other operations, rebuild TensorFlow with the appropriate compiler flags.\n","2023-04-26 04:01:02.513933: W tensorflow/compiler/tf2tensorrt/utils/py_utils.cc:38] TF-TRT Warning: Could not find TensorRT\n","Loading model: resnet50_final.h5\n","2023-04-26 04:01:05.266415: I tensorflow/compiler/xla/stream_executor/cuda/cuda_gpu_executor.cc:996] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero. See more at https://github.com/torvalds/linux/blob/v6.0/Documentation/ABI/testing/sysfs-bus-pci#L344-L355\n","2023-04-26 04:01:05.297560: I tensorflow/compiler/xla/stream_executor/cuda/cuda_gpu_executor.cc:996] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero. See more at https://github.com/torvalds/linux/blob/v6.0/Documentation/ABI/testing/sysfs-bus-pci#L344-L355\n","2023-04-26 04:01:05.297855: I tensorflow/compiler/xla/stream_executor/cuda/cuda_gpu_executor.cc:996] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero. See more at https://github.com/torvalds/linux/blob/v6.0/Documentation/ABI/testing/sysfs-bus-pci#L344-L355\n","2023-04-26 04:01:05.298534: I tensorflow/compiler/xla/stream_executor/cuda/cuda_gpu_executor.cc:996] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero. See more at https://github.com/torvalds/linux/blob/v6.0/Documentation/ABI/testing/sysfs-bus-pci#L344-L355\n","2023-04-26 04:01:05.298798: I tensorflow/compiler/xla/stream_executor/cuda/cuda_gpu_executor.cc:996] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero. See more at https://github.com/torvalds/linux/blob/v6.0/Documentation/ABI/testing/sysfs-bus-pci#L344-L355\n","2023-04-26 04:01:05.298987: I tensorflow/compiler/xla/stream_executor/cuda/cuda_gpu_executor.cc:996] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero. See more at https://github.com/torvalds/linux/blob/v6.0/Documentation/ABI/testing/sysfs-bus-pci#L344-L355\n","2023-04-26 04:01:06.326146: I tensorflow/compiler/xla/stream_executor/cuda/cuda_gpu_executor.cc:996] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero. See more at https://github.com/torvalds/linux/blob/v6.0/Documentation/ABI/testing/sysfs-bus-pci#L344-L355\n","2023-04-26 04:01:06.326472: I tensorflow/compiler/xla/stream_executor/cuda/cuda_gpu_executor.cc:996] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero. See more at https://github.com/torvalds/linux/blob/v6.0/Documentation/ABI/testing/sysfs-bus-pci#L344-L355\n","2023-04-26 04:01:06.326732: I tensorflow/compiler/xla/stream_executor/cuda/cuda_gpu_executor.cc:996] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero. See more at https://github.com/torvalds/linux/blob/v6.0/Documentation/ABI/testing/sysfs-bus-pci#L344-L355\n","2023-04-26 04:01:06.326920: W tensorflow/core/common_runtime/gpu/gpu_bfc_allocator.cc:47] Overriding orig_value setting because the TF_FORCE_GPU_ALLOW_GROWTH environment variable is set. Original config value was 0.\n","2023-04-26 04:01:06.326975: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1635] Created device /job:localhost/replica:0/task:0/device:GPU:0 with 13678 MB memory:  -> device: 0, name: Tesla T4, pci bus id: 0000:00:04.0, compute capability: 7.5\n","Loaded in: 7.679355144500732\n","Generating predictions on image: ./data/test/winter.jpg\n","2023-04-26 04:01:14.682419: I tensorflow/compiler/xla/stream_executor/cuda/cuda_dnn.cc:424] Loaded cuDNN version 8700\n","1/1 [==============================] - 3s 3s/step\n","<class 'numpy.ndarray'>\n","[[0.22940652 0.23059711 0.23804356 0.30195278]]\n"]}]},{"cell_type":"code","source":["!python resnet50_predict.py resnet50_final.h5 ./data/test/fall.jpg"],"metadata":{"colab":{"base_uri":"https://localhost:8080/"},"id":"2aEsckTne-rA","executionInfo":{"status":"ok","timestamp":1682481730404,"user_tz":-540,"elapsed":11798,"user":{"displayName":"박민수","userId":"07044323519487942912"}},"outputId":"c91cdb13-37f9-457b-b687-9cd4c445a5f0"},"execution_count":null,"outputs":[{"output_type":"stream","name":"stdout","text":["2023-04-26 04:01:59.123915: I tensorflow/core/platform/cpu_feature_guard.cc:182] This TensorFlow binary is optimized to use available CPU instructions in performance-critical operations.\n","To enable the following instructions: AVX2 FMA, in other operations, rebuild TensorFlow with the appropriate compiler flags.\n","2023-04-26 04:02:00.003283: W tensorflow/compiler/tf2tensorrt/utils/py_utils.cc:38] TF-TRT Warning: Could not find TensorRT\n","Loading model: resnet50_final.h5\n","2023-04-26 04:02:01.720909: I tensorflow/compiler/xla/stream_executor/cuda/cuda_gpu_executor.cc:996] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero. See more at https://github.com/torvalds/linux/blob/v6.0/Documentation/ABI/testing/sysfs-bus-pci#L344-L355\n","2023-04-26 04:02:01.757616: I tensorflow/compiler/xla/stream_executor/cuda/cuda_gpu_executor.cc:996] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero. See more at https://github.com/torvalds/linux/blob/v6.0/Documentation/ABI/testing/sysfs-bus-pci#L344-L355\n","2023-04-26 04:02:01.757924: I tensorflow/compiler/xla/stream_executor/cuda/cuda_gpu_executor.cc:996] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero. See more at https://github.com/torvalds/linux/blob/v6.0/Documentation/ABI/testing/sysfs-bus-pci#L344-L355\n","2023-04-26 04:02:01.758637: I tensorflow/compiler/xla/stream_executor/cuda/cuda_gpu_executor.cc:996] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero. See more at https://github.com/torvalds/linux/blob/v6.0/Documentation/ABI/testing/sysfs-bus-pci#L344-L355\n","2023-04-26 04:02:01.758886: I tensorflow/compiler/xla/stream_executor/cuda/cuda_gpu_executor.cc:996] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero. See more at https://github.com/torvalds/linux/blob/v6.0/Documentation/ABI/testing/sysfs-bus-pci#L344-L355\n","2023-04-26 04:02:01.759075: I tensorflow/compiler/xla/stream_executor/cuda/cuda_gpu_executor.cc:996] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero. See more at https://github.com/torvalds/linux/blob/v6.0/Documentation/ABI/testing/sysfs-bus-pci#L344-L355\n","2023-04-26 04:02:02.887755: I tensorflow/compiler/xla/stream_executor/cuda/cuda_gpu_executor.cc:996] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero. See more at https://github.com/torvalds/linux/blob/v6.0/Documentation/ABI/testing/sysfs-bus-pci#L344-L355\n","2023-04-26 04:02:02.888177: I tensorflow/compiler/xla/stream_executor/cuda/cuda_gpu_executor.cc:996] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero. See more at https://github.com/torvalds/linux/blob/v6.0/Documentation/ABI/testing/sysfs-bus-pci#L344-L355\n","2023-04-26 04:02:02.888486: I tensorflow/compiler/xla/stream_executor/cuda/cuda_gpu_executor.cc:996] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero. See more at https://github.com/torvalds/linux/blob/v6.0/Documentation/ABI/testing/sysfs-bus-pci#L344-L355\n","2023-04-26 04:02:02.888733: W tensorflow/core/common_runtime/gpu/gpu_bfc_allocator.cc:47] Overriding orig_value setting because the TF_FORCE_GPU_ALLOW_GROWTH environment variable is set. Original config value was 0.\n","2023-04-26 04:02:02.888808: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1635] Created device /job:localhost/replica:0/task:0/device:GPU:0 with 13678 MB memory:  -> device: 0, name: Tesla T4, pci bus id: 0000:00:04.0, compute capability: 7.5\n","Loaded in: 3.2248013019561768\n","Generating predictions on image: ./data/test/fall.jpg\n","2023-04-26 04:02:07.040613: I tensorflow/compiler/xla/stream_executor/cuda/cuda_dnn.cc:424] Loaded cuDNN version 8700\n","1/1 [==============================] - 4s 4s/step\n","<class 'numpy.ndarray'>\n","[[0.23222922 0.22072385 0.25825357 0.2887933 ]]\n"]}]},{"cell_type":"code","source":["!python resnet50_predict.py resnet50_final.h5 ./data/test/summer.jpg"],"metadata":{"colab":{"base_uri":"https://localhost:8080/"},"id":"TRn9_D3AuUFJ","executionInfo":{"status":"ok","timestamp":1682481753989,"user_tz":-540,"elapsed":11740,"user":{"displayName":"박민수","userId":"07044323519487942912"}},"outputId":"307afeda-a44f-422a-9273-efeda6ce8e22"},"execution_count":null,"outputs":[{"output_type":"stream","name":"stdout","text":["2023-04-26 04:02:22.867957: I tensorflow/core/platform/cpu_feature_guard.cc:182] This TensorFlow binary is optimized to use available CPU instructions in performance-critical operations.\n","To enable the following instructions: AVX2 FMA, in other operations, rebuild TensorFlow with the appropriate compiler flags.\n","2023-04-26 04:02:23.756066: W tensorflow/compiler/tf2tensorrt/utils/py_utils.cc:38] TF-TRT Warning: Could not find TensorRT\n","Loading model: resnet50_final.h5\n","2023-04-26 04:02:25.650792: I tensorflow/compiler/xla/stream_executor/cuda/cuda_gpu_executor.cc:996] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero. See more at https://github.com/torvalds/linux/blob/v6.0/Documentation/ABI/testing/sysfs-bus-pci#L344-L355\n","2023-04-26 04:02:25.681384: I tensorflow/compiler/xla/stream_executor/cuda/cuda_gpu_executor.cc:996] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero. See more at https://github.com/torvalds/linux/blob/v6.0/Documentation/ABI/testing/sysfs-bus-pci#L344-L355\n","2023-04-26 04:02:25.681691: I tensorflow/compiler/xla/stream_executor/cuda/cuda_gpu_executor.cc:996] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero. See more at https://github.com/torvalds/linux/blob/v6.0/Documentation/ABI/testing/sysfs-bus-pci#L344-L355\n","2023-04-26 04:02:25.682423: I tensorflow/compiler/xla/stream_executor/cuda/cuda_gpu_executor.cc:996] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero. See more at https://github.com/torvalds/linux/blob/v6.0/Documentation/ABI/testing/sysfs-bus-pci#L344-L355\n","2023-04-26 04:02:25.682683: I tensorflow/compiler/xla/stream_executor/cuda/cuda_gpu_executor.cc:996] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero. See more at https://github.com/torvalds/linux/blob/v6.0/Documentation/ABI/testing/sysfs-bus-pci#L344-L355\n","2023-04-26 04:02:25.682893: I tensorflow/compiler/xla/stream_executor/cuda/cuda_gpu_executor.cc:996] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero. See more at https://github.com/torvalds/linux/blob/v6.0/Documentation/ABI/testing/sysfs-bus-pci#L344-L355\n","2023-04-26 04:02:26.755865: I tensorflow/compiler/xla/stream_executor/cuda/cuda_gpu_executor.cc:996] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero. See more at https://github.com/torvalds/linux/blob/v6.0/Documentation/ABI/testing/sysfs-bus-pci#L344-L355\n","2023-04-26 04:02:26.756273: I tensorflow/compiler/xla/stream_executor/cuda/cuda_gpu_executor.cc:996] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero. See more at https://github.com/torvalds/linux/blob/v6.0/Documentation/ABI/testing/sysfs-bus-pci#L344-L355\n","2023-04-26 04:02:26.756559: I tensorflow/compiler/xla/stream_executor/cuda/cuda_gpu_executor.cc:996] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero. See more at https://github.com/torvalds/linux/blob/v6.0/Documentation/ABI/testing/sysfs-bus-pci#L344-L355\n","2023-04-26 04:02:26.756789: W tensorflow/core/common_runtime/gpu/gpu_bfc_allocator.cc:47] Overriding orig_value setting because the TF_FORCE_GPU_ALLOW_GROWTH environment variable is set. Original config value was 0.\n","2023-04-26 04:02:26.756852: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1635] Created device /job:localhost/replica:0/task:0/device:GPU:0 with 13678 MB memory:  -> device: 0, name: Tesla T4, pci bus id: 0000:00:04.0, compute capability: 7.5\n","Loaded in: 3.771909713745117\n","Generating predictions on image: ./data/test/summer.jpg\n","2023-04-26 04:02:30.941611: I tensorflow/compiler/xla/stream_executor/cuda/cuda_dnn.cc:424] Loaded cuDNN version 8700\n","1/1 [==============================] - 3s 3s/step\n","<class 'numpy.ndarray'>\n","[[0.23492478 0.2352766  0.2533867  0.27641198]]\n"]}]},{"cell_type":"code","source":["!python resnet50_predict.py resnet50_final.h5 ./data/test/spring.jpg"],"metadata":{"colab":{"base_uri":"https://localhost:8080/"},"id":"ZmHdE3EquUM4","executionInfo":{"status":"ok","timestamp":1682481764837,"user_tz":-540,"elapsed":10879,"user":{"displayName":"박민수","userId":"07044323519487942912"}},"outputId":"1ec51ba0-dc38-49e1-b6d9-8e58d66a256c"},"execution_count":null,"outputs":[{"output_type":"stream","name":"stdout","text":["2023-04-26 04:02:34.352748: I tensorflow/core/platform/cpu_feature_guard.cc:182] This TensorFlow binary is optimized to use available CPU instructions in performance-critical operations.\n","To enable the following instructions: AVX2 FMA, in other operations, rebuild TensorFlow with the appropriate compiler flags.\n","2023-04-26 04:02:35.586630: W tensorflow/compiler/tf2tensorrt/utils/py_utils.cc:38] TF-TRT Warning: Could not find TensorRT\n","Loading model: resnet50_final.h5\n","2023-04-26 04:02:37.463143: I tensorflow/compiler/xla/stream_executor/cuda/cuda_gpu_executor.cc:996] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero. See more at https://github.com/torvalds/linux/blob/v6.0/Documentation/ABI/testing/sysfs-bus-pci#L344-L355\n","2023-04-26 04:02:37.492674: I tensorflow/compiler/xla/stream_executor/cuda/cuda_gpu_executor.cc:996] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero. See more at https://github.com/torvalds/linux/blob/v6.0/Documentation/ABI/testing/sysfs-bus-pci#L344-L355\n","2023-04-26 04:02:37.492950: I tensorflow/compiler/xla/stream_executor/cuda/cuda_gpu_executor.cc:996] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero. See more at https://github.com/torvalds/linux/blob/v6.0/Documentation/ABI/testing/sysfs-bus-pci#L344-L355\n","2023-04-26 04:02:37.493657: I tensorflow/compiler/xla/stream_executor/cuda/cuda_gpu_executor.cc:996] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero. See more at https://github.com/torvalds/linux/blob/v6.0/Documentation/ABI/testing/sysfs-bus-pci#L344-L355\n","2023-04-26 04:02:37.493893: I tensorflow/compiler/xla/stream_executor/cuda/cuda_gpu_executor.cc:996] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero. See more at https://github.com/torvalds/linux/blob/v6.0/Documentation/ABI/testing/sysfs-bus-pci#L344-L355\n","2023-04-26 04:02:37.494079: I tensorflow/compiler/xla/stream_executor/cuda/cuda_gpu_executor.cc:996] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero. See more at https://github.com/torvalds/linux/blob/v6.0/Documentation/ABI/testing/sysfs-bus-pci#L344-L355\n","2023-04-26 04:02:38.487512: I tensorflow/compiler/xla/stream_executor/cuda/cuda_gpu_executor.cc:996] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero. See more at https://github.com/torvalds/linux/blob/v6.0/Documentation/ABI/testing/sysfs-bus-pci#L344-L355\n","2023-04-26 04:02:38.487878: I tensorflow/compiler/xla/stream_executor/cuda/cuda_gpu_executor.cc:996] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero. See more at https://github.com/torvalds/linux/blob/v6.0/Documentation/ABI/testing/sysfs-bus-pci#L344-L355\n","2023-04-26 04:02:38.488075: I tensorflow/compiler/xla/stream_executor/cuda/cuda_gpu_executor.cc:996] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero. See more at https://github.com/torvalds/linux/blob/v6.0/Documentation/ABI/testing/sysfs-bus-pci#L344-L355\n","2023-04-26 04:02:38.488218: W tensorflow/core/common_runtime/gpu/gpu_bfc_allocator.cc:47] Overriding orig_value setting because the TF_FORCE_GPU_ALLOW_GROWTH environment variable is set. Original config value was 0.\n","2023-04-26 04:02:38.488260: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1635] Created device /job:localhost/replica:0/task:0/device:GPU:0 with 13678 MB memory:  -> device: 0, name: Tesla T4, pci bus id: 0000:00:04.0, compute capability: 7.5\n","Loaded in: 2.985593795776367\n","Generating predictions on image: ./data/test/spring.jpg\n","2023-04-26 04:02:42.102292: I tensorflow/compiler/xla/stream_executor/cuda/cuda_dnn.cc:424] Loaded cuDNN version 8700\n","1/1 [==============================] - 3s 3s/step\n","<class 'numpy.ndarray'>\n","[[0.23494479 0.22126475 0.25557446 0.28821602]]\n"]}]},{"cell_type":"code","source":[],"metadata":{"id":"3dnOOEAluUUb"},"execution_count":null,"outputs":[]},{"cell_type":"code","source":[],"metadata":{"id":"Guff8qrkuUa6"},"execution_count":null,"outputs":[]},{"cell_type":"code","source":["!python resnet50_train.py"],"metadata":{"colab":{"base_uri":"https://localhost:8080/"},"id":"REigyYu7b8Wg","executionInfo":{"status":"ok","timestamp":1682482476571,"user_tz":-540,"elapsed":161555,"user":{"displayName":"박민수","userId":"07044323519487942912"}},"outputId":"7609ae95-8119-47b9-c470-964de8048ecd"},"execution_count":null,"outputs":[{"output_type":"stream","name":"stdout","text":["2023-04-26 04:11:55.196137: I tensorflow/core/platform/cpu_feature_guard.cc:182] This TensorFlow binary is optimized to use available CPU instructions in performance-critical operations.\n","To enable the following instructions: AVX2 FMA, in other operations, rebuild TensorFlow with the appropriate compiler flags.\n","2023-04-26 04:11:56.159087: W tensorflow/compiler/tf2tensorrt/utils/py_utils.cc:38] TF-TRT Warning: Could not find TensorRT\n","2023-04-26 04:11:57.551491: I tensorflow/compiler/xla/stream_executor/cuda/cuda_gpu_executor.cc:996] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero. See more at https://github.com/torvalds/linux/blob/v6.0/Documentation/ABI/testing/sysfs-bus-pci#L344-L355\n","2023-04-26 04:11:57.583145: I tensorflow/compiler/xla/stream_executor/cuda/cuda_gpu_executor.cc:996] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero. See more at https://github.com/torvalds/linux/blob/v6.0/Documentation/ABI/testing/sysfs-bus-pci#L344-L355\n","2023-04-26 04:11:57.583417: I tensorflow/compiler/xla/stream_executor/cuda/cuda_gpu_executor.cc:996] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero. See more at https://github.com/torvalds/linux/blob/v6.0/Documentation/ABI/testing/sysfs-bus-pci#L344-L355\n","Found 213 images belonging to 4 classes.\n","Found 204 images belonging to 4 classes.\n","2023-04-26 04:11:57.776094: I tensorflow/compiler/xla/stream_executor/cuda/cuda_gpu_executor.cc:996] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero. See more at https://github.com/torvalds/linux/blob/v6.0/Documentation/ABI/testing/sysfs-bus-pci#L344-L355\n","2023-04-26 04:11:57.776426: I tensorflow/compiler/xla/stream_executor/cuda/cuda_gpu_executor.cc:996] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero. See more at https://github.com/torvalds/linux/blob/v6.0/Documentation/ABI/testing/sysfs-bus-pci#L344-L355\n","2023-04-26 04:11:57.776646: I tensorflow/compiler/xla/stream_executor/cuda/cuda_gpu_executor.cc:996] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero. See more at https://github.com/torvalds/linux/blob/v6.0/Documentation/ABI/testing/sysfs-bus-pci#L344-L355\n","2023-04-26 04:11:58.797386: I tensorflow/compiler/xla/stream_executor/cuda/cuda_gpu_executor.cc:996] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero. See more at https://github.com/torvalds/linux/blob/v6.0/Documentation/ABI/testing/sysfs-bus-pci#L344-L355\n","2023-04-26 04:11:58.797691: I tensorflow/compiler/xla/stream_executor/cuda/cuda_gpu_executor.cc:996] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero. See more at https://github.com/torvalds/linux/blob/v6.0/Documentation/ABI/testing/sysfs-bus-pci#L344-L355\n","2023-04-26 04:11:58.797896: I tensorflow/compiler/xla/stream_executor/cuda/cuda_gpu_executor.cc:996] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero. See more at https://github.com/torvalds/linux/blob/v6.0/Documentation/ABI/testing/sysfs-bus-pci#L344-L355\n","2023-04-26 04:11:58.798070: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1635] Created device /job:localhost/replica:0/task:0/device:GPU:0 with 13678 MB memory:  -> device: 0, name: Tesla T4, pci bus id: 0000:00:04.0, compute capability: 7.5\n","/usr/local/lib/python3.9/dist-packages/keras/optimizers/legacy/adam.py:117: UserWarning: The `lr` argument is deprecated, use `learning_rate` instead.\n","  super().__init__(name, **kwargs)\n","/content/drive/MyDrive/finetuned-resnet50-keras/resnet50_train.py:73: UserWarning: `Model.fit_generator` is deprecated and will be removed in a future version. Please use `Model.fit`, which supports generators.\n","  finetuned_model.fit_generator(batches, steps_per_epoch=num_train_steps, epochs=1000, callbacks=[early_stopping, checkpointer], validation_data=val_batches, validation_steps=num_valid_steps)\n","2023-04-26 04:12:01.087036: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'Placeholder/_0' with dtype int32\n","\t [[{{node Placeholder/_0}}]]\n","Epoch 1/1000\n","2023-04-26 04:12:06.342160: I tensorflow/compiler/xla/stream_executor/cuda/cuda_dnn.cc:424] Loaded cuDNN version 8700\n","1/1 [==============================] - ETA: 0s - loss: 1.3846 - accuracy: 0.26562023-04-26 04:12:37.590282: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'Placeholder/_0' with dtype int32\n","\t [[{{node Placeholder/_0}}]]\n","\n","Epoch 1: val_loss improved from inf to 1.38456, saving model to resnet50_best.h5\n","1/1 [==============================] - 54s 54s/step - loss: 1.3846 - accuracy: 0.2656 - val_loss: 1.3846 - val_accuracy: 0.2891\n","Epoch 2/1000\n","1/1 [==============================] - ETA: 0s - loss: 1.3844 - accuracy: 0.2969\n","Epoch 2: val_loss did not improve from 1.38456\n","1/1 [==============================] - 2s 2s/step - loss: 1.3844 - accuracy: 0.2969 - val_loss: 1.3865 - val_accuracy: 0.2891\n","Epoch 3/1000\n","1/1 [==============================] - ETA: 0s - loss: 1.3851 - accuracy: 0.2344\n","Epoch 3: val_loss improved from 1.38456 to 1.38387, saving model to resnet50_best.h5\n","1/1 [==============================] - 3s 3s/step - loss: 1.3851 - accuracy: 0.2344 - val_loss: 1.3839 - val_accuracy: 0.3281\n","Epoch 4/1000\n","1/1 [==============================] - ETA: 0s - loss: 1.3845 - accuracy: 0.3125\n","Epoch 4: val_loss did not improve from 1.38387\n","1/1 [==============================] - 2s 2s/step - loss: 1.3845 - accuracy: 0.3125 - val_loss: 1.3855 - val_accuracy: 0.2188\n","Epoch 5/1000\n","1/1 [==============================] - ETA: 0s - loss: 1.3868 - accuracy: 0.2353\n","Epoch 5: val_loss did not improve from 1.38387\n","1/1 [==============================] - 4s 4s/step - loss: 1.3868 - accuracy: 0.2353 - val_loss: 1.3863 - val_accuracy: 0.2500\n","Epoch 6/1000\n","1/1 [==============================] - ETA: 0s - loss: 1.3857 - accuracy: 0.2500\n","Epoch 6: val_loss did not improve from 1.38387\n","1/1 [==============================] - 2s 2s/step - loss: 1.3857 - accuracy: 0.2500 - val_loss: 1.3852 - val_accuracy: 0.2422\n","Epoch 7/1000\n","1/1 [==============================] - ETA: 0s - loss: 1.3852 - accuracy: 0.2353\n","Epoch 7: val_loss did not improve from 1.38387\n","1/1 [==============================] - 2s 2s/step - loss: 1.3852 - accuracy: 0.2353 - val_loss: 1.3873 - val_accuracy: 0.2344\n","Epoch 8/1000\n","1/1 [==============================] - ETA: 0s - loss: 1.3828 - accuracy: 0.3059\n","Epoch 8: val_loss did not improve from 1.38387\n","1/1 [==============================] - 2s 2s/step - loss: 1.3828 - accuracy: 0.3059 - val_loss: 1.3842 - val_accuracy: 0.2891\n","Epoch 9/1000\n","1/1 [==============================] - ETA: 0s - loss: 1.3856 - accuracy: 0.2656\n","Epoch 9: val_loss did not improve from 1.38387\n","1/1 [==============================] - 2s 2s/step - loss: 1.3856 - accuracy: 0.2656 - val_loss: 1.3865 - val_accuracy: 0.2891\n","Epoch 10/1000\n","1/1 [==============================] - ETA: 0s - loss: 1.3853 - accuracy: 0.2824\n","Epoch 10: val_loss did not improve from 1.38387\n","1/1 [==============================] - 2s 2s/step - loss: 1.3853 - accuracy: 0.2824 - val_loss: 1.3866 - val_accuracy: 0.2891\n","Epoch 11/1000\n","1/1 [==============================] - ETA: 0s - loss: 1.3852 - accuracy: 0.2941\n","Epoch 11: val_loss did not improve from 1.38387\n","1/1 [==============================] - 2s 2s/step - loss: 1.3852 - accuracy: 0.2941 - val_loss: 1.3852 - val_accuracy: 0.2656\n","Epoch 12/1000\n","1/1 [==============================] - ETA: 0s - loss: 1.3873 - accuracy: 0.2000\n","Epoch 12: val_loss did not improve from 1.38387\n","1/1 [==============================] - 2s 2s/step - loss: 1.3873 - accuracy: 0.2000 - val_loss: 1.3849 - val_accuracy: 0.3125\n","Epoch 13/1000\n","1/1 [==============================] - ETA: 0s - loss: 1.3848 - accuracy: 0.3125\n","Epoch 13: val_loss did not improve from 1.38387\n","1/1 [==============================] - 2s 2s/step - loss: 1.3848 - accuracy: 0.3125 - val_loss: 1.3857 - val_accuracy: 0.2344\n","Epoch 14/1000\n","1/1 [==============================] - ETA: 0s - loss: 1.3841 - accuracy: 0.3203\n","Epoch 14: val_loss did not improve from 1.38387\n","1/1 [==============================] - 2s 2s/step - loss: 1.3841 - accuracy: 0.3203 - val_loss: 1.3877 - val_accuracy: 0.2734\n","Epoch 15/1000\n","1/1 [==============================] - ETA: 0s - loss: 1.3863 - accuracy: 0.2734\n","Epoch 15: val_loss did not improve from 1.38387\n","1/1 [==============================] - 2s 2s/step - loss: 1.3863 - accuracy: 0.2734 - val_loss: 1.3854 - val_accuracy: 0.2812\n","Epoch 16/1000\n","1/1 [==============================] - ETA: 0s - loss: 1.3846 - accuracy: 0.2969\n","Epoch 16: val_loss did not improve from 1.38387\n","1/1 [==============================] - 2s 2s/step - loss: 1.3846 - accuracy: 0.2969 - val_loss: 1.3858 - val_accuracy: 0.3281\n","Epoch 17/1000\n","1/1 [==============================] - ETA: 0s - loss: 1.3850 - accuracy: 0.3059\n","Epoch 17: val_loss did not improve from 1.38387\n","1/1 [==============================] - 2s 2s/step - loss: 1.3850 - accuracy: 0.3059 - val_loss: 1.3867 - val_accuracy: 0.2266\n","Epoch 18/1000\n","1/1 [==============================] - ETA: 0s - loss: 1.3854 - accuracy: 0.3059\n","Epoch 18: val_loss improved from 1.38387 to 1.38368, saving model to resnet50_best.h5\n","1/1 [==============================] - 4s 4s/step - loss: 1.3854 - accuracy: 0.3059 - val_loss: 1.3837 - val_accuracy: 0.2734\n","Epoch 19/1000\n","1/1 [==============================] - ETA: 0s - loss: 1.3847 - accuracy: 0.2734\n","Epoch 19: val_loss did not improve from 1.38368\n","1/1 [==============================] - 2s 2s/step - loss: 1.3847 - accuracy: 0.2734 - val_loss: 1.3863 - val_accuracy: 0.2656\n","Epoch 20/1000\n","1/1 [==============================] - ETA: 0s - loss: 1.3842 - accuracy: 0.2824\n","Epoch 20: val_loss did not improve from 1.38368\n","1/1 [==============================] - 2s 2s/step - loss: 1.3842 - accuracy: 0.2824 - val_loss: 1.3890 - val_accuracy: 0.2031\n","Epoch 21/1000\n","1/1 [==============================] - ETA: 0s - loss: 1.3851 - accuracy: 0.2588\n","Epoch 21: val_loss did not improve from 1.38368\n","1/1 [==============================] - 2s 2s/step - loss: 1.3851 - accuracy: 0.2588 - val_loss: 1.3850 - val_accuracy: 0.2969\n","Epoch 22/1000\n","1/1 [==============================] - ETA: 0s - loss: 1.3863 - accuracy: 0.2578\n","Epoch 22: val_loss did not improve from 1.38368\n","1/1 [==============================] - 2s 2s/step - loss: 1.3863 - accuracy: 0.2578 - val_loss: 1.3853 - val_accuracy: 0.2891\n","Epoch 23/1000\n","1/1 [==============================] - ETA: 0s - loss: 1.3859 - accuracy: 0.2656\n","Epoch 23: val_loss did not improve from 1.38368\n","1/1 [==============================] - 2s 2s/step - loss: 1.3859 - accuracy: 0.2656 - val_loss: 1.3866 - val_accuracy: 0.2656\n","Epoch 24/1000\n","1/1 [==============================] - ETA: 0s - loss: 1.3847 - accuracy: 0.2706\n","Epoch 24: val_loss did not improve from 1.38368\n","1/1 [==============================] - 2s 2s/step - loss: 1.3847 - accuracy: 0.2706 - val_loss: 1.3867 - val_accuracy: 0.2344\n","Epoch 25/1000\n","1/1 [==============================] - ETA: 0s - loss: 1.3830 - accuracy: 0.2706\n","Epoch 25: val_loss improved from 1.38368 to 1.38331, saving model to resnet50_best.h5\n","1/1 [==============================] - 3s 3s/step - loss: 1.3830 - accuracy: 0.2706 - val_loss: 1.3833 - val_accuracy: 0.3047\n","Epoch 26/1000\n","1/1 [==============================] - ETA: 0s - loss: 1.3829 - accuracy: 0.2824\n","Epoch 26: val_loss did not improve from 1.38331\n","1/1 [==============================] - 2s 2s/step - loss: 1.3829 - accuracy: 0.2824 - val_loss: 1.3836 - val_accuracy: 0.3281\n","Epoch 27/1000\n","1/1 [==============================] - ETA: 0s - loss: 1.3844 - accuracy: 0.2353\n","Epoch 27: val_loss did not improve from 1.38331\n","1/1 [==============================] - 2s 2s/step - loss: 1.3844 - accuracy: 0.2353 - val_loss: 1.3858 - val_accuracy: 0.2266\n","Epoch 28/1000\n","1/1 [==============================] - ETA: 0s - loss: 1.3854 - accuracy: 0.2000\n","Epoch 28: val_loss did not improve from 1.38331\n","1/1 [==============================] - 2s 2s/step - loss: 1.3854 - accuracy: 0.2000 - val_loss: 1.3860 - val_accuracy: 0.2500\n","Epoch 29/1000\n","1/1 [==============================] - ETA: 0s - loss: 1.3827 - accuracy: 0.3529\n","Epoch 29: val_loss did not improve from 1.38331\n","1/1 [==============================] - 2s 2s/step - loss: 1.3827 - accuracy: 0.3529 - val_loss: 1.3868 - val_accuracy: 0.2422\n","Epoch 30/1000\n","1/1 [==============================] - ETA: 0s - loss: 1.3848 - accuracy: 0.2734\n","Epoch 30: val_loss did not improve from 1.38331\n","1/1 [==============================] - 2s 2s/step - loss: 1.3848 - accuracy: 0.2734 - val_loss: 1.3856 - val_accuracy: 0.2734\n","Epoch 31/1000\n","1/1 [==============================] - ETA: 0s - loss: 1.3848 - accuracy: 0.2578\n","Epoch 31: val_loss did not improve from 1.38331\n","1/1 [==============================] - 2s 2s/step - loss: 1.3848 - accuracy: 0.2578 - val_loss: 1.3846 - val_accuracy: 0.3281\n","Epoch 32/1000\n","1/1 [==============================] - ETA: 0s - loss: 1.3839 - accuracy: 0.2941\n","Epoch 32: val_loss did not improve from 1.38331\n","1/1 [==============================] - 2s 2s/step - loss: 1.3839 - accuracy: 0.2941 - val_loss: 1.3841 - val_accuracy: 0.3125\n","Epoch 33/1000\n","1/1 [==============================] - ETA: 0s - loss: 1.3846 - accuracy: 0.2824\n","Epoch 33: val_loss did not improve from 1.38331\n","1/1 [==============================] - 2s 2s/step - loss: 1.3846 - accuracy: 0.2824 - val_loss: 1.3867 - val_accuracy: 0.2266\n","Epoch 34/1000\n","1/1 [==============================] - ETA: 0s - loss: 1.3844 - accuracy: 0.3059\n","Epoch 34: val_loss did not improve from 1.38331\n","1/1 [==============================] - 2s 2s/step - loss: 1.3844 - accuracy: 0.3059 - val_loss: 1.3850 - val_accuracy: 0.3047\n","Epoch 35/1000\n","1/1 [==============================] - ETA: 0s - loss: 1.3828 - accuracy: 0.3516\n","Epoch 35: val_loss did not improve from 1.38331\n","1/1 [==============================] - 2s 2s/step - loss: 1.3828 - accuracy: 0.3516 - val_loss: 1.3865 - val_accuracy: 0.2500\n","Epoch 36/1000\n","1/1 [==============================] - ETA: 0s - loss: 1.3861 - accuracy: 0.2969\n","Epoch 36: val_loss did not improve from 1.38331\n","1/1 [==============================] - 2s 2s/step - loss: 1.3861 - accuracy: 0.2969 - val_loss: 1.3853 - val_accuracy: 0.2578\n","Epoch 37/1000\n","1/1 [==============================] - ETA: 0s - loss: 1.3823 - accuracy: 0.3412\n","Epoch 37: val_loss did not improve from 1.38331\n","1/1 [==============================] - 2s 2s/step - loss: 1.3823 - accuracy: 0.3412 - val_loss: 1.3854 - val_accuracy: 0.2500\n","Epoch 38/1000\n","1/1 [==============================] - ETA: 0s - loss: 1.3848 - accuracy: 0.2891\n","Epoch 38: val_loss did not improve from 1.38331\n","1/1 [==============================] - 2s 2s/step - loss: 1.3848 - accuracy: 0.2891 - val_loss: 1.3869 - val_accuracy: 0.3281\n","Epoch 39/1000\n","1/1 [==============================] - ETA: 0s - loss: 1.3853 - accuracy: 0.2734\n","Epoch 39: val_loss did not improve from 1.38331\n","1/1 [==============================] - 2s 2s/step - loss: 1.3853 - accuracy: 0.2734 - val_loss: 1.3858 - val_accuracy: 0.3047\n","Epoch 40/1000\n","1/1 [==============================] - ETA: 0s - loss: 1.3844 - accuracy: 0.3125\n","Epoch 40: val_loss did not improve from 1.38331\n","1/1 [==============================] - 2s 2s/step - loss: 1.3844 - accuracy: 0.3125 - val_loss: 1.3855 - val_accuracy: 0.2969\n","Epoch 41/1000\n","1/1 [==============================] - ETA: 0s - loss: 1.3841 - accuracy: 0.3125\n","Epoch 41: val_loss improved from 1.38331 to 1.38322, saving model to resnet50_best.h5\n","1/1 [==============================] - 4s 4s/step - loss: 1.3841 - accuracy: 0.3125 - val_loss: 1.3832 - val_accuracy: 0.2891\n","Epoch 42/1000\n","1/1 [==============================] - ETA: 0s - loss: 1.3863 - accuracy: 0.2118\n","Epoch 42: val_loss improved from 1.38322 to 1.38319, saving model to resnet50_best.h5\n","1/1 [==============================] - 3s 3s/step - loss: 1.3863 - accuracy: 0.2118 - val_loss: 1.3832 - val_accuracy: 0.2656\n","Traceback (most recent call last):\n","  File \"/content/drive/MyDrive/finetuned-resnet50-keras/resnet50_train.py\", line 73, in <module>\n","    finetuned_model.fit_generator(batches, steps_per_epoch=num_train_steps, epochs=1000, callbacks=[early_stopping, checkpointer], validation_data=val_batches, validation_steps=num_valid_steps)\n","  File \"/usr/local/lib/python3.9/dist-packages/keras/engine/training.py\", line 2636, in fit_generator\n","    return self.fit(\n","  File \"/usr/local/lib/python3.9/dist-packages/keras/utils/traceback_utils.py\", line 65, in error_handler\n","    return fn(*args, **kwargs)\n","  File \"/usr/local/lib/python3.9/dist-packages/keras/engine/training.py\", line 1673, in fit\n","    self.reset_metrics()\n","KeyboardInterrupt\n","^C\n"]}]},{"cell_type":"code","source":[],"metadata":{"id":"vUpypkeRb8aT"},"execution_count":null,"outputs":[]},{"cell_type":"code","source":["!python resnet50_predict.py resnet50_best2.h5 ./data/test/spring.jpg"],"metadata":{"colab":{"base_uri":"https://localhost:8080/"},"id":"8tYbg0lXxZQo","executionInfo":{"status":"ok","timestamp":1682482538167,"user_tz":-540,"elapsed":13560,"user":{"displayName":"박민수","userId":"07044323519487942912"}},"outputId":"b04772f9-8ae0-4a79-c6d6-9ef797eb7b17"},"execution_count":null,"outputs":[{"output_type":"stream","name":"stdout","text":["2023-04-26 04:15:25.356670: I tensorflow/core/platform/cpu_feature_guard.cc:182] This TensorFlow binary is optimized to use available CPU instructions in performance-critical operations.\n","To enable the following instructions: AVX2 FMA, in other operations, rebuild TensorFlow with the appropriate compiler flags.\n","2023-04-26 04:15:26.173364: W tensorflow/compiler/tf2tensorrt/utils/py_utils.cc:38] TF-TRT Warning: Could not find TensorRT\n","Loading model: resnet50_best2.h5\n","2023-04-26 04:15:29.537513: I tensorflow/compiler/xla/stream_executor/cuda/cuda_gpu_executor.cc:996] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero. See more at https://github.com/torvalds/linux/blob/v6.0/Documentation/ABI/testing/sysfs-bus-pci#L344-L355\n","2023-04-26 04:15:29.580826: I tensorflow/compiler/xla/stream_executor/cuda/cuda_gpu_executor.cc:996] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero. See more at https://github.com/torvalds/linux/blob/v6.0/Documentation/ABI/testing/sysfs-bus-pci#L344-L355\n","2023-04-26 04:15:29.581162: I tensorflow/compiler/xla/stream_executor/cuda/cuda_gpu_executor.cc:996] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero. See more at https://github.com/torvalds/linux/blob/v6.0/Documentation/ABI/testing/sysfs-bus-pci#L344-L355\n","2023-04-26 04:15:29.582067: I tensorflow/compiler/xla/stream_executor/cuda/cuda_gpu_executor.cc:996] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero. See more at https://github.com/torvalds/linux/blob/v6.0/Documentation/ABI/testing/sysfs-bus-pci#L344-L355\n","2023-04-26 04:15:29.582356: I tensorflow/compiler/xla/stream_executor/cuda/cuda_gpu_executor.cc:996] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero. See more at https://github.com/torvalds/linux/blob/v6.0/Documentation/ABI/testing/sysfs-bus-pci#L344-L355\n","2023-04-26 04:15:29.582580: I tensorflow/compiler/xla/stream_executor/cuda/cuda_gpu_executor.cc:996] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero. See more at https://github.com/torvalds/linux/blob/v6.0/Documentation/ABI/testing/sysfs-bus-pci#L344-L355\n","2023-04-26 04:15:30.625755: I tensorflow/compiler/xla/stream_executor/cuda/cuda_gpu_executor.cc:996] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero. See more at https://github.com/torvalds/linux/blob/v6.0/Documentation/ABI/testing/sysfs-bus-pci#L344-L355\n","2023-04-26 04:15:30.626046: I tensorflow/compiler/xla/stream_executor/cuda/cuda_gpu_executor.cc:996] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero. See more at https://github.com/torvalds/linux/blob/v6.0/Documentation/ABI/testing/sysfs-bus-pci#L344-L355\n","2023-04-26 04:15:30.626278: I tensorflow/compiler/xla/stream_executor/cuda/cuda_gpu_executor.cc:996] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero. See more at https://github.com/torvalds/linux/blob/v6.0/Documentation/ABI/testing/sysfs-bus-pci#L344-L355\n","2023-04-26 04:15:30.626444: W tensorflow/core/common_runtime/gpu/gpu_bfc_allocator.cc:47] Overriding orig_value setting because the TF_FORCE_GPU_ALLOW_GROWTH environment variable is set. Original config value was 0.\n","2023-04-26 04:15:30.626497: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1635] Created device /job:localhost/replica:0/task:0/device:GPU:0 with 13678 MB memory:  -> device: 0, name: Tesla T4, pci bus id: 0000:00:04.0, compute capability: 7.5\n","Loaded in: 5.3371453285217285\n","Generating predictions on image: ./data/test/spring.jpg\n","2023-04-26 04:15:35.180183: I tensorflow/compiler/xla/stream_executor/cuda/cuda_dnn.cc:424] Loaded cuDNN version 8700\n","1/1 [==============================] - 3s 3s/step\n","<class 'numpy.ndarray'>\n","[[0.25005168 0.23983881 0.24864015 0.26146936]]\n"]}]},{"cell_type":"code","source":[],"metadata":{"id":"JL3wTkxgb8d-"},"execution_count":null,"outputs":[]},{"cell_type":"code","source":["import tensorflow as tf\n","import matplotlib.pyplot as plt\n","from tensorflow.keras.layers import *\n","from tensorflow.keras.models import Model,Sequential\n","\n","from keras.callbacks import EarlyStopping, ModelCheckpoint\n","\n","train_path = '../color_data/train/'\n","val_path = '../train_face/'\n","\n","# 예측할 클래스 수\n","classes = 4\n","\n","# Input으로 사용될 크기와 채널수\n","height = 256\n","width = 256\n","channels = 3\n","\n","\n","image_generator = tf.keras.preprocessing.image.ImageDataGenerator(rescale=1/255, validation_split=0.33)\n","image_data_train = image_generator.flow_from_directory(train_path,subset='training')\n","image_data_test = image_generator.flow_from_directory(val_path,subset='validation')\n","\n","print(image_data_train.class_indices)\n","\n","resnetv2 = tf.keras.applications.ResNet152V2(include_top=False,input_shape=(height,width,channels))\n","\n","\n","resnetv2.trainable=False\n","\n","\n","model = Sequential([\n","                 resnetv2,\n","                 Dense(512,activation='relu'),\n","                 BatchNormalization(),\n","                 GlobalAveragePooling2D(),\n","                 Dense(classes,activation='softmax')\n","])\n","\n","model.compile(optimizer='adam',loss='categorical_crossentropy',metrics='accuracy')\n","\n","def scheduler(epoch, lr):\n","    if epoch < 2:\n","        return lr\n","    else:\n","        return lr * tf.math.exp(-0.1)\n","\n","lrs = tf.keras.callbacks.LearningRateScheduler(scheduler)\n","\n","# model.fit(image_data_train,batch_size=32,epochs=10,callbacks=[lrs],validation_data=(image_data_test),\n","#           validation_steps =image_data_test.samples/image_data_test.batch_size)\n","\n","early_stopping = EarlyStopping(patience=50)\n","checkpointer = ModelCheckpoint('resnet152v2_best.h5', verbose=1, save_best_only=True)\n","\n","model.fit(image_data_train,batch_size=256,epochs=100, callbacks=[lrs, early_stopping, checkpointer],validation_data=(image_data_test),\n","          validation_steps =image_data_test.samples/image_data_test.batch_size)\n","model.save('resnet152v2_final.h5')"],"metadata":{"colab":{"base_uri":"https://localhost:8080/"},"id":"PgHu_bBbzU-k","executionInfo":{"status":"ok","timestamp":1682490143502,"user_tz":-540,"elapsed":364389,"user":{"displayName":"박민수","userId":"07044323519487942912"}},"outputId":"0a83216c-6da0-4f76-f66e-a265e2933620"},"execution_count":null,"outputs":[{"output_type":"stream","name":"stdout","text":["Found 341 images belonging to 4 classes.\n","Found 68 images belonging to 4 classes.