Update submodule models and load CNN for odbjdet just once

Author: fedjo

models

@@ -0,0 +1,19 @@
+# Tensorflow Graph load
+import os
+import tensorflow as tf
+
+
+CWD_PATH = os.path.join(os.getenv('FACEREC_APP_DIR', '..'), 'aat')
+
+# Path to frozen detection graph. This is the actual model that is used for the object detection.
+MODEL_NAME = 'ssd_mobilenet_v1_coco_2017_11_17'
+# MODEL_NAME = 'faster_rcnn_inception_resnet_v2_atrous_oid_2018_01_28'
+PATH_TO_CKPT = os.path.join(CWD_PATH, 'object_detection', MODEL_NAME, 'frozen_inference_graph.pb')
+
+_detection_graph = tf.Graph()
+with _detection_graph.as_default():
+    od_graph_def = tf.GraphDef()
+    with tf.gfile.GFile(PATH_TO_CKPT, 'rb') as fid:
+        serialized_graph = fid.read()
+        od_graph_def.ParseFromString(serialized_graph)
+        tf.import_graph_def(od_graph_def, name='')

project/aat/__init__.py

@@ -21,6 +21,7 @@
 
 from django.conf import settings
 
+from aat import _detection_graph, CWD_PATH
 from .models import Cascade
 from .utils.utils import exec_cmd
 import aat.utils.recognizer_utils as recognizer
@@ -143,13 +144,6 @@ def object_detection2(self, video_path, framerate):
         log.debug("Cannot open video path {}".format(video_path))
         return {'od_error': 'Cannot open video'}
 
-    CWD_PATH = os.path.join(os.getenv('FACEREC_APP_DIR', '..'), 'aat')
-
-    # Path to frozen detection graph. This is the actual model that is used for the object detection.
-    MODEL_NAME = 'ssd_mobilenet_v1_coco_2017_11_17'
-    # MODEL_NAME = 'faster_rcnn_inception_resnet_v2_atrous_oid_2018_01_28'
-    PATH_TO_CKPT = os.path.join(CWD_PATH, 'object_detection', MODEL_NAME, 'frozen_inference_graph.pb')
-
     # List of the strings that is used to add correct label for each box.
     PATH_TO_LABELS = os.path.join(CWD_PATH, 'object_detection', 'data', 'mscoco_label_map.pbtxt')
     # PATH_TO_LABELS = os.path.join(CWD_PATH, 'object_detection', 'data', ' oid_bbox_trainable_label_map.pbtxt')
@@ -186,15 +180,9 @@ def object_detection2(self, video_path, framerate):
 
             # frame_with_objects = cv2.cvtColor(frame, cv2.COLOR_BGR2RGB)
             frame_with_objects = frame.copy()
-            detection_graph = tf.Graph()
-            with detection_graph.as_default():
-                od_graph_def = tf.GraphDef()
-                with tf.gfile.GFile(PATH_TO_CKPT, 'rb') as fid:
-                    serialized_graph = fid.read()
-                    od_graph_def.ParseFromString(serialized_graph)
-                    tf.import_graph_def(od_graph_def, name='')
-
-                sess = tf.Session(graph=detection_graph)
+
+            detection_graph = _detection_graph
+            sess = tf.Session(graph=detection_graph)
 
              # Expand dimensions since the model expects images to have shape:
              # [1, None, None, 3]