Face Detection on WIDER FACE 17 (val)
90.72AP (Easy)Quantized RetinaFace
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| Quantized RetinaFaceBackbone=MobileNetV1-0.25, Training=Pre-trained, Context Module (CM)=true, Depthwise and Dilated Convolutions (DDC)=false, Quantized=true2021.04 | 90.72 | 87.45 | 73.56 | |
| RetinaFaceBackbone=MobileNetV1-0.25, Training=Pre-trained, Context Module (CM)=true, Depthwise and Dilated Convolutions (DDC)=false, Quantized=false2021.04 | 90.7 | 87.88 | 73.5 | |
| Quantized RetinaFaceBackbone=MobileNetV1-0.25, Training=Pre-trained, Context Module (CM)=true, Depthwise and Dilated Convolutions (DDC)=true, Quantized=true2021.04 | 90.32 | 87.68 | 73.53 | |
| RetinaFaceBackbone=MobileNetV1-0.25, Training=Pre-trained, Context Module (CM)=true, Depthwise and Dilated Convolutions (DDC)=true, Quantized=false2021.04 | 90.28 | 87.13 | 73.24 | |
| RetinaFaceBackbone=MobileNetV1-0.25, Training=Scratch, Context Module (CM)=true, Depthwise and Dilated Convolutions (DDC)=false, Quantized=false2021.04 | 89.98 | 87.11 | 72.01 | |
| Quantized RetinaFaceBackbone=MobileNetV1-0.25, Training=Scratch, Context Module (CM)=true, Depthwise and Dilated Convolutions (DDC)=false, Quantized=true2021.04 | 89.7 | 86.91 | 71.89 | |
| RetinaFaceBackbone=MobileNetV1-0.25, Training=Scratch, Context Module (CM)=true, Depthwise and Dilated Convolutions (DDC)=true, