Training for libfacedetection in PyTorch

It is the training program for libfacedetection. The source code is based on FaceBoxes.PyTorch and ssd.pytorch.

Visualization of our network architecture: [netron].

Contents

• Installation
• Preparation
• Training
• Detection
• Evaluation on WIDER Face
• Export CPP source code
• Export to ONNX model
• Design your own model
• Citation

Installation

1. Install PyTorch >= v1.7.0 following official instruction.

2. Clone this repository. We will call the cloned directory as $TRAIN_ROOT.

   git clone https://github.com/ShiqiYu/libfacedetection.train

3. Install dependencies.

   pip install -r requirements.txt

Note: Codes are based on Python 3+.

Preparation

1. Download the WIDER Face dataset and its evaluation tools.
2. Extract zip files under $TRAIN_ROOT/data/widerface as follows:

   $ tree data/widerface
   data/widerface
   ├── eval_tools
   ├── wider_face_split
   ├── WIDER_test
   ├── WIDER_train
   ├── WIDER_val
   └── trainset.json           

NOTE:
We relabled the WIDER Face train set using RetinaFace. New labels are in $TRAIN_ROOT/data/widerface/trainset.json, which is the COCO_format annotations file used in DALI dataloader.

Training

cd $TRAIN_ROOT/tasks/task1/
python train.py

Detection

cd $TRAIN_ROOT/tasks/task1/
python detect.py -m weights/yunet_final.pth --image_file=filename.jpg

Evaluation on WIDER Face

1. Build NMS module.

   cd $TRAIN_ROOT/src/widerface_eval
   python setup.py build_ext --inplace

2. Perform evaluation. To reproduce the following performance, run on the default settings. Run python test.py --help for more options.

   cd $TRAIN_ROOT/tasks/task1/
   python test.py -m weights/yunet_final.pth

NOTE: We now use the Python version of eval_tools from here.

Performance on WIDER Face (Val): scales=[1.], confidence_threshold=0.3:

AP_easy=0.856, AP_medium=0.842, AP_hard=0.727

Export CPP source code

The following bash code can export a CPP file for project libfacedetection

cd $TRAIN_ROOT/tasks/task1/
python exportcpp.py -m weights/yunet_final.pth -o output.cpp

Export to onnx model

Export to onnx model for libfacedetection/example/opencv_dnn.

cd $TRAIN_ROOT/tasks/task1/
python exportonnx.py -m weights/yunet_final.pth

Design your own model

You can copy $TRAIN_ROOT/tasks/task1/ to $TRAIN_ROOT/tasks/task2/ or other similar directory, and then modify the model defined in file: tasks/task2/yufacedetectnet.py .

Citation

The loss used in training is EIoU, a novel extended IoU. More details can be found in:

@article{eiou,
 author={Peng, Hanyang and Yu, Shiqi},
 journal={IEEE Transactions on Image Processing}, 
 title={A Systematic IoU-Related Method: Beyond Simplified Regression for Better Localization}, 
 year={2021},
 volume={30},
 pages={5032-5044},
 doi={10.1109/TIP.2021.3077144}
 }

The paper can be open accessed at https://ieeexplore.ieee.org/document/9429909.

We also published a paper on face detection to evaluate different methods.

@article{facedetect-yu,
 author={Yuantao Feng and Shiqi Yu and Hanyang Peng and Yan-ran Li and Jianguo Zhang}
 title={Detect Faces Efficiently: A Survey and Evaluations},
 journal={IEEE Transactions on Biometrics, Behavior, and Identity Science},
 year={2021}
 }

The paper can be open accessed at https://ieeexplore.ieee.org/document/9580485