\n","{'autumn_warm': 0, 'spring_warm': 1, 'summer_cool': 2, 'winter_cool': 3}\n","Epoch 1/100\n","11/11 [==============================] - ETA: 0s - loss: 1.3539 - accuracy: 0.4047\n","Epoch 1: val_loss improved from inf to 1.69041, saving model to resnet152v2_best.h5\n","11/11 [==============================] - 22s 1s/step - loss: 1.3539 - accuracy: 0.4047 - val_loss: 1.6904 - val_accuracy: 0.3529 - lr: 0.0010\n","Epoch 2/100\n","11/11 [==============================] - ETA: 0s - loss: 0.6811 - accuracy: 0.7713\n","Epoch 2: val_loss improved from 1.69041 to 1.62097, saving model to resnet152v2_best.h5\n","11/11 [==============================] - 6s 591ms/step - loss: 0.6811 - accuracy: 0.7713 - val_loss: 1.6210 - val_accuracy: 0.4118 - lr: 0.0010\n","Epoch 3/100\n","11/11 [==============================] - ETA: 0s - loss: 0.3878 - accuracy: 0.9150\n","Epoch 3: val_loss improved from 1.62097 to 1.44283, saving model to resnet152v2_best.h5\n","11/11 [==============================] - 6s 594ms/step - loss: 0.3878 - accuracy: 0.9150 - val_loss: 1.4428 - val_accuracy: 0.5147 - lr: 9.0484e-04\n","Epoch 4/100\n","11/11 [==============================] - ETA: 0s - loss: 0.2781 - accuracy: 0.9648\n","Epoch 4: val_loss improved from 1.44283 to 1.04254, saving model to resnet152v2_best.h5\n","11/11 [==============================] - 6s 600ms/step - loss: 0.2781 - accuracy: 0.9648 - val_loss: 1.0425 - val_accuracy: 0.6029 - lr: 8.1873e-04\n","Epoch 5/100\n","11/11 [==============================] - ETA: 0s - loss: 0.2115 - accuracy: 0.9853\n","Epoch 5: val_loss improved from 1.04254 to 0.90312, saving model to resnet152v2_best.h5\n","11/11 [==============================] - 6s 543ms/step - loss: 0.2115 - accuracy: 0.9853 - val_loss: 0.9031 - val_accuracy: 0.6765 - lr: 7.4082e-04\n","Epoch 6/100\n","11/11 [==============================] - ETA: 0s - loss: 0.1397 - accuracy: 0.9912\n","Epoch 6: val_loss improved from 0.90312 to 0.87753, saving model to resnet152v2_best.h5\n","11/11 [==============================] - 6s 573ms/step - loss: 0.1397 - accuracy: 0.9912 - val_loss: 0.8775 - val_accuracy: 0.6618 - lr: 6.7032e-04\n","Epoch 7/100\n","11/11 [==============================] - ETA: 0s - loss: 0.1015 - accuracy: 1.0000\n","Epoch 7: val_loss improved from 0.87753 to 0.79294, saving model to resnet152v2_best.h5\n","11/11 [==============================] - 6s 534ms/step - loss: 0.1015 - accuracy: 1.0000 - val_loss: 0.7929 - val_accuracy: 0.6912 - lr: 6.0653e-04\n","Epoch 8/100\n","11/11 [==============================] - ETA: 0s - loss: 0.0716 - accuracy: 1.0000\n","Epoch 8: val_loss improved from 0.79294 to 0.78481, saving model to resnet152v2_best.h5\n","11/11 [==============================] - 6s 600ms/step - loss: 0.0716 - accuracy: 1.0000 - val_loss: 0.7848 - val_accuracy: 0.7059 - lr: 5.4881e-04\n","Epoch 9/100\n","11/11 [==============================] - ETA: 0s - loss: 0.0574 - accuracy: 1.0000\n","Epoch 9: val_loss improved from 0.78481 to 0.73102, saving model to resnet152v2_best.h5\n","11/11 [==============================] - 6s 603ms/step - loss: 0.0574 - accuracy: 1.0000 - val_loss: 0.7310 - val_accuracy: 0.7206 - lr: 4.9659e-04\n","Epoch 10/100\n","11/11 [==============================] - ETA: 0s - loss: 0.0468 - accuracy: 1.0000\n","Epoch 10: val_loss improved from 0.73102 to 0.72562, saving model to resnet152v2_best.h5\n","11/11 [==============================] - 6s 545ms/step - loss: 0.0468 - accuracy: 1.0000 - val_loss: 0.7256 - val_accuracy: 0.7059 - lr: 4.4933e-04\n","Epoch 11/100\n","11/11 [==============================] - ETA: 0s - loss: 0.0430 - accuracy: 1.0000\n","Epoch 11: val_loss improved from 0.72562 to 0.71177, saving model to resnet152v2_best.h5\n","11/11 [==============================] - 7s 646ms/step - loss: 0.0430 - accuracy: 1.0000 - val_loss: 0.7118 - val_accuracy: 0.7206 - lr: 4.0657e-04\n","Epoch 12/100\n","11/11 [==============================] - ETA: 0s - loss: 0.0385 - accuracy: 1.0000\n","Epoch 12: val_loss improved from 0.71177 to 0.70569, saving model to resnet152v2_best.h5\n","11/11 [==============================] - 6s 539ms/step - loss: 0.0385 - accuracy: 1.0000 - val_loss: 0.7057 - val_accuracy: 0.7059 - lr: 3.6788e-04\n","Epoch 13/100\n","11/11 [==============================] - ETA: 0s - loss: 0.0404 - accuracy: 1.0000\n","Epoch 13: val_loss did not improve from 0.70569\n","11/11 [==============================] - 5s 407ms/step - loss: 0.0404 - accuracy: 1.0000 - val_loss: 0.7098 - val_accuracy: 0.7059 - lr: 3.3287e-04\n","Epoch 14/100\n","11/11 [==============================] - ETA: 0s - loss: 0.0337 - accuracy: 1.0000\n","Epoch 14: val_loss improved from 0.70569 to 0.69988, saving model to resnet152v2_best.h5\n","11/11 [==============================] - 9s 857ms/step - loss: 0.0337 - accuracy: 1.0000 - val_loss: 0.6999 - val_accuracy: 0.7206 - lr: 3.0119e-04\n","Epoch 15/100\n","11/11 [==============================] - ETA: 0s - loss: 0.0313 - accuracy: 1.0000\n","Epoch 15: val_loss did not improve from 0.69988\n","11/11 [==============================] - 4s 355ms/step - loss: 0.0313 - accuracy: 1.0000 - val_loss: 0.7039 - val_accuracy: 0.7500 - lr: 2.7253e-04\n","Epoch 16/100\n","11/11 [==============================] - ETA: 0s - loss: 0.0295 - accuracy: 1.0000\n","Epoch 16: val_loss improved from 0.69988 to 0.69530, saving model to resnet152v2_best.h5\n","11/11 [==============================] - 7s 647ms/step - loss: 0.0295 - accuracy: 1.0000 - val_loss: 0.6953 - val_accuracy: 0.7353 - lr: 2.4660e-04\n","Epoch 17/100\n","11/11 [==============================] - ETA: 0s - loss: 0.0269 - accuracy: 1.0000\n","Epoch 17: val_loss improved from 0.69530 to 0.69462, saving model to resnet152v2_best.h5\n","11/11 [==============================] - 6s 547ms/step - loss: 0.0269 - accuracy: 1.0000 - val_loss: 0.6946 - val_accuracy: 0.7500 - lr: 2.2313e-04\n","Epoch 18/100\n","11/11 [==============================] - ETA: 0s - loss: 0.0240 - accuracy: 1.0000\n","Epoch 18: val_loss did not improve from 0.69462\n","11/11 [==============================] - 4s 351ms/step - loss: 0.0240 - accuracy: 1.0000 - val_loss: 0.7034 - val_accuracy: 0.7500 - lr: 2.0190e-04\n","Epoch 19/100\n","11/11 [==============================] - ETA: 0s - loss: 0.0213 - accuracy: 1.0000\n","Epoch 19: val_loss did not improve from 0.69462\n","11/11 [==============================] - 5s 427ms/step - loss: 0.0213 - accuracy: 1.0000 - val_loss: 0.7077 - val_accuracy: 0.7500 - lr: 1.8268e-04\n","Epoch 20/100\n","11/11 [==============================] - ETA: 0s - loss: 0.0240 - accuracy: 1.0000\n","Epoch 20: val_loss did not improve from 0.69462\n","11/11 [==============================] - 4s 363ms/step - loss: 0.0240 - accuracy: 1.0000 - val_loss: 0.7128 - val_accuracy: 0.7500 - lr: 1.6530e-04\n","Epoch 21/100\n","11/11 [==============================] - ETA: 0s - loss: 0.0221 - accuracy: 1.0000\n","Epoch 21: val_loss did not improve from 0.69462\n","11/11 [==============================] - 4s 353ms/step - loss: 0.0221 - accuracy: 1.0000 - val_loss: 0.7171 - val_accuracy: 0.7500 - lr: 1.4957e-04\n","Epoch 22/100\n","11/11 [==============================] - ETA: 0s - loss: 0.0206 - accuracy: 1.0000\n","Epoch 22: val_loss did not improve from 0.69462\n","11/11 [==============================] - 5s 414ms/step - loss: 0.0206 - accuracy: 1.0000 - val_loss: 0.7202 - val_accuracy: 0.7500 - lr: 1.3534e-04\n","Epoch 23/100\n","11/11 [==============================] - ETA: 0s - loss: 0.0181 - accuracy: 1.0000\n","Epoch 23: val_loss did not improve from 0.69462\n","11/11 [==============================] - 4s 360ms/step - loss: 0.0181 - accuracy: 1.0000 - val_loss: 0.7244 - val_accuracy: 0.7500 - lr: 1.2246e-04\n","Epoch 24/100\n","11/11 [==============================] - ETA: 0s - loss: 0.0219 - accuracy: 1.0000\n","Epoch 24: val_loss did not improve from 0.69462\n","11/11 [==============================] - 4s 350ms/step - loss: 0.0219 - accuracy: 1.0000 - val_loss: 0.7268 - val_accuracy: 0.7500 - lr: 1.1080e-04\n","Epoch 25/100\n","11/11 [==============================] - ETA: 0s - loss: 0.0199 - accuracy: 1.0000\n","Epoch 25: val_loss did not improve from 0.69462\n","11/11 [==============================] - 5s 411ms/step - loss: 0.0199 - accuracy: 1.0000 - val_loss: 0.7294 - val_accuracy: 0.7500 - lr: 1.0026e-04\n","Epoch 26/100\n","11/11 [==============================] - ETA: 0s - loss: 0.0175 - accuracy: 1.0000\n","Epoch 26: val_loss did not improve from 0.69462\n","11/11 [==============================] - 4s 404ms/step - loss: 0.0175 - accuracy: 1.0000 - val_loss: 0.7335 - val_accuracy: 0.7353 - lr: 9.0718e-05\n","Epoch 27/100\n","11/11 [==============================] - ETA: 0s - loss: 0.0174 - accuracy: 1.0000\n","Epoch 27: val_loss did not improve from 0.69462\n","11/11 [==============================] - 4s 351ms/step - loss: 0.0174 - accuracy: 1.0000 - val_loss: 0.7341 - val_accuracy: 0.7500 - lr: 8.2085e-05\n","Epoch 28/100\n","11/11 [==============================] - ETA: 0s - loss: 0.0201 - accuracy: 1.0000\n","Epoch 28: val_loss did not improve from 0.69462\n","11/11 [==============================] - 4s 371ms/step - loss: 0.0201 - accuracy: 1.0000 - val_loss: 0.7355 - val_accuracy: 0.7500 - lr: 7.4273e-05\n","Epoch 29/100\n","11/11 [==============================] - ETA: 0s - loss: 0.0224 - accuracy: 1.0000\n","Epoch 29: val_loss did not improve from 0.69462\n","11/11 [==============================] - 4s 351ms/step - loss: 0.0224 - accuracy: 1.0000 - val_loss: 0.7375 - val_accuracy: 0.7500 - lr: 6.7205e-05\n","Epoch 30/100\n","11/11 [==============================] - ETA: 0s - loss: 0.0163 - accuracy: 1.0000\n","Epoch 30: val_loss did not improve from 0.69462\n","11/11 [==============================] - 4s 350ms/step - loss: 0.0163 - accuracy: 1.0000 - val_loss: 0.7410 - val_accuracy: 0.7500 - lr: 6.0810e-05\n","Epoch 31/100\n","11/11 [==============================] - ETA: 0s - loss: 0.0172 - accuracy: 1.0000\n","Epoch 31: val_loss did not improve from 0.69462\n","11/11 [==============================] - 4s 407ms/step - loss: 0.0172 - accuracy: 1.0000 - val_loss: 0.7423 - val_accuracy: 0.7500 - lr: 5.5023e-05\n","Epoch 32/100\n","11/11 [==============================] - ETA: 0s - loss: 0.0202 - accuracy: 1.0000\n","Epoch 32: val_loss did not improve from 0.69462\n","11/11 [==============================] - 4s 357ms/step - loss: 0.0202 - accuracy: 1.0000 - val_loss: 0.7428 - val_accuracy: 0.7500 - lr: 4.9787e-05\n","Epoch 33/100\n","11/11 [==============================] - ETA: 0s - loss: 0.0182 - accuracy: 1.0000\n","Epoch 33: val_loss did not improve from 0.69462\n","11/11 [==============================] - 4s 356ms/step - loss: 0.0182 - accuracy: 1.0000 - val_loss: 0.7445 - val_accuracy: 0.7500 - lr: 4.5049e-05\n","Epoch 34/100\n","11/11 [==============================] - ETA: 0s - loss: 0.0190 - accuracy: 1.0000\n","Epoch 34: val_loss did not improve from 0.69462\n","11/11 [==============================] - 4s 409ms/step - loss: 0.0190 - accuracy: 1.0000 - val_loss: 0.7476 - val_accuracy: 0.7500 - lr: 4.0762e-05\n","Epoch 35/100\n","11/11 [==============================] - ETA: 0s - loss: 0.0156 - accuracy: 1.0000\n","Epoch 35: val_loss did not improve from 0.69462\n","11/11 [==============================] - 4s 364ms/step - loss: 0.0156 - accuracy: 1.0000 - val_loss: 0.7494 - val_accuracy: 0.7500 - lr: 3.6883e-05\n","Epoch 36/100\n","11/11 [==============================] - ETA: 0s - loss: 0.0187 - accuracy: 1.0000\n","Epoch 36: val_loss did not improve from 0.69462\n","11/11 [==============================] - 4s 354ms/step - loss: 0.0187 - accuracy: 1.0000 - val_loss: 0.7521 - val_accuracy: 0.7500 - lr: 3.3373e-05\n","Epoch 37/100\n","11/11 [==============================] - ETA: 0s - loss: 0.0186 - accuracy: 1.0000\n","Epoch 37: val_loss did not improve from 0.69462\n","11/11 [==============================] - 5s 408ms/step - loss: 0.0186 - accuracy: 1.0000 - val_loss: 0.7548 - val_accuracy: 0.7500 - lr: 3.0197e-05\n","Epoch 38/100\n","11/11 [==============================] - ETA: 0s - loss: 0.0144 - accuracy: 1.0000\n","Epoch 38: val_loss did not improve from 0.69462\n","11/11 [==============================] - 4s 359ms/step - loss: 0.0144 - accuracy: 1.0000 - val_loss: 0.7561 - val_accuracy: 0.7500 - lr: 2.7324e-05\n","Epoch 39/100\n","11/11 [==============================] - ETA: 0s - loss: 0.0154 - accuracy: 1.0000\n","Epoch 39: val_loss did not improve from 0.69462\n","11/11 [==============================] - 4s 370ms/step - loss: 0.0154 - accuracy: 1.0000 - val_loss: 0.7582 - val_accuracy: 0.7500 - lr: 2.4723e-05\n","Epoch 40/100\n","11/11 [==============================] - ETA: 0s - loss: 0.0171 - accuracy: 1.0000\n","Epoch 40: val_loss did not improve from 0.69462\n","11/11 [==============================] - 4s 361ms/step - loss: 0.0171 - accuracy: 1.0000 - val_loss: 0.7598 - val_accuracy: 0.7500 - lr: 2.2371e-05\n","Epoch 41/100\n","11/11 [==============================] - ETA: 0s - loss: 0.0151 - accuracy: 1.0000\n","Epoch 41: val_loss did not improve from 0.69462\n","11/11 [==============================] - 4s 361ms/step - loss: 0.0151 - accuracy: 1.0000 - val_loss: 0.7607 - val_accuracy: 0.7500 - lr: 2.0242e-05\n","Epoch 42/100\n","11/11 [==============================] - ETA: 0s - loss: 0.0201 - accuracy: 1.0000\n","Epoch 42: val_loss did not improve from 0.69462\n","11/11 [==============================] - 4s 352ms/step - loss: 0.0201 - accuracy: 1.0000 - val_loss: 0.7623 - val_accuracy: 0.7500 - lr: 1.8316e-05\n","Epoch 43/100\n","11/11 [==============================] - ETA: 0s - loss: 0.0178 - accuracy: 1.0000\n","Epoch 43: val_loss did not improve from 0.69462\n","11/11 [==============================] - 4s 361ms/step - loss: 0.0178 - accuracy: 1.0000 - val_loss: 0.7636 - val_accuracy: 0.7500 - lr: 1.6573e-05\n","Epoch 44/100\n","11/11 [==============================] - ETA: 0s - loss: 0.0142 - accuracy: 1.0000\n","Epoch 44: val_loss did not improve from 0.69462\n","11/11 [==============================] - 4s 370ms/step - loss: 0.0142 - accuracy: 1.0000 - val_loss: 0.7652 - val_accuracy: 0.7353 - lr: 1.4996e-05\n","Epoch 45/100\n","11/11 [==============================] - ETA: 0s - loss: 0.0148 - accuracy: 1.0000\n","Epoch 45: val_loss did not improve from 0.69462\n","11/11 [==============================] - 4s 351ms/step - loss: 0.0148 - accuracy: 1.0000 - val_loss: 0.7665 - val_accuracy: 0.7353 - lr: 1.3569e-05\n","Epoch 46/100\n","11/11 [==============================] - ETA: 0s - loss: 0.0127 - accuracy: 1.0000\n","Epoch 46: val_loss did not improve from 0.69462\n","11/11 [==============================] - 4s 350ms/step - loss: 0.0127 - accuracy: 1.0000 - val_loss: 0.7674 - val_accuracy: 0.7353 - lr: 1.2277e-05\n","Epoch 47/100\n","11/11 [==============================] - ETA: 0s - loss: 0.0178 - accuracy: 1.0000\n","Epoch 47: val_loss did not improve from 0.69462\n","11/11 [==============================] - 5s 413ms/step - loss: 0.0178 - accuracy: 1.0000 - val_loss: 0.7682 - val_accuracy: 0.7353 - lr: 1.1109e-05\n","Epoch 48/100\n","11/11 [==============================] - ETA: 0s - loss: 0.0147 - accuracy: 1.0000\n","Epoch 48: val_loss did not improve from 0.69462\n","11/11 [==============================] - 4s 349ms/step - loss: 0.0147 - accuracy: 1.0000 - val_loss: 0.7688 - val_accuracy: 0.7353 - lr: 1.0052e-05\n","Epoch 49/100\n","11/11 [==============================] - ETA: 0s - loss: 0.0170 - accuracy: 1.0000\n","Epoch 49: val_loss did not improve from 0.69462\n","11/11 [==============================] - 4s 357ms/step - loss: 0.0170 - accuracy: 1.0000 - val_loss: 0.7695 - val_accuracy: 0.7353 - lr: 9.0953e-06\n","Epoch 50/100\n","11/11 [==============================] - ETA: 0s - loss: 0.0156 - accuracy: 1.0000\n","Epoch 50: val_loss did not improve from 0.69462\n","11/11 [==============================] - 5s 413ms/step - loss: 0.0156 - accuracy: 1.0000 - val_loss: 0.7700 - val_accuracy: 0.7353 - lr: 8.2297e-06\n","Epoch 51/100\n","11/11 [==============================] - ETA: 0s - loss: 0.0156 - accuracy: 1.0000\n","Epoch 51: val_loss did not improve from 0.69462\n","11/11 [==============================] - 4s 356ms/step - loss: 0.0156 - accuracy: 1.0000 - val_loss: 0.7709 - val_accuracy: 0.7353 - lr: 7.4466e-06\n","Epoch 52/100\n","11/11 [==============================] - ETA: 0s - loss: 0.0162 - accuracy: 1.0000\n","Epoch 52: val_loss did not improve from 0.69462\n","11/11 [==============================] - 4s 353ms/step - loss: 0.0162 - accuracy: 1.0000 - val_loss: 0.7713 - val_accuracy: 0.7353 - lr: 6.7379e-06\n","Epoch 53/100\n","11/11 [==============================] - ETA: 0s - loss: 0.0143 - accuracy: 1.0000\n","Epoch 53: val_loss did not improve from 0.69462\n","11/11 [==============================] - 4s 405ms/step - loss: 0.0143 - accuracy: 1.0000 - val_loss: 0.7718 - val_accuracy: 0.7353 - lr: 6.0967e-06\n","Epoch 54/100\n","11/11 [==============================] - ETA: 0s - loss: 0.0130 - accuracy: 1.0000\n","Epoch 54: val_loss did not improve from 0.69462\n","11/11 [==============================] - 4s 369ms/step - loss: 0.0130 - accuracy: 1.0000 - val_loss: 0.7723 - val_accuracy: 0.7353 - lr: 5.5165e-06\n","Epoch 55/100\n","11/11 [==============================] - ETA: 0s - loss: 0.0139 - accuracy: 1.0000\n","Epoch 55: val_loss did not improve from 0.69462\n","11/11 [==============================] - 4s 351ms/step - loss: 0.0139 - accuracy: 1.0000 - val_loss: 0.7729 - val_accuracy: 0.7353 - lr: 4.9916e-06\n","Epoch 56/100\n","11/11 [==============================] - ETA: 0s - loss: 0.0141 - accuracy: 1.0000\n","Epoch 56: val_loss did not improve from 0.69462\n","11/11 [==============================] - 4s 353ms/step - loss: 0.0141 - accuracy: 1.0000 - val_loss: 0.7734 - val_accuracy: 0.7353 - lr: 4.5166e-06\n","Epoch 57/100\n","11/11 [==============================] - ETA: 0s - loss: 0.0171 - accuracy: 1.0000\n","Epoch 57: val_loss did not improve from 0.69462\n","11/11 [==============================] - 5s 418ms/step - loss: 0.0171 - accuracy: 1.0000 - val_loss: 0.7738 - val_accuracy: 0.7353 - lr: 4.0868e-06\n","Epoch 58/100\n","11/11 [==============================] - ETA: 0s - loss: 0.0135 - accuracy: 1.0000\n","Epoch 58: val_loss did not improve from 0.69462\n","11/11 [==============================] - 4s 365ms/step - loss: 0.0135 - accuracy: 1.0000 - val_loss: 0.7740 - val_accuracy: 0.7353 - lr: 3.6979e-06\n","Epoch 59/100\n","11/11 [==============================] - ETA: 0s - loss: 0.0159 - accuracy: 1.0000\n","Epoch 59: val_loss did not improve from 0.69462\n","11/11 [==============================] - 4s 353ms/step - loss: 0.0159 - accuracy: 1.0000 - val_loss: 0.7743 - val_accuracy: 0.7353 - lr: 3.3460e-06\n","Epoch 60/100\n","11/11 [==============================] - ETA: 0s - loss: 0.0175 - accuracy: 1.0000\n","Epoch 60: val_loss did not improve from 0.69462\n","11/11 [==============================] - 5s 413ms/step - loss: 0.0175 - accuracy: 1.0000 - val_loss: 0.7745 - val_accuracy: 0.7353 - lr: 3.0275e-06\n","Epoch 61/100\n","11/11 [==============================] - ETA: 0s - loss: 0.0162 - accuracy: 1.0000\n","Epoch 61: val_loss did not improve from 0.69462\n","11/11 [==============================] - 4s 351ms/step - loss: 0.0162 - accuracy: 1.0000 - val_loss: 0.7747 - val_accuracy: 0.7353 - lr: 2.7394e-06\n","Epoch 62/100\n","11/11 [==============================] - ETA: 0s - loss: 0.0153 - accuracy: 1.0000\n","Epoch 62: val_loss did not improve from 0.69462\n","11/11 [==============================] - 4s 354ms/step - loss: 0.0153 - accuracy: 1.0000 - val_loss: 0.7748 - val_accuracy: 0.7353 - lr: 2.4787e-06\n","Epoch 63/100\n","11/11 [==============================] - ETA: 0s - loss: 0.0157 - accuracy: 1.0000\n","Epoch 63: val_loss did not improve from 0.69462\n","11/11 [==============================] - 4s 417ms/step - loss: 0.0157 - accuracy: 1.0000 - val_loss: 0.7748 - val_accuracy: 0.7353 - lr: 2.2429e-06\n","Epoch 64/100\n","11/11 [==============================] - ETA: 0s - loss: 0.0138 - accuracy: 1.0000\n","Epoch 64: val_loss did not improve from 0.69462\n","11/11 [==============================] - 4s 366ms/step - loss: 0.0138 - accuracy: 1.0000 - val_loss: 0.7751 - val_accuracy: 0.7353 - lr: 2.0294e-06\n","Epoch 65/100\n","11/11 [==============================] - ETA: 0s - loss: 0.0166 - accuracy: 1.0000\n","Epoch 65: val_loss did not improve from 0.69462\n","11/11 [==============================] - 4s 351ms/step - loss: 0.0166 - accuracy: 1.0000 - val_loss: 0.7752 - val_accuracy: 0.7353 - lr: 1.8363e-06\n","Epoch 66/100\n","11/11 [==============================] - ETA: 0s - loss: 0.0131 - accuracy: 1.0000\n","Epoch 66: val_loss did not improve from 0.69462\n","11/11 [==============================] - 5s 420ms/step - loss: 0.0131 - accuracy: 1.0000 - val_loss: 0.7751 - val_accuracy: 0.7353 - lr: 1.6616e-06\n","Epoch 67/100\n","11/11 [==============================] - ETA: 0s - loss: 0.0135 - accuracy: 1.0000\n","Epoch 67: val_loss did not improve from 0.69462\n","11/11 [==============================] - 4s 361ms/step - loss: 0.0135 - accuracy: 1.0000 - val_loss: 0.7754 - val_accuracy: 0.7353 - lr: 1.5034e-06\n"]}]},{"cell_type":"code","source":["import tensorflow as tf\n","import matplotlib.pyplot as plt\n","from tensorflow.keras.layers import *\n","from tensorflow.keras.models import Model,Sequential\n","\n","from keras.callbacks import EarlyStopping, ModelCheckpoint\n","\n","import datetime as dt\n","\n","train_path = '../color_data/train/'\n","val_path = '../train_face/'\n","\n","# 예측할 클래스 수\n","classes = 4\n","\n","# Input으로 사용될 크기와 채널수\n","height = 256\n","width = 256\n","channels = 3\n","\n","\n","image_generator = tf.keras.preprocessing.image.ImageDataGenerator(rescale=1/255,\n","                                                                  rotation_range = 45,\n","                                                                  width_shift_range = 0.2,\n","                                                                  height_shift_range = 0.2,\n","                                                                  shear_range = 0.2,\n","                                                                  zoom_range = 0.7,\n","                                                                  horizontal_flip = True,\n","                                                                  validation_split=0.33,)\n","\n","image_data_train = image_generator.flow_from_directory(train_path,subset='training')\n","image_data_test = image_generator.flow_from_directory(val_path,subset='validation')\n","\n","print(image_data_train.class_indices)\n","\n","resnetv2 = tf.keras.applications.ResNet152V2(include_top=False,input_shape=(height,width,channels))\n","\n","\n","resnetv2.trainable=False\n","\n","\n","model = Sequential([\n","                 resnetv2,\n","                 Dense(512,activation='relu'),\n","                 BatchNormalization(),\n","                 GlobalAveragePooling2D(),\n","                 Dense(classes,activation='softmax')\n","])\n","\n","model.compile(optimizer='adam',loss='categorical_crossentropy',metrics='accuracy')\n","\n","def scheduler(epoch, lr):\n","    if epoch < 2:\n","        return lr\n","    else:\n","        return lr * tf.math.exp(-0.1)\n","\n","lrs = tf.keras.callbacks.LearningRateScheduler(scheduler)\n","\n","# model.fit(image_data_train,batch_size=32,epochs=10,callbacks=[lrs],validation_data=(image_data_test),\n","#           validation_steps =image_data_test.samples/image_data_test.batch_size)\n","\n","early_stopping = EarlyStopping(patience=50)\n","\n","checkpointer = ModelCheckpoint(f'resnet152v2_best_{dt.datetime.now()}.h5', verbose=1, save_best_only=True)\n","\n","model.fit(image_data_train,batch_size=256,epochs=100, callbacks=[lrs, early_stopping, checkpointer],validation_data=(image_data_test),\n","          validation_steps =image_data_test.samples/image_data_test.batch_size)\n","model.save('resnet152v2_final4.h5')"],"metadata":{"colab":{"base_uri":"https://localhost:8080/"},"id":"b05ebb7D_TKS","executionInfo":{"status":"ok","timestamp":1682572612537,"user_tz":-540,"elapsed":983862,"user":{"displayName":"박민수","userId":"07044323519487942912"}},"outputId":"8d69c168-fd74-468b-a195-9a3d7ab12575"},"execution_count":null,"outputs":[{"output_type":"stream","name":"stdout","text":["Found 341 images belonging to 4 classes.\n","Found 68 images belonging to 4 classes.\n","{'autumn_warm': 0, 'spring_warm': 1, 'summer_cool': 2, 'winter_cool': 3}\n","Epoch 1/100\n","11/11 [==============================] - ETA: 0s - loss: 1.5368 - accuracy: 0.3314\n","Epoch 1: val_loss improved from inf to 2.13313, saving model to resnet152v2_best_2023-04-27 05:00:34.633544.h5\n","11/11 [==============================] - 23s 1s/step - loss: 1.5368 - accuracy: 0.3314 - val_loss: 2.1331 - val_accuracy: 0.2794 - lr: 0.0010\n","Epoch 2/100\n","11/11 [==============================] - ETA: 0s - loss: 1.3602 - accuracy: 0.4223\n","Epoch 2: val_loss improved from 2.13313 to 1.98205, saving model to resnet152v2_best_2023-04-27 05:00:34.633544.h5\n","11/11 [==============================] - 11s 975ms/step - loss: 1.3602 - accuracy: 0.4223 - val_loss: 1.9820 - val_accuracy: 0.2794 - lr: 0.0010\n","Epoch 3/100\n","11/11 [==============================] - ETA: 0s - loss: 1.3506 - accuracy: 0.3930\n","Epoch 3: val_loss improved from 1.98205 to 1.50984, saving model to resnet152v2_best_2023-04-27 05:00:34.633544.h5\n","11/11 [==============================] - 12s 1s/step - loss: 1.3506 - accuracy: 0.3930 - val_loss: 1.5098 - val_accuracy: 0.3088 - lr: 9.0484e-04\n","Epoch 4/100\n","11/11 [==============================] - ETA: 0s - loss: 1.2557 - accuracy: 0.4457\n","Epoch 4: val_loss improved from 1.50984 to 1.26778, saving model to resnet152v2_best_2023-04-27 05:00:34.633544.h5\n","11/11 [==============================] - 10s 889ms/step - loss: 1.2557 - accuracy: 0.4457 - val_loss: 1.2678 - val_accuracy: 0.4412 - lr: 8.1873e-04\n","Epoch 5/100\n","11/11 [==============================] - ETA: 0s - loss: 1.1621 - accuracy: 0.5396\n","Epoch 5: val_loss did not improve from 1.26778\n","11/11 [==============================] - 9s 816ms/step - loss: 1.1621 - accuracy: 0.5396 - val_loss: 1.4288 - val_accuracy: 0.4118 - lr: 7.4082e-04\n","Epoch 6/100\n","11/11 [==============================] - ETA: 0s - loss: 1.1680 - accuracy: 0.4780\n","Epoch 6: val_loss improved from 1.26778 to 1.21390, saving model to resnet152v2_best_2023-04-27 05:00:34.633544.h5\n","11/11 [==============================] - 12s 1s/step - loss: 1.1680 - accuracy: 0.4780 - val_loss: 1.2139 - val_accuracy: 0.5147 - lr: 6.7032e-04\n","Epoch 7/100\n","11/11 [==============================] - ETA: 0s - loss: 1.1060 - accuracy: 0.5220\n","Epoch 7: val_loss did not improve from 1.21390\n","11/11 [==============================] - 10s 818ms/step - loss: 1.1060 - accuracy: 0.5220 - val_loss: 1.2437 - val_accuracy: 0.5000 - lr: 6.0653e-04\n","Epoch 8/100\n","11/11 [==============================] - ETA: 0s - loss: 1.1184 - accuracy: 0.4868\n","Epoch 8: val_loss did not improve from 1.21390\n","11/11 [==============================] - 9s 823ms/step - loss: 1.1184 - accuracy: 0.4868 - val_loss: 1.3854 - val_accuracy: 0.3824 - lr: 5.4881e-04\n","Epoch 9/100\n","11/11 [==============================] - ETA: 0s - loss: 1.1420 - accuracy: 0.4927\n","Epoch 9: val_loss improved from 1.21390 to 1.20288, saving model to resnet152v2_best_2023-04-27 05:00:34.633544.h5\n","11/11 [==============================] - 12s 1s/step - loss: 1.1420 - accuracy: 0.4927 - val_loss: 1.2029 - val_accuracy: 0.3676 - lr: 4.9659e-04\n","Epoch 10/100\n","11/11 [==============================] - ETA: 0s - loss: 1.0845 - accuracy: 0.5220\n","Epoch 10: val_loss did not improve from 1.20288\n","11/11 [==============================] - 10s 868ms/step - loss: 1.0845 - accuracy: 0.5220 - val_loss: 1.2967 - val_accuracy: 0.3971 - lr: 4.4933e-04\n","Epoch 11/100\n","11/11 [==============================] - ETA: 0s - loss: 1.0909 - accuracy: 0.5543\n","Epoch 11: val_loss did not improve from 1.20288\n","11/11 [==============================] - 9s 826ms/step - loss: 1.0909 - accuracy: 0.5543 - val_loss: 1.3001 - val_accuracy: 0.3382 - lr: 4.0657e-04\n","Epoch 12/100\n","11/11 [==============================] - ETA: 0s - loss: 1.0089 - accuracy: 0.5660\n","Epoch 12: val_loss did not improve from 1.20288\n","11/11 [==============================] - 8s 690ms/step - loss: 1.0089 - accuracy: 0.5660 - val_loss: 1.3873 - val_accuracy: 0.4118 - lr: 3.6788e-04\n","Epoch 13/100\n","11/11 [==============================] - ETA: 0s - loss: 1.0164 - accuracy: 0.5894\n","Epoch 13: val_loss did not improve from 1.20288\n","11/11 [==============================] - 8s 684ms/step - loss: 1.0164 - accuracy: 0.5894 - val_loss: 1.2472 - val_accuracy: 0.4412 - lr: 3.3287e-04\n","Epoch 14/100\n","11/11 [==============================] - ETA: 0s - loss: 1.0077 - accuracy: 0.6012\n","Epoch 14: val_loss improved from 1.20288 to 1.17091, saving model to resnet152v2_best_2023-04-27 05:00:34.633544.h5\n","11/11 [==============================] - 12s 1s/step - loss: 1.0077 - accuracy: 0.6012 - val_loss: 1.1709 - val_accuracy: 0.4706 - lr: 3.0119e-04\n","Epoch 15/100\n","11/11 [==============================] - ETA: 0s - loss: 0.9823 - accuracy: 0.5982\n","Epoch 15: val_loss improved from 1.17091 to 1.06626, saving model to resnet152v2_best_2023-04-27 05:00:34.633544.h5\n","11/11 [==============================] - 11s 1s/step - loss: 0.9823 - accuracy: 0.5982 - val_loss: 1.0663 - val_accuracy: 0.5147 - lr: 2.7253e-04\n","Epoch 16/100\n","11/11 [==============================] - ETA: 0s - loss: 1.0276 - accuracy: 0.5630\n","Epoch 16: val_loss did not improve from 1.06626\n","11/11 [==============================] - 8s 690ms/step - loss: 1.0276 - accuracy: 0.5630 - val_loss: 1.1646 - val_accuracy: 0.4706 - lr: 2.4660e-04\n","Epoch 17/100\n","11/11 [==============================] - ETA: 0s - loss: 1.0257 - accuracy: 0.5718\n","Epoch 17: val_loss did not improve from 1.06626\n","11/11 [==============================] - 9s 823ms/step - loss: 1.0257 - accuracy: 0.5718 - val_loss: 1.1402 - val_accuracy: 0.5000 - lr: 2.2313e-04\n","Epoch 18/100\n","11/11 [==============================] - ETA: 0s - loss: 0.9701 - accuracy: 0.5982\n","Epoch 18: val_loss did not improve from 1.06626\n","11/11 [==============================] - 9s 818ms/step - loss: 0.9701 - accuracy: 0.5982 - val_loss: 1.1490 - val_accuracy: 0.4706 - lr: 2.0190e-04\n","Epoch 19/100\n","11/11 [==============================] - ETA: 0s - loss: 0.9737 - accuracy: 0.5718\n","Epoch 19: val_loss improved from 1.06626 to 1.01442, saving model to resnet152v2_best_2023-04-27 05:00:34.633544.h5\n","11/11 [==============================] - 11s 1s/step - loss: 0.9737 - accuracy: 0.5718 - val_loss: 1.0144 - val_accuracy: 0.6324 - lr: 1.8268e-04\n","Epoch 20/100\n","11/11 [==============================] - ETA: 0s - loss: 1.0245 - accuracy: 0.5513\n","Epoch 20: val_loss did not improve from 1.01442\n","11/11 [==============================] - 8s 715ms/step - loss: 1.0245 - accuracy: 0.5513 - val_loss: 1.1400 - val_accuracy: 0.5147 - lr: 1.6530e-04\n","Epoch 21/100\n","11/11 [==============================] - ETA: 0s - loss: 0.9699 - accuracy: 0.5953\n","Epoch 21: val_loss did not improve from 1.01442\n","11/11 [==============================] - 9s 854ms/step - loss: 0.9699 - accuracy: 0.5953 - val_loss: 1.2001 - val_accuracy: 0.5147 - lr: 1.4957e-04\n","Epoch 22/100\n","11/11 [==============================] - ETA: 0s - loss: 0.9663 - accuracy: 0.6070\n","Epoch 22: val_loss did not improve from 1.01442\n","11/11 [==============================] - 9s 816ms/step - loss: 0.9663 - accuracy: 0.6070 - val_loss: 1.1125 - val_accuracy: 0.5441 - lr: 1.3534e-04\n","Epoch 23/100\n","11/11 [==============================] - ETA: 0s - loss: 0.9780 - accuracy: 0.5982\n","Epoch 23: val_loss did not improve from 1.01442\n","11/11 [==============================] - 8s 739ms/step - loss: 0.9780 - accuracy: 0.5982 - val_loss: 1.1658 - val_accuracy: 0.4706 - lr: 1.2246e-04\n","Epoch 24/100\n","11/11 [==============================] - ETA: 0s - loss: 0.9679 - accuracy: 0.6129\n","Epoch 24: val_loss did not improve from 1.01442\n","11/11 [==============================] - 9s 806ms/step - loss: 0.9679 - accuracy: 0.6129 - val_loss: 1.0887 - val_accuracy: 0.5294 - lr: 1.1080e-04\n","Epoch 25/100\n","11/11 [==============================] - ETA: 0s - loss: 0.9913 - accuracy: 0.5894\n","Epoch 25: val_loss did not improve from 1.01442\n","11/11 [==============================] - 8s 700ms/step - loss: 0.9913 - accuracy: 0.5894 - val_loss: 1.1989 - val_accuracy: 0.4559 - lr: 1.0026e-04\n","Epoch 26/100\n","11/11 [==============================] - ETA: 0s - loss: 0.9491 - accuracy: 0.6305\n","Epoch 26: val_loss did not improve from 1.01442\n","11/11 [==============================] - 9s 796ms/step - loss: 0.9491 - accuracy: 0.6305 - val_loss: 1.1571 - val_accuracy: 0.4853 - lr: 9.0718e-05\n","Epoch 27/100\n","11/11 [==============================] - ETA: 0s - loss: 0.9925 - accuracy: 0.5894\n","Epoch 27: val_loss did not improve from 1.01442\n","11/11 [==============================] - 9s 851ms/step - loss: 0.9925 - accuracy: 0.5894 - val_loss: 1.0153 - val_accuracy: 0.5000 - lr: 8.2085e-05\n","Epoch 28/100\n","11/11 [==============================] - ETA: 0s - loss: 0.9413 - accuracy: 0.5982\n","Epoch 28: val_loss did not improve from 1.01442\n","11/11 [==============================] - 8s 696ms/step - loss: 0.9413 - accuracy: 0.5982 - val_loss: 1.1119 - val_accuracy: 0.5000 - lr: 7.4273e-05\n","Epoch 29/100\n","11/11 [==============================] - ETA: 0s - loss: 0.9483 - accuracy: 0.5982\n","Epoch 29: val_loss did not improve from 1.01442\n","11/11 [==============================] - 9s 812ms/step - loss: 0.9483 - accuracy: 0.5982 - val_loss: 1.1183 - val_accuracy: 0.5294 - lr: 6.7205e-05\n","Epoch 30/100\n","11/11 [==============================] - ETA: 0s - loss: 0.9750 - accuracy: 0.6012\n","Epoch 30: val_loss did not improve from 1.01442\n","11/11 [==============================] - 9s 812ms/step - loss: 0.9750 - accuracy: 0.6012 - val_loss: 1.0708 - val_accuracy: 0.5735 - lr: 6.0810e-05\n","Epoch 31/100\n","11/11 [==============================] - ETA: 0s - loss: 0.9406 - accuracy: 0.6598\n","Epoch 31: val_loss improved from 1.01442 to 1.00261, saving model to resnet152v2_best_2023-04-27 05:00:34.633544.h5\n","11/11 [==============================] - 11s 1s/step - loss: 0.9406 - accuracy: 0.6598 - val_loss: 1.0026 - val_accuracy: 0.6324 - lr: 5.5023e-05\n","Epoch 32/100\n","11/11 [==============================] - ETA: 0s - loss: 0.9635 - accuracy: 0.5660\n","Epoch 32: val_loss did not improve from 1.00261\n","11/11 [==============================] - 10s 890ms/step - loss: 0.9635 - accuracy: 0.5660 - val_loss: 1.0391 - val_accuracy: 0.5882 - lr: 4.9787e-05\n","Epoch 33/100\n","11/11 [==============================] - ETA: 0s - loss: 0.9315 - accuracy: 0.6657\n","Epoch 33: val_loss did not improve from 1.00261\n","11/11 [==============================] - 8s 764ms/step - loss: 0.9315 - accuracy: 0.6657 - val_loss: 1.1515 - val_accuracy: 0.5000 - lr: 4.5049e-05\n","Epoch 34/100\n","11/11 [==============================] - ETA: 0s - loss: 0.9419 - accuracy: 0.6276\n","Epoch 34: val_loss did not improve from 1.00261\n","11/11 [==============================] - 8s 719ms/step - loss: 0.9419 - accuracy: 0.6276 - val_loss: 1.1390 - val_accuracy: 0.4853 - lr: 4.0762e-05\n","Epoch 35/100\n","11/11 [==============================] - ETA: 0s - loss: 0.9217 - accuracy: 0.6276\n","Epoch 35: val_loss did not improve from 1.00261\n","11/11 [==============================] - 9s 847ms/step - loss: 0.9217 - accuracy: 0.6276 - val_loss: 1.0435 - val_accuracy: 0.5588 - lr: 3.6883e-05\n","Epoch 36/100\n","11/11 [==============================] - ETA: 0s - loss: 0.9332 - accuracy: 0.6129\n","Epoch 36: val_loss did not improve from 1.00261\n","11/11 [==============================] - 8s 713ms/step - loss: 0.9332 - accuracy: 0.6129 - val_loss: 1.0899 - val_accuracy: 0.4706 - lr: 3.3373e-05\n","Epoch 37/100\n","11/11 [==============================] - ETA: 0s - loss: 0.9189 - accuracy: 0.6217\n","Epoch 37: val_loss did not improve from 1.00261\n","11/11 [==============================] - 9s 787ms/step - loss: 0.9189 - accuracy: 0.6217 - val_loss: 1.1293 - val_accuracy: 0.4706 - lr: 3.0197e-05\n","Epoch 38/100\n","11/11 [==============================] - ETA: 0s - loss: 0.9519 - accuracy: 0.5982\n","Epoch 38: val_loss did not improve from 1.00261\n","11/11 [==============================] - 9s 815ms/step - loss: 0.9519 - accuracy: 0.5982 - val_loss: 1.0181 - val_accuracy: 0.5882 - lr: 2.7324e-05\n","Epoch 39/100\n","11/11 [==============================] - ETA: 0s - loss: 0.9596 - accuracy: 0.5982\n","Epoch 39: val_loss did not improve from 1.00261\n","11/11 [==============================] - 8s 707ms/step - loss: 0.9596 - accuracy: 0.5982 - val_loss: 1.0301 - val_accuracy: 0.5147 - lr: 2.4723e-05\n","Epoch 40/100\n","11/11 [==============================] - ETA: 0s - loss: 0.9069 - accuracy: 0.6217\n","Epoch 40: val_loss did not improve from 1.00261\n","11/11 [==============================] - 9s 806ms/step - loss: 0.9069 - accuracy: 0.6217 - val_loss: 1.0204 - val_accuracy: 0.5147 - lr: 2.2371e-05\n","Epoch 41/100\n","11/11 [==============================] - ETA: 0s - loss: 0.9066 - accuracy: 0.6305\n","Epoch 41: val_loss did not improve from 1.00261\n","11/11 [==============================] - 8s 738ms/step - loss: 0.9066 - accuracy: 0.6305 - val_loss: 1.0888 - val_accuracy: 0.5588 - lr: 2.0242e-05\n","Epoch 42/100\n","11/11 [==============================] - ETA: 0s - loss: 0.9723 - accuracy: 0.5777\n","Epoch 42: val_loss did not improve from 1.00261\n","11/11 [==============================] - 8s 731ms/step - loss: 0.9723 - accuracy: 0.5777 - val_loss: 1.1010 - val_accuracy: 0.4706 - lr: 1.8316e-05\n","Epoch 43/100\n","11/11 [==============================] - ETA: 0s - loss: 0.9118 - accuracy: 0.6452\n","Epoch 43: val_loss improved from 1.00261 to 0.95324, saving model to resnet152v2_best_2023-04-27 05:00:34.633544.h5\n","11/11 [==============================] - 12s 1s/step - loss: 0.9118 - accuracy: 0.6452 - val_loss: 0.9532 - val_accuracy: 0.6176 - lr: 1.6573e-05\n","Epoch 44/100\n","11/11 [==============================] - ETA: 0s - loss: 0.9307 - accuracy: 0.6393\n","Epoch 44: val_loss did not improve from 0.95324\n","11/11 [==============================] - 9s 830ms/step - loss: 0.9307 - accuracy: 0.6393 - val_loss: 1.0182 - val_accuracy: 0.5735 - lr: 1.4996e-05\n","Epoch 45/100\n","11/11 [==============================] - ETA: 0s - loss: 0.9435 - accuracy: 0.6540\n","Epoch 45: val_loss did not improve from 0.95324\n","11/11 [==============================] - 8s 699ms/step - loss: 0.9435 - accuracy: 0.6540 - val_loss: 1.0715 - val_accuracy: 0.5441 - lr: 1.3569e-05\n","Epoch 46/100\n","11/11 [==============================] - ETA: 0s - loss: 0.9114 - accuracy: 0.6598\n","Epoch 46: val_loss did not improve from 0.95324\n","11/11 [==============================] - 9s 803ms/step - loss: 0.9114 - accuracy: 0.6598 - val_loss: 1.0051 - val_accuracy: 0.6912 - lr: 1.2277e-05\n","Epoch 47/100\n","11/11 [==============================] - ETA: 0s - loss: 0.9418 - accuracy: 0.6100\n","Epoch 47: val_loss did not improve from 0.95324\n","11/11 [==============================] - 8s 755ms/step - loss: 0.9418 - accuracy: 0.6100 - val_loss: 1.0982 - val_accuracy: 0.5441 - lr: 1.1109e-05\n","Epoch 48/100\n","11/11 [==============================] - ETA: 0s - loss: 0.9485 - accuracy: 0.6246\n","Epoch 48: val_loss did not improve from 0.95324\n","11/11 [==============================] - 8s 694ms/step - loss: 0.9485 - accuracy: 0.6246 - val_loss: 1.1140 - val_accuracy: 0.4706 - lr: 1.0052e-05\n","Epoch 49/100\n","11/11 [==============================] - ETA: 0s - loss: 0.9290 - accuracy: 0.6070\n","Epoch 49: val_loss did not improve from 0.95324\n","11/11 [==============================] - 9s 817ms/step - loss: 0.9290 - accuracy: 0.6070 - val_loss: 1.0179 - val_accuracy: 0.6176 - lr: 9.0953e-06\n","Epoch 50/100\n","11/11 [==============================] - ETA: 0s - loss: 0.9337 - accuracy: 0.6158\n","Epoch 50: val_loss did not improve from 0.95324\n","11/11 [==============================] - 8s 761ms/step - loss: 0.9337 - accuracy: 0.6158 - val_loss: 1.1897 - val_accuracy: 0.4559 - lr: 8.2297e-06\n","Epoch 51/100\n","11/11 [==============================] - ETA: 0s - loss: 0.9004 - accuracy: 0.6422\n","Epoch 51: val_loss did not improve from 0.95324\n","11/11 [==============================] - 8s 719ms/step - loss: 0.9004 - accuracy: 0.6422 - val_loss: 1.0349 - val_accuracy: 0.5735 - lr: 7.4466e-06\n","Epoch 52/100\n","11/11 [==============================] - ETA: 0s - loss: 0.9192 - accuracy: 0.6188\n","Epoch 52: val_loss improved from 0.95324 to 0.93622, saving model to resnet152v2_best_2023-04-27 05:00:34.633544.h5\n","11/11 [==============================] - 12s 1s/step - loss: 0.9192 - accuracy: 0.6188 - val_loss: 0.9362 - val_accuracy: 0.6029 - lr: 6.7379e-06\n","Epoch 53/100\n","11/11 [==============================] - ETA: 0s - loss: 0.9054 - accuracy: 0.6481\n","Epoch 53: val_loss did not improve from 0.93622\n","11/11 [==============================] - 9s 792ms/step - loss: 0.9054 - accuracy: 0.6481 - val_loss: 1.1182 - val_accuracy: 0.5588 - lr: 6.0967e-06\n","Epoch 54/100\n","11/11 [==============================] - ETA: 0s - loss: 0.9457 - accuracy: 0.6276\n","Epoch 54: val_loss improved from 0.93622 to 0.89906, saving model to resnet152v2_best_2023-04-27 05:00:34.633544.h5\n","11/11 [==============================] - 12s 1s/step - loss: 0.9457 - accuracy: 0.6276 - val_loss: 0.8991 - val_accuracy: 0.6765 - lr: 5.5165e-06\n","Epoch 55/100\n","11/11 [==============================] - ETA: 0s - loss: 0.9072 - accuracy: 0.6540\n","Epoch 55: val_loss did not improve from 0.89906\n","11/11 [==============================] - 8s 692ms/step - loss: 0.9072 - accuracy: 0.6540 - val_loss: 1.1164 - val_accuracy: 0.4853 - lr: 4.9916e-06\n","Epoch 56/100\n","11/11 [==============================] - ETA: 0s - loss: 0.9428 - accuracy: 0.6158\n","Epoch 56: val_loss did not improve from 0.89906\n","11/11 [==============================] - 9s 892ms/step - loss: 0.9428 - accuracy: 0.6158 - val_loss: 1.0874 - val_accuracy: 0.5441 - lr: 4.5166e-06\n","Epoch 57/100\n","11/11 [==============================] - ETA: 0s - loss: 0.9012 - accuracy: 0.6393\n","Epoch 57: val_loss did not improve from 0.89906\n","11/11 [==============================] - 8s 755ms/step - loss: 0.9012 - accuracy: 0.6393 - val_loss: 1.0670 - val_accuracy: 0.5000 - lr: 4.0868e-06\n","Epoch 58/100\n","11/11 [==============================] - ETA: 0s - loss: 0.9644 - accuracy: 0.5865\n","Epoch 58: val_loss did not improve from 0.89906\n","11/11 [==============================] - 8s 697ms/step - loss: 0.9644 - accuracy: 0.5865 - val_loss: 1.0532 - val_accuracy: 0.5588 - lr: 3.6979e-06\n","Epoch 59/100\n","11/11 [==============================] - ETA: 0s - loss: 0.9466 - accuracy: 0.6276\n","Epoch 59: val_loss did not improve from 0.89906\n","11/11 [==============================] - 10s 913ms/step - loss: 0.9466 - accuracy: 0.6276 - val_loss: 1.1405 - val_accuracy: 0.4265 - lr: 3.3460e-06\n","Epoch 60/100\n","11/11 [==============================] - ETA: 