/DBFace

DBFace is a real-time, single-stage detector for face detection, with faster speed and higher accuracy

Primary LanguagePython

DBFace

DBFace is a real-time, single-stage detector for face detection, with faster speed and higher accuracy

MAP@0.5 Result on validation set of WiderFace

  • Single Scale Inference on the Original Image
Method Version Size Easy Medium Hard
RetinaFace-MobileNetV2 Small 1.68MB 0.896 0.871 0.681
DBFace-MobileNetV3-ReLU (Ours) Large 7.03MB 0.883 0.880 0.786
DBFace-MobileNetV3 (Ours) Large 7.03MB 0.905 0.896 0.794
CenterFace-MobileNetV2 Large 7.3MB ? ? ?

Result on DBFace (threshold = 0.2)

selfie


Result on RetinaFace-MobileNetV2 (threshold=0.2)

retinaface

Result on CenterFace-MobileNetV2 (threshold=?)

selfie.centerface.draw.jpg

MAP@0.5 Result on validation set of WiderFace

MAP@0.4 Result on validation set of WiderFace

Method Version Size Easy Medium Hard
DBFace-MobileNetV3 (Ours) Large 7.03MB 0.925 0.920 0.847
CenterFace-MobileNetV2 Large 7.3MB 0.922 0.911 0.782
chenjun2hao re-implementation Large 7.3MB 0.921 0.909 0.785

Speed and Train

coming soon...

Onnx And TensorRT

tensorRTIntegrate

Author

References

  1. Hamid Rezatofighi1, Generalized Intersection over Union: A Metric and A Loss for Bounding Box Regression:https://arxiv.org/abs/1902.09630

  2. Xingyi Zhou, Objects as Points:https://arxiv.org/abs/1904.07850

  3. Zili Liu, Training-Time-Friendly Network for Real-Time Object Detection:https://arxiv.org/abs/1909.00700

  4. Zhen-Hua Feng, Wing Loss for Robust Facial Landmark Localisation with Convolutional Neural Networks: https://arxiv.org/abs/1711.06753v4

  5. Mahyar Najib, SSH: Single Stage Headless Face Detector: https://arxiv.org/abs/1708.03979

  6. MobileNet: https://github.com/xiaolai-sqlai/mobilenetv3