A PyTorch Implementation of PyramidBox
I train pyramidbox with pytorch and the result approaches the original paper result,the pretrained model can be downloaded in vgg,the final model can be downloaded in Pyramidbox.the AP in WIDER FACE as following:
Easy MAP | Medium MAP | hard MAP | |
---|---|---|---|
origin paper | 0.960 | 0.948 | 0.888 |
this repo | 0.948 | 0.938 | 0.880 |
the AP in AFW,PASCAL,FDDB as following:
AFW | PASCAL | FDDB |
---|---|---|
99.65 | 99.02 | 0.983 |
the gap is small with origin paper,I train 120k batch_size 4 which is different from paper,which maybe cause the gap,if you have more gpu ,the final result maybe better.
- pytorch 0.3
- opencv
- numpy
- easydict
- download WIDER face dataset
- modify data/config.py
python prepare_wider_data.py
python train.py --batch_size 4
--lr 5e-4
according to yourself dataset path,modify data/config.py
- Evaluate on AFW.
python tools/afw_test.py
- Evaluate on FDDB
python tools/fddb_test.py
- Evaluate on PASCAL face
python tools/pascal_test.py
- test on WIDER FACE
python tools/wider_test.py
you can test yourself image
python demo.py