0s - loss: 0.9009 - accuracy: 0.6393\n","Epoch 60: val_loss did not improve from 0.89906\n","11/11 [==============================] - 9s 840ms/step - loss: 0.9009 - accuracy: 0.6393 - val_loss: 1.0809 - val_accuracy: 0.4706 - lr: 3.0275e-06\n","Epoch 61/100\n","11/11 [==============================] - ETA: 0s - loss: 0.8929 - accuracy: 0.6246\n","Epoch 61: val_loss did not improve from 0.89906\n","11/11 [==============================] - 8s 692ms/step - loss: 0.8929 - accuracy: 0.6246 - val_loss: 1.0934 - val_accuracy: 0.5882 - lr: 2.7394e-06\n","Epoch 62/100\n","11/11 [==============================] - ETA: 0s - loss: 0.9440 - accuracy: 0.6364\n","Epoch 62: val_loss did not improve from 0.89906\n","11/11 [==============================] - 9s 796ms/step - loss: 0.9440 - accuracy: 0.6364 - val_loss: 1.0539 - val_accuracy: 0.6029 - lr: 2.4787e-06\n","Epoch 63/100\n","11/11 [==============================] - ETA: 0s - loss: 0.9059 - accuracy: 0.6569\n","Epoch 63: val_loss did not improve from 0.89906\n","11/11 [==============================] - 8s 706ms/step - loss: 0.9059 - accuracy: 0.6569 - val_loss: 0.9447 - val_accuracy: 0.6618 - lr: 2.2429e-06\n","Epoch 64/100\n","11/11 [==============================] - ETA: 0s - loss: 0.8986 - accuracy: 0.6540\n","Epoch 64: val_loss did not improve from 0.89906\n","11/11 [==============================] - 9s 742ms/step - loss: 0.8986 - accuracy: 0.6540 - val_loss: 0.9393 - val_accuracy: 0.6471 - lr: 2.0294e-06\n","Epoch 65/100\n","11/11 [==============================] - ETA: 0s - loss: 0.9033 - accuracy: 0.6276\n","Epoch 65: val_loss did not improve from 0.89906\n","11/11 [==============================] - 9s 802ms/step - loss: 0.9033 - accuracy: 0.6276 - val_loss: 0.9527 - val_accuracy: 0.6324 - lr: 1.8363e-06\n","Epoch 66/100\n","11/11 [==============================] - ETA: 0s - loss: 0.8970 - accuracy: 0.6334\n","Epoch 66: val_loss did not improve from 0.89906\n","11/11 [==============================] - 9s 805ms/step - loss: 0.8970 - accuracy: 0.6334 - val_loss: 0.9810 - val_accuracy: 0.6029 - lr: 1.6616e-06\n","Epoch 67/100\n","11/11 [==============================] - ETA: 0s - loss: 0.9123 - accuracy: 0.6364\n","Epoch 67: val_loss did not improve from 0.89906\n","11/11 [==============================] - 8s 710ms/step - loss: 0.9123 - accuracy: 0.6364 - val_loss: 1.0934 - val_accuracy: 0.5441 - lr: 1.5034e-06\n","Epoch 68/100\n","11/11 [==============================] - ETA: 0s - loss: 0.9047 - accuracy: 0.6422\n","Epoch 68: val_loss did not improve from 0.89906\n","11/11 [==============================] - 9s 801ms/step - loss: 0.9047 - accuracy: 0.6422 - val_loss: 1.0966 - val_accuracy: 0.5588 - lr: 1.3604e-06\n","Epoch 69/100\n","11/11 [==============================] - ETA: 0s - loss: 0.9101 - accuracy: 0.6452\n","Epoch 69: val_loss did not improve from 0.89906\n","11/11 [==============================] - 9s 814ms/step - loss: 0.9101 - accuracy: 0.6452 - val_loss: 1.0363 - val_accuracy: 0.5735 - lr: 1.2309e-06\n","Epoch 70/100\n","11/11 [==============================] - ETA: 0s - loss: 0.9073 - accuracy: 0.6393\n","Epoch 70: val_loss did not improve from 0.89906\n","11/11 [==============================] - 8s 704ms/step - loss: 0.9073 - accuracy: 0.6393 - val_loss: 1.0360 - val_accuracy: 0.5735 - lr: 1.1138e-06\n","Epoch 71/100\n","11/11 [==============================] - ETA: 0s - loss: 0.9365 - accuracy: 0.5924\n","Epoch 71: val_loss did not improve from 0.89906\n","11/11 [==============================] - 8s 726ms/step - loss: 0.9365 - accuracy: 0.5924 - val_loss: 1.0240 - val_accuracy: 0.5441 - lr: 1.0078e-06\n","Epoch 72/100\n","11/11 [==============================] - ETA: 0s - loss: 0.8914 - accuracy: 0.6393\n","Epoch 72: val_loss did not improve from 0.89906\n","11/11 [==============================] - 9s 809ms/step - loss: 0.8914 - accuracy: 0.6393 - val_loss: 0.9978 - val_accuracy: 0.5294 - lr: 9.1188e-07\n","Epoch 73/100\n","11/11 [==============================] - ETA: 0s - loss: 0.9415 - accuracy: 0.6070\n","Epoch 73: val_loss did not improve from 0.89906\n","11/11 [==============================] - 8s 694ms/step - loss: 0.9415 - accuracy: 0.6070 - val_loss: 1.0481 - val_accuracy: 0.5441 - lr: 8.2510e-07\n","Epoch 74/100\n","11/11 [==============================] - ETA: 0s - loss: 0.9371 - accuracy: 0.6129\n","Epoch 74: val_loss did not improve from 0.89906\n","11/11 [==============================] - 9s 804ms/step - loss: 0.9371 - accuracy: 0.6129 - val_loss: 1.0701 - val_accuracy: 0.5882 - lr: 7.4658e-07\n","Epoch 75/100\n","11/11 [==============================] - ETA: 0s - loss: 0.9608 - accuracy: 0.5806\n","Epoch 75: val_loss did not improve from 0.89906\n","11/11 [==============================] - 8s 758ms/step - loss: 0.9608 - accuracy: 0.5806 - val_loss: 1.0021 - val_accuracy: 0.6029 - lr: 6.7554e-07\n","Epoch 76/100\n","11/11 [==============================] - ETA: 0s - loss: 0.9405 - accuracy: 0.6276\n","Epoch 76: val_loss did not improve from 0.89906\n","11/11 [==============================] - 8s 678ms/step - loss: 0.9405 - accuracy: 0.6276 - val_loss: 1.0888 - val_accuracy: 0.5000 - lr: 6.1125e-07\n","Epoch 77/100\n","11/11 [==============================] - ETA: 0s - loss: 0.9141 - accuracy: 0.6158\n","Epoch 77: val_loss did not improve from 0.89906\n","11/11 [==============================] - 9s 809ms/step - loss: 0.9141 - accuracy: 0.6158 - val_loss: 0.9843 - val_accuracy: 0.5735 - lr: 5.5308e-07\n","Epoch 78/100\n","11/11 [==============================] - ETA: 0s - loss: 0.9260 - accuracy: 0.6393\n","Epoch 78: val_loss did not improve from 0.89906\n","11/11 [==============================] - 8s 737ms/step - loss: 0.9260 - accuracy: 0.6393 - val_loss: 1.0701 - val_accuracy: 0.5000 - lr: 5.0045e-07\n","Epoch 79/100\n","11/11 [==============================] - ETA: 0s - loss: 0.9173 - accuracy: 0.6158\n","Epoch 79: val_loss did not improve from 0.89906\n","11/11 [==============================] - 8s 715ms/step - loss: 0.9173 - accuracy: 0.6158 - val_loss: 1.0731 - val_accuracy: 0.5294 - lr: 4.5283e-07\n","Epoch 80/100\n","11/11 [==============================] - ETA: 0s - loss: 0.9499 - accuracy: 0.6305\n","Epoch 80: val_loss did not improve from 0.89906\n","11/11 [==============================] - 9s 792ms/step - loss: 0.9499 - accuracy: 0.6305 - val_loss: 1.1112 - val_accuracy: 0.4853 - lr: 4.0973e-07\n","Epoch 81/100\n","11/11 [==============================] - ETA: 0s - loss: 0.8933 - accuracy: 0.6716\n","Epoch 81: val_loss did not improve from 0.89906\n","11/11 [==============================] - 8s 699ms/step - loss: 0.8933 - accuracy: 0.6716 - val_loss: 0.9947 - val_accuracy: 0.6029 - lr: 3.7074e-07\n","Epoch 82/100\n","11/11 [==============================] - ETA: 0s - loss: 0.9227 - accuracy: 0.6276\n","Epoch 82: val_loss did not improve from 0.89906\n","11/11 [==============================] - 9s 811ms/step - loss: 0.9227 - accuracy: 0.6276 - val_loss: 1.0730 - val_accuracy: 0.5441 - lr: 3.3546e-07\n","Epoch 83/100\n","11/11 [==============================] - ETA: 0s - loss: 0.8769 - accuracy: 0.6657\n","Epoch 83: val_loss did not improve from 0.89906\n","11/11 [==============================] - 8s 771ms/step - loss: 0.8769 - accuracy: 0.6657 - val_loss: 1.0253 - val_accuracy: 0.5735 - lr: 3.0354e-07\n","Epoch 84/100\n","11/11 [==============================] - ETA: 0s - loss: 0.9119 - accuracy: 0.6276\n","Epoch 84: val_loss did not improve from 0.89906\n","11/11 [==============================] - 8s 700ms/step - loss: 0.9119 - accuracy: 0.6276 - val_loss: 1.0536 - val_accuracy: 0.5441 - lr: 2.7465e-07\n","Epoch 85/100\n","11/11 [==============================] - ETA: 0s - loss: 0.9466 - accuracy: 0.5953\n","Epoch 85: val_loss did not improve from 0.89906\n","11/11 [==============================] - 8s 699ms/step - loss: 0.9466 - accuracy: 0.5953 - val_loss: 1.0985 - val_accuracy: 0.5441 - lr: 2.4852e-07\n","Epoch 86/100\n","11/11 [==============================] - ETA: 0s - loss: 0.9618 - accuracy: 0.6041\n","Epoch 86: val_loss did not improve from 0.89906\n","11/11 [==============================] - 9s 792ms/step - loss: 0.9618 - accuracy: 0.6041 - val_loss: 1.0558 - val_accuracy: 0.5000 - lr: 2.2487e-07\n","Epoch 87/100\n","11/11 [==============================] - ETA: 0s - loss: 0.9433 - accuracy: 0.6276\n","Epoch 87: val_loss did not improve from 0.89906\n","11/11 [==============================] - 9s 826ms/step - loss: 0.9433 - accuracy: 0.6276 - val_loss: 1.0793 - val_accuracy: 0.5588 - lr: 2.0347e-07\n","Epoch 88/100\n","11/11 [==============================] - ETA: 0s - loss: 0.9208 - accuracy: 0.6070\n","Epoch 88: val_loss did not improve from 0.89906\n","11/11 [==============================] - 8s 712ms/step - loss: 0.9208 - accuracy: 0.6070 - val_loss: 1.0049 - val_accuracy: 0.5882 - lr: 1.8410e-07\n","Epoch 89/100\n","11/11 [==============================] - ETA: 0s - loss: 0.9464 - accuracy: 0.5953\n","Epoch 89: val_loss did not improve from 0.89906\n","11/11 [==============================] - 9s 797ms/step - loss: 0.9464 - accuracy: 0.5953 - val_loss: 1.0200 - val_accuracy: 0.5735 - lr: 1.6659e-07\n","Epoch 90/100\n","11/11 [==============================] - ETA: 0s - loss: 0.9390 - accuracy: 0.6070\n","Epoch 90: val_loss did not improve from 0.89906\n","11/11 [==============================] - 9s 799ms/step - loss: 0.9390 - accuracy: 0.6070 - val_loss: 1.0658 - val_accuracy: 0.5000 - lr: 1.5073e-07\n","Epoch 91/100\n","11/11 [==============================] - ETA: 0s - loss: 0.9385 - accuracy: 0.5982\n","Epoch 91: val_loss did not improve from 0.89906\n","11/11 [==============================] - 8s 691ms/step - loss: 0.9385 - accuracy: 0.5982 - val_loss: 1.0999 - val_accuracy: 0.5588 - lr: 1.3639e-07\n","Epoch 92/100\n","11/11 [==============================] - ETA: 0s - loss: 0.9240 - accuracy: 0.6012\n","Epoch 92: val_loss did not improve from 0.89906\n","11/11 [==============================] - 9s 800ms/step - loss: 0.9240 - accuracy: 0.6012 - val_loss: 1.0119 - val_accuracy: 0.5588 - lr: 1.2341e-07\n","Epoch 93/100\n","11/11 [==============================] - ETA: 0s - loss: 0.8870 - accuracy: 0.6364\n","Epoch 93: val_loss did not improve from 0.89906\n","11/11 [==============================] - 8s 683ms/step - loss: 0.8870 - accuracy: 0.6364 - val_loss: 1.0593 - val_accuracy: 0.5441 - lr: 1.1167e-07\n","Epoch 94/100\n","11/11 [==============================] - ETA: 0s - loss: 0.9542 - accuracy: 0.5982\n","Epoch 94: val_loss did not improve from 0.89906\n","11/11 [==============================] - 9s 781ms/step - loss: 0.9542 - accuracy: 0.5982 - val_loss: 0.9843 - val_accuracy: 0.5735 - lr: 1.0104e-07\n","Epoch 95/100\n","11/11 [==============================] - ETA: 0s - loss: 0.9433 - accuracy: 0.5806\n","Epoch 95: val_loss did not improve from 0.89906\n","11/11 [==============================] - 9s 806ms/step - loss: 0.9433 - accuracy: 0.5806 - val_loss: 1.0352 - val_accuracy: 0.5441 - lr: 9.1424e-08\n","Epoch 96/100\n","11/11 [==============================] - ETA: 0s - loss: 0.9401 - accuracy: 0.6188\n","Epoch 96: val_loss did not improve from 0.89906\n","11/11 [==============================] - 8s 722ms/step - loss: 0.9401 - accuracy: 0.6188 - val_loss: 0.9797 - val_accuracy: 0.5588 - lr: 8.2724e-08\n","Epoch 97/100\n","11/11 [==============================] - ETA: 0s - loss: 0.9496 - accuracy: 0.5953\n","Epoch 97: val_loss did not improve from 0.89906\n","11/11 [==============================] - 9s 839ms/step - loss: 0.9496 - accuracy: 0.5953 - val_loss: 1.1502 - val_accuracy: 0.5000 - lr: 7.4851e-08\n","Epoch 98/100\n","11/11 [==============================] - ETA: 0s - loss: 0.9248 - accuracy: 0.6276\n","Epoch 98: val_loss did not improve from 0.89906\n","11/11 [==============================] - 9s 832ms/step - loss: 0.9248 - accuracy: 0.6276 - val_loss: 1.0034 - val_accuracy: 0.5441 - lr: 6.7728e-08\n","Epoch 99/100\n","11/11 [==============================] - ETA: 0s - loss: 0.9760 - accuracy: 0.6129\n","Epoch 99: val_loss did not improve from 0.89906\n","11/11 [==============================] - 8s 703ms/step - loss: 0.9760 - accuracy: 0.6129 - val_loss: 1.0248 - val_accuracy: 0.5588 - lr: 6.1283e-08\n","Epoch 100/100\n","11/11 [==============================] - ETA: 0s - loss: 0.9038 - accuracy: 0.6422\n","Epoch 100: val_loss did not improve from 0.89906\n","11/11 [==============================] - 9s 818ms/step - loss: 0.9038 - accuracy: 0.6422 - val_loss: 0.9883 - val_accuracy: 0.6176 - lr: 5.5451e-08\n"]}]},{"cell_type":"code","source":[],"metadata":{"id":"_nEG46DLSUKi"},"execution_count":null,"outputs":[]},{"cell_type":"code","source":[],"metadata":{"id":"tJB5TfPESUP5"},"execution_count":null,"outputs":[]},{"cell_type":"code","source":["!ls ../color_data/train"],"metadata":{"colab":{"base_uri":"https://localhost:8080/"},"id":"GhmJBTAk1ero","executionInfo":{"status":"ok","timestamp":1682483591344,"user_tz":-540,"elapsed":670,"user":{"displayName":"박민수","userId":"07044323519487942912"}},"outputId":"03682ef3-e9f3-450c-e08b-6d9551a18cda"},"execution_count":null,"outputs":[{"output_type":"stream","name":"stdout","text":["autumn_warm  spring_warm  summer_cool  winter_cool\n"]}]},{"cell_type":"code","source":["from tensorflow.keras.preprocessing.image import load_img\n","from tensorflow.keras.preprocessing.image import img_to_array\n","import matplotlib.pyplot as plt\n","import numpy as np\n","import os\n","\n","from keras.models import load_model\n","\n","\n","img_height = 256\n","img_width = 256\n","\n","\n","# filename = ['../test_face/autumn_warm/face_000.jpg', \"../test_face/spring_warm/face_002.jpg\", \"../test_face/summer_cool/face_001.jpg\", \"../test_face/winter_cool/face_000.jpg\" ]\n","\n","filename = []\n","\n","path = \"../test_face2/\"\n","img_list = os.listdir(path)\n","for i in img_list:\n","  filename.append(path+i)\n","                    \n","predict_dictionary = {0:'autumn_warm',1:'spring_warm',2:'summer_cool', 3:\"winter_cool\"}\n","\n","model_path = \"./resnet152v2_best.h5\"\n","# model_path = \"./resnet50v2_final.h5\"\n","\n","model = load_model(model_path)\n","\n","answer_count = 0\n","f = 0\n","tone = 0\n","\n","a_list = []\n","f_list = []\n","\n","for file in filename:\n","    original = load_img(file, target_size = (img_height,img_width))\n","\n","    \n","    numpy_image = img_to_array(original)\n","    plt.text(30, 10, str(file.split(\"/\")[2]))\n","    plt.imshow(np.uint8(numpy_image))\n","    plt.show()\n","    print(\"numpy array size : \", numpy_image.shape)\n","    image_batch = np.expand_dims(numpy_image , axis = 0)\n","\n","    predict = np.argmax(model.predict(image_batch/255.))