backup

c++/readme.md

@@ -1,35 +1,39 @@
 # C++ port
 
 The C++ port of this repo is originally created and contributed by [Ammar Ali](https://github.com/ammarali32), and slightly modified by [cleardusk](https://github.com/cleardusk).
+The implementation relies on the new OpenCV (4.2.0+) framework to read onnx and do the inference.
 
 ## Foreword
 
 * The implementation is not optimized
 * The converted onnx model and the yolo-face detector weights can be downloads in [GoogleDrive](https://drive.google.com/drive/folders/1Iprh93fpGkgebc7rw1OSOEemvxpxFXIB) or [BaiduDrive](https://pan.baidu.com/s/1jSLUK3FEEHf1DE1UqOrO1g) (password: 6kjr)
 * Dependency: >= OpenCV 4.2.0
-* yolo face detector is tested in this port to substitute dlib
+* yolo face detector is tested in this port to substitute dlib (welcome for contributing light-weighted face detector of the c++ port)
 
 ## How to build and run
 
 ### macOS
 ```shell script
-# Install or upgrade OpenCV, the default is 4.4.0. This may take a long long time !!!
+# Step 1: Install or upgrade OpenCV, the default is 4.4.0. This may take a long long time !!!
 brew upgrade opencv # or `brew install opencv` if opencv is not installed
 
-# build cpp demo
+# Step 2: build cpp demo
 cd c++ # cd into the c++ dir
 mkdir build
 cd build
 cmake ..
 make
 
-# run
+# Step 3: put the downloaded weights `mb_1.onnx` and `tiny-yolo-azface-fddb_82000.weights` into `weights` dir
+
+# Run
+cd ..
 ./build/demo
 ```
 
 ### Linux
 ```shell script
-# build opencv, >= 4.2.0 will be ok
+# Step 1: build opencv, >= 4.2.0 will be ok
 git clone https://github.com/opencv/opencv.git # you can checkout the desired branch version
 cd opencv
 mkdir build
@@ -38,22 +42,25 @@ cmake -D CMAKE_BUILD_TYPE=Release -D CMAKE_INSTALL_PREFIX=/usr/local ..
 make -j
 sudo make install
 
-# build cpp demo
+# Step 2: build cpp demo
 cd c++ # cd into the c++ dir
 mkdir build
 cd build
 cmake ..
 make
 
-# run
+# Step 3: put the downloaded weights `mb_1.onnx` and `tiny-yolo-azface-fddb_82000.weights` into `weights` dir
+
+# Run
+cd ..
 ./build/demo
 ```
 
 ### Result
 The result will be stored in `res` by default. Below is a demo running by this c++ port.
 
 <p align="center">
-  <img src="demo.jpg" alt="A demo" width="600px">
+  <img src="demo.jpg" alt="A demo" width="750px">
 </p>
 
 
@@ -72,7 +79,7 @@ Inference time: 6.62105 ms
 See [convert_to_onnx.py](./convert_to_onnx.py).
 
 ### TODO
-More functionalities are to be implemented.
+More functionalities are to be implemented. And welcome for discussions or PRs.
 <!-- I hope that I will have time to complete all functionality and not just facial landmarks but i am sure that from here it is not hard to be implemented. -->
 
 

readme.md

@@ -14,8 +14,8 @@ By [Jianzhu Guo](http://guojianzhu.com).
 </p>
 
 **\[Updates\]**
- - `2020.8.2`: Update a simple c++ port of this project, see [here](./c++/readme.md). The implementation relies on the new OpenCV (4.2.0+) framework to read onnx and do the inference.  
- - `2020.7.3`: The extended work [Towards Fast, Accurate and Stable 3D Dense Face Alignment](https://guojianzhu.com/assets/pdfs/3162.pdf) is accepted by <strong>[ECCV 2020](https://eccv2020.eu/)</strong>.
+ - `2020.8.2`: Update a <strong>[simple c++ port](./c++/readme.md)</strong> of this project.
+ - `2020.7.3`: The extended work <strong>[Towards Fast, Accurate and Stable 3D Dense Face Alignment](https://guojianzhu.com/assets/pdfs/3162.pdf)</strong> is accepted by [ECCV 2020](https://eccv2020.eu/). See [my page](https://guojianzhu.com) for more details.
  - `2019.9.15`: Some updates, see the commits for details.
  - `2019.6.17`: Adding a [video demo](./video_demo.py) contributed by [zjjMaiMai](https://github.com/zjjMaiMai).
  - `2019.5.2`: Evaluating inference speed on CPU with PyTorch v1.1.0, see [here](#CPU) and [speed_cpu.py](./speed_cpu.py).
@@ -32,6 +32,7 @@ By [Jianzhu Guo](http://guojianzhu.com).
 
 **\[Todo\]**
 
+- [x] Add c++ port.
 - [x] Depth image estimation.
 - [x] PNCC (Projected Normalized Coordinate Code).
 - [x] PAF (Pose Adaptive Feature).
@@ -311,4 +312,4 @@ Welcome to focus on my 3D face related works: [MeGlass](https://github.com/clear
 
 
 ## Contact
-**Jianzhu Guo (郭建珠)** [[Homepage-Academic](http://guojianzhu.com), [Homepage-Blog](http://vlog.guojianzhu.com), [Google Scholar](https://scholar.google.com/citations?user=W8_JzNcAAAAJ&hl=en&oi=ao)]:  **[email protected]**. 
+**Jianzhu Guo (郭建珠)** [[Homepage](http://guojianzhu.com), [Google Scholar](https://scholar.google.com/citations?user=W8_JzNcAAAAJ&hl=en&oi=ao)]:  **[email protected]** or **[email protected]**.