YUNet Face and 5 Point Landmark Detection implementation using PyTorch

Installation

conda create -n YUNet python=3.11.11
conda activate YUNet
conda install python=3.11.11 pytorch torchvision torchaudio pytorch-cuda=12.1 -c pytorch -c nvidia
pip install opencv-python
pip install PyYAML
pip install tqdm

Train

• Configure your dataset path in main.py for training
• Run bash main.sh $ --train for training, $ is number of GPUs

Test

• Configure your dataset path in main.py for testing
• Run python main.py --test for testing

Results

Model	AP_easy	AP_medium	AP_hard	#Params	Params Ratio	MFlops (320x320)	FPS(320x320)
SCRFD0.5(ICLR2022)	0.892	0.885	0.819	631,410	8.32x	184	284
Retinaface0.5(CVPR2020)	0.907	0.883	0.742	426,608	5.62X	245	235
YuNet_n(Original)	0.892	0.883	0.811	75,856	1.00x	149	456
YuNet_n(Ours)	0.896	0.887	0.818	72,928	1.00x	133	456

Dataset structure

├── WIDERFace 
    ├── images
        ├── train
            ├── 1111.jpg
            ├── 2222.jpg
        ├── val
            ├── 1111.jpg
            ├── 2222.jpg
    ├── labels
        ├── train
            ├── 1111.txt
            ├── 2222.txt
        ├── val
            ├── 1111.txt
            ├── 2222.txt

Evaluation on WIDERFace Dataset

• https://github.com/jahongir7174/YUNet-eval

Reference

• https://github.com/jahongir7174/YUNet-eval
• https://github.com/ShiqiYu/libfacedetection.train