\n","\n","    print('결과 : ',predict_dictionary[predict])\n","    \n","    s1 = file.split(\"/\")[2].split(\"_\")[0]         \n","    s2 = file.split(\"/\")[2].split(\"_\")[1][:4]\n","\n","    if predict_dictionary[predict].split(\"_\")[1] == s2:\n","      tone += 1\n","      if predict_dictionary[predict].split(\"_\")[0] == s1:\n","        answer_count+=1\n","    else:\n","      f += 1\n","\n","\n","print(\"a :\", answer_count, \"tone :\", tone-answer_count, \" f :\", f)"],"metadata":{"colab":{"base_uri":"https://localhost:8080/","height":1000,"output_embedded_package_id":"1Y6GbR2nvmNE22bCl7pc8yIQkUmdNtjXS"},"id":"tC3jbPqwzVD3","executionInfo":{"status":"ok","timestamp":1682489497160,"user_tz":-540,"elapsed":34479,"user":{"displayName":"박민수","userId":"07044323519487942912"}},"outputId":"0be0ceb4-4b56-4a14-94e4-308b476f95b6"},"execution_count":null,"outputs":[{"output_type":"display_data","data":{"text/plain":"Output hidden; open in https://colab.research.google.com to view."},"metadata":{}}]},{"cell_type":"code","source":["from tensorflow.keras.preprocessing.image import load_img\n","from tensorflow.keras.preprocessing.image import img_to_array\n","import matplotlib.pyplot as plt\n","import numpy as np\n","import os\n","\n","from keras.models import load_model\n","\n","\n","img_height = 256\n","img_width = 256\n","\n","\n","# filename = ['../test_face/autumn_warm/face_000.jpg', \"../test_face/spring_warm/face_002.jpg\", \"../test_face/summer_cool/face_001.jpg\", \"../test_face/winter_cool/face_000.jpg\" ]\n","\n","filename = []\n","\n","path = \"../test_face2/\"\n","img_list = os.listdir(path)\n","for i in img_list:\n","  filename.append(path+i)\n","                    \n","predict_dictionary = {0:'autumn_warm',1:'spring_warm',2:'summer_cool', 3:\"winter_cool\"}\n","\n","model_path = \"./resnet152v2_best.h5\"\n","# model_path = \"./resnet50v2_final.h5\"\n","\n","model = load_model(model_path)\n","\n","answer_count = 0\n","f = 0\n","tone = 0\n","\n","a_list = []\n","f_list = []\n","\n","for file in filename:\n","    original = load_img(file, target_size = (img_height,img_width))\n","\n","    \n","    numpy_image = img_to_array(original)\n","    plt.text(30, 10, str(file.split(\"/\")[2]))\n","    plt.imshow(np.uint8(numpy_image))\n","    plt.show()\n","    print(\"numpy array size : \", numpy_image.shape)\n","    image_batch = np.expand_dims(numpy_image , axis = 0)\n","\n","    predict = np.argmax(model.predict(image_batch/255.))\n","\n","    print('결과 : ',predict_dictionary[predict])\n","    \n","    s1 = file.split(\"/\")[2].split(\"_\")[0]         \n","    s2 = file.split(\"/\")[2].split(\"_\")[1][:4]\n","\n","    if predict_dictionary[predict].split(\"_\")[1] == s2:\n","      tone += 1\n","      if predict_dictionary[predict].split(\"_\")[0] == s1:\n","        answer_count+=1\n","    else:\n","      f += 1\n","      f_list.append()\n","\n","\n","print(\"a :\", answer_count, \"tone :\", tone-answer_count, \" f :\", f)"],"metadata":{"colab":{"base_uri":"https://localhost:8080/","height":1000,"output_embedded_package_id":"1R4crdeXgpOLLb3i6e9D-tIO9OXrqVqZK"},"id":"y27qdLMXOtXQ","executionInfo":{"status":"ok","timestamp":1682490234700,"user_tz":-540,"elapsed":45722,"user":{"displayName":"박민수","userId":"07044323519487942912"}},"outputId":"b15f0484-45b3-490d-a2b6-68ebbf8c3b3f"},"execution_count":null,"outputs":[{"output_type":"display_data","data":{"text/plain":"Output hidden; open in https://colab.research.google.com to view."},"metadata":{}}]},{"cell_type":"code","source":["from tensorflow.keras.preprocessing.image import load_img\n","from tensorflow.keras.preprocessing.image import img_to_array\n","import matplotlib.pyplot as plt\n","import numpy as np\n","import os\n","\n","from keras.models import load_model\n","\n","\n","img_height = 256\n","img_width = 256\n","\n","\n","# filename = ['../test_face/autumn_warm/face_000.jpg', \"../test_face/spring_warm/face_002.jpg\", \"../test_face/summer_cool/face_001.jpg\", \"../test_face/winter_cool/face_000.jpg\" ]\n","\n","filename = []\n","\n","path = \"../test_face2/\"\n","img_list = os.listdir(path)\n","for i in img_list:\n","  filename.append(path+i)\n","                    \n","predict_dictionary = {0:'autumn_warm',1:'spring_warm',2:'summer_cool', 3:\"winter_cool\"}\n","\n","model_path = \"./resnet152v2_best_1.h5\"\n","# model_path = \"./resnet50v2_final.h5\"\n","\n","model = load_model(model_path)\n","\n","answer_count = 0\n","f = 0\n","tone = 0\n","\n","a_list = []\n","f_list = []\n","\n","for file in filename:\n","    original = load_img(file, target_size = (img_height,img_width))\n","\n","    \n","    numpy_image = img_to_array(original)\n","    plt.text(30, 10, str(file.split(\"/\")[2]))\n","    plt.imshow(np.uint8(numpy_image))\n","    plt.show()\n","    print(\"numpy array size : \", numpy_image.shape)\n","    image_batch = np.expand_dims(numpy_image , axis = 0)\n","\n","    predict = np.argmax(model.predict(image_batch/255.))\n","\n","    print('결과 : ',predict_dictionary[predict])\n","    \n","    s1 = file.split(\"/\")[2].split(\"_\")[0]         \n","    s2 = file.split(\"/\")[2].split(\"_\")[1][:4]\n","\n","    if predict_dictionary[predict].split(\"_\")[1] == s2:\n","      tone += 1\n","      if predict_dictionary[predict].split(\"_\")[0] == s1:\n","        answer_count+=1\n","    else:\n","      f += 1\n","\n","\n","print(\"a :\", answer_count, \"tone :\", tone-answer_count, \" f :\", f)"],"metadata":{"id":"VYHuMC-kzsdb","colab":{"base_uri":"https://localhost:8080/","height":1000,"output_embedded_package_id":"1aPJq2s1fT2DTTlxVJnFulGX4pWdG3JIB"},"executionInfo":{"status":"ok","timestamp":1682573145254,"user_tz":-540,"elapsed":38029,"user":{"displayName":"박민수","userId":"07044323519487942912"}},"outputId":"a6665a5d-43be-4c0c-f41a-cec9e44b8a58"},"execution_count":null,"outputs":[{"output_type":"display_data","data":{"text/plain":"Output hidden; open in https://colab.research.google.com to view."},"metadata":{}}]},{"cell_type":"code","source":["from tensorflow.keras.preprocessing.image import load_img\n","from tensorflow.keras.preprocessing.image import img_to_array\n","import matplotlib.pyplot as plt\n","import numpy as np\n","import os\n","\n","from keras.models import load_model\n","\n","\n","img_height = 256\n","img_width = 256\n","\n","\n","# filename = ['../test_face/autumn_warm/face_000.jpg', \"../test_face/spring_warm/face_002.jpg\", \"../test_face/summer_cool/face_001.jpg\", \"../test_face/winter_cool/face_000.jpg\" ]\n","\n","filename = []\n","\n","path = \"../test_face2/\"\n","img_list = os.listdir(path)\n","for i in img_list:\n","  filename.append(path+i)\n","                    \n","predict_dictionary = {0:'autumn_warm',1:'spring_warm',2:'summer_cool', 3:\"winter_cool\"}\n","\n","model_path = \"./resnet152v2_best_2.h5\"\n","# model_path = \"./resnet50v2_final.h5\"\n","\n","model = load_model(model_path)\n","\n","answer_count = 0\n","f = 0\n","tone = 0\n","\n","a_list = []\n","f_list = []\n","\n","for file in filename:\n","    original = load_img(file, target_size = (img_height,img_width))\n","\n","    \n","    numpy_image = img_to_array(original)\n","    plt.text(30, 10, str(file.split(\"/\")[2]))\n","    plt.imshow(np.uint8(numpy_image))\n","    plt.show()\n","    print(\"numpy array size : \", numpy_image.shape)\n","    image_batch = np.expand_dims(numpy_image , axis = 0)\n","\n","    predict = np.argmax(model.predict(image_batch/255.))\n","\n","    print('결과 : ',predict_dictionary[predict])\n","    \n","    s1 = file.split(\"/\")[2].split(\"_\")[0]         \n","    s2 = file.split(\"/\")[2].split(\"_\")[1][:4]\n","\n","    if predict_dictionary[predict].split(\"_\")[1] == s2:\n","      tone += 1\n","      if predict_dictionary[predict].split(\"_\")[0] == s1:\n","        answer_count+=1\n","    else:\n","      f += 1\n","\n","\n","print(\"a :\", answer_count, \"tone :\", tone-answer_count, \" f :\", f)"],"metadata":{"id":"rc4rzLw0zsRL","colab":{"base_uri":"https://localhost:8080/","height":1000,"output_embedded_package_id":"1tSPhVgPf8EV6OAQ0NLLE-3ge3Q2Yszyz"},"executionInfo":{"status":"ok","timestamp":1682573177237,"user_tz":-540,"elapsed":31991,"user":{"displayName":"박민수","userId":"07044323519487942912"}},"outputId":"1c53a047-dacc-4bce-d027-4c4bbd158217"},"execution_count":null,"outputs":[{"output_type":"display_data","data":{"text/plain":"Output hidden; open in https://colab.research.google.com to view."},"metadata":{}}]},{"cell_type":"code","source":["from tensorflow.keras.preprocessing.image import load_img\n","from tensorflow.keras.preprocessing.image import img_to_array\n","import matplotlib.pyplot as plt\n","import numpy as np\n","import os\n","\n","from keras.models import load_model\n","\n","\n","img_height = 256\n","img_width = 256\n","\n","\n","# filename = ['../test_face/autumn_warm/face_000.jpg', \"../test_face/spring_warm/face_002.jpg\", \"../test_face/summer_cool/face_001.jpg\", \"../test_face/winter_cool/face_000.jpg\" ]\n","\n","filename = []\n","\n","path = \"./test/\"\n","img_list = os.listdir(path)\n","for i in img_list:\n","  filename.append(path+i)\n","                    \n","predict_dictionary = {0:'autumn_warm',1:'spring_warm',2:'summer_cool', 3:\"winter_cool\"}\n","\n","model_path = \"./resnet152v2_best_2.h5\"\n","# model_path = \"./resnet50v2_final.h5\"\n","\n","model = load_model(model_path)\n","\n","answer_count = 0\n","f = 0\n","tone = 0\n","\n","a_list = []\n","f_list = []\n","\n","for file in filename:\n","    original = load_img(file, target_size = (img_height,img_width))\n","\n","    \n","    numpy_image = img_to_array(original)\n","    plt.text(30, 10, str(file.split(\"/\")[2]))\n","    plt.imshow(np.uint8(numpy_image))\n","    plt.show()\n","    print(\"numpy array size : \", numpy_image.shape)\n","    image_batch = np.expand_dims(numpy_image , axis = 0)\n","\n","    predict = np.argmax(model.predict(image_batch/255.))\n","\n","    print('결과 : ',predict_dictionary[predict])\n","    \n","    s1 = file.split(\"/\")[2].split(\"_\")[0]         \n","    s2 = file.split(\"/\")[2].split(\"_\")[1][:4]\n","\n","    if predict_dictionary[predict].split(\"_\")[1] == s2:\n","      tone += 1\n","      if predict_dictionary[predict].split(\"_\")[0] == s1:\n","        answer_count+=1\n","    else:\n","      f += 1\n","\n","\n","print(\"a :\", answer_count, \"tone :\", tone-answer_count, \" f :\", f)"],"metadata":{"colab":{"base_uri":"https://localhost:8080/","height":705},"id":"9nX-s7SQK9yK","executionInfo":{"status":"error","timestamp":1682585051047,"user_tz":-540,"elapsed":29931,"user":{"displayName":"박민수","userId":"07044323519487942912"}},"outputId":"da2655b0-4cdd-453e-e911-2f9bf143843d"},"execution_count":null,"outputs":[{"output_type":"display_data","data":{"text/plain":["<Figure size 640x480 with 1 Axes>"],"image/png":"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\n"},"metadata":{}},{"output_type":"stream","name":"stdout","text":["numpy array size :  (256, 256, 3)\n","1/1 [==============================] - 11s 11s/step\n","결과 :  winter_cool\n"]},{"output_type":"error","ename":"IndexError","evalue":"ignored","traceback":["\u001b[0;31m---------------------------------------------------------------------------\u001b[0m","\u001b[0;31mIndexError\u001b[0m                                Traceback (most recent call last)","\u001b[0;32m<ipython-input-4-6ac6057e0347>\u001b[0m in \u001b[0;36m<cell line: 37>\u001b[0;34m()\u001b[0m\n\u001b[1;32m     51\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     52\u001b[0m     \u001b[0ms1\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mfile\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0msplit\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m\"/\"\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0;36m2\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0msplit\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m\"_\"\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0;36m0\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m---> 53\u001b[0;31m     \u001b[0ms2\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mfile\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0msplit\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m\"/\"\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0;36m2\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0msplit\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m\"_\"\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0;36m1\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;36m4\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m     54\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     55\u001b[0m     \u001b[0;32mif\u001b[0m \u001b[0mpredict_dictionary\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0mpredict\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0msplit\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m\"_\"\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0;36m1\u001b[0m\u001b[0;34m]\u001b[0m \u001b[0;34m==\u001b[0m \u001b[0ms2\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n","\u001b[0;31mIndexError\u001b[0m: list index out of range"]}]},{"cell_type":"code","source":[],"metadata":{"id":"6OFbe6kiK9q1"},"execution_count":null,"outputs":[]},{"cell_type":"code","source":[],"metadata":{"id":"BwNhPKZuK9js"},"execution_count":null,"outputs":[]},{"cell_type":"code","source":[],"metadata":{"id":"WFKXyMzE5QlN"},"execution_count":null,"outputs":[]}]}