https://github.com/Krasjet-Yu/YOLO-FaceV2
YOLO-FaceV2: A Scale and Occlusion Aware Face Detector
https://github.com/Krasjet-Yu/YOLO-FaceV2
Last synced: 4 days ago
JSON representation
YOLO-FaceV2: A Scale and Occlusion Aware Face Detector
- Host: GitHub
- URL: https://github.com/Krasjet-Yu/YOLO-FaceV2
- Owner: Krasjet-Yu
- Created: 2022-08-02T18:26:09.000Z (over 2 years ago)
- Default Branch: master
- Last Pushed: 2024-02-20T14:03:15.000Z (about 1 year ago)
- Last Synced: 2024-02-20T15:28:10.213Z (about 1 year ago)
- Language: Jupyter Notebook
- Homepage:
- Size: 2.26 MB
- Stars: 145
- Watchers: 2
- Forks: 24
- Open Issues: 32
-
Metadata Files:
- Readme: README.md
Awesome Lists containing this project
- awesome-face-detection-and-recognition - YOLO-FaceV2 - Yu/YOLO-FaceV2?style=social"/> : "YOLO-FaceV2: A Scale and Occlusion Aware Face Detector ". (**[arXiv 2022](https://arxiv.org/abs/2208.02019)**). "微信公众号「江大白」《[超越Yolo5-Face，Yolo-Facev2开源，各类Trick优化，值得学习！](https://mp.weixin.qq.com/s?__biz=Mzg5NzgyNTU2Mg==&mid=2247498561&idx=1&sn=b7ff0592644ab6bc5b716e07294e1c0a&source=41#wechat_redirect)》" (Face Detection)
- awesome-yolo-object-detection - YOLO-FaceV2 - Yu/YOLO-FaceV2?style=social"/> : "YOLO-FaceV2: A Scale and Occlusion Aware Face Detector ". (**[arXiv 2022](https://arxiv.org/abs/2208.02019)**). "微信公众号「江大白」《[超越Yolo5-Face，Yolo-Facev2开源，各类Trick优化，值得学习！](https://mp.weixin.qq.com/s?__biz=Mzg5NzgyNTU2Mg==&mid=2247498561&idx=1&sn=b7ff0592644ab6bc5b716e07294e1c0a&source=41#wechat_redirect)》" (Object Detection Applications)
- awesome-yolo-object-detection - YOLO-FaceV2 - Yu/YOLO-FaceV2?style=social"/> : "YOLO-FaceV2: A Scale and Occlusion Aware Face Detector ". (**[arXiv 2022](https://arxiv.org/abs/2208.02019)**). "微信公众号「江大白」《[超越Yolo5-Face，Yolo-Facev2开源，各类Trick优化，值得学习！](https://mp.weixin.qq.com/s?__biz=Mzg5NzgyNTU2Mg==&mid=2247498561&idx=1&sn=b7ff0592644ab6bc5b716e07294e1c0a&source=41#wechat_redirect)》" (Applications)
README
# YOLO-FaceV2
### Introduction
YOLO-FaceV2: A Scale and Occlusion Aware Face Detector
*[Pattern Recognition](https://www.sciencedirect.com/science/article/pii/S0031320324004655)* or *[arxiv](https://arxiv.org/abs/2208.02019)*### Framework Structure
### Requirments
Create a Python Virtual Environment.
```shell
conda create -n {name} python=x.x
```Enter Python Virtual Environment.
```shell
conda activate {name}
```Install pytorch in *[this](https://pytorch.org/get-started/previous-versions/)*.
```shell
pip install torch==1.10.0+cu111 torchvision==0.11.0+cu111 torchaudio==0.10.0 -f https://download.pytorch.org/whl/torch_stable.html
```Install other python package.
```shell
pip install -r requirements.txt
```
### Step-Through Example
#### Downloaded Dataset
```shell
bash data/scripts/get_widerface.sh
```#### Dataset
Download the [WIDER FACE](http://shuoyang1213.me/WIDERFACE/) dataset. Then convert it to YOLO format.
```shell
# You can modify convert.py and voc_label.py if needed.
python3 data/convert.py
python3 data/voc_label.py
``````shell
cd data
python3 train2yolo.py /path/to/original/widerface/train [/path/to/save/widerface/train]
python3 val2yolo.py /path/to/original/widerface [/path/to/save/widerface/val]
```## Preweight
| Model | Easy | Medium | Hard | PreWeight |
|------------------|-------|--------|-------|------------|
| YOLO-Facev2n | 0.936 | 0.921 | 0.812 | [yolo-facev2n.pt](https://github.com/Krasjet-Yu/YOLO-FaceV2/releases/download/v2.1.0/yolo-facev2m-preweight.pt) |
| YOLO-FaceV2s | 0.983 | 0.970 | 0.893 | [yolo-facev2s.pt](https://github.com/Krasjet-Yu/YOLO-FaceV2/releases/download/v2.1.0/yolo-facev2s-preweight.pt) |
| YOLO-FaceV2m | 0.984 | 0.973 | 0.905 | [yolo-facev2m.pt](https://github.com/Krasjet-Yu/YOLO-FaceV2/releases/download/v2.1.0/yolo-facev2m-preweight.pt) |
| **YOLO-FaceV2l** | 0.986 | 0.979 | 0.919 | [yolo-facev2l.pt](https://github.com/Krasjet-Yu/YOLO-FaceV2/releases/download/v2.1.0/yolo-facev2l-preweight.pt) |#### Training
```shell
python train.py --weights preweight.pt
--data data/WIDER_FACE.yaml
--cfg models/yolov5s_v2_RFEM_MultiSEAM.yaml
--batch-size 32
--epochs 250
```#### Evaluate
```shell
python3 test_widerface.py --weights 'your test model' --img-size 640
cd widerface_evaluate/
python evaluation.py --pred ./widerface_txt_x
```
### Visualize
Visualization of small-scale face detection effect:
Visualization of the heat map effect of attention covering human faces:
### Finetune
see in *[https://github.com/ultralytics/yolov5/issues/607](https://github.com/ultralytics/yolov5/issues/607)*
```shell
# Single-GPU
python train.py --epochs 10 --data coco128.yaml --weights yolov5s.pt --cache --evolve# Multi-GPU
for i in 0 1 2 3 4 5 6 7; do
sleep $(expr 30 \* $i) && # 30-second delay (optional)
echo 'Starting GPU '$i'...' &&
nohup python train.py --epochs 10 --data coco128.yaml --weights yolov5s.pt --cache --device $i --evolve > evolve_gpu_$i.log &
done# Multi-GPU bash-while (not recommended)
for i in 0 1 2 3 4 5 6 7; do
sleep $(expr 30 \* $i) && # 30-second delay (optional)
echo 'Starting GPU '$i'...' &&
"$(while true; do nohup python train.py... --device $i --evolve 1 > evolve_gpu_$i.log; done)" &
done
```### Reference
*[https://github.com/ultralytics/yolov5](https://github.com/ultralytics/yolov5)*
*[https://github.com/deepcam-cn/yolov5-face](https://github.com/deepcam-cn/yolov5-face)*
*[https://github.com/open-mmlab/mmdetection](https://github.com/open-mmlab/mmdetection)*
*[https://github.com/dongdonghy/repulsion_loss_pytorch](https://github.com/dongdonghy/repulsion_loss_pytorch)*### Cite
If you think this work is helpful for you, please cite
```shell
@article{YU2024110714,
title = {YOLO-FaceV2: A scale and occlusion aware face detector},
journal = {Pattern Recognition},
pages = {110714},
year = {2024},
issn = {0031-3203},
url = {https://www.sciencedirect.com/science/article/pii/S0031320324004655},
author = {Ziping Yu and Hongbo Huang and Weijun Chen and Yongxin Su and Yahui Liu and Xiuying Wang},
keywords = {Face detection, YOLO, Scale-aware, Occlusion, Imbalance problem},
}```
### Contact
We use code's license is MIT License. The code can be used for business inquiries or professional support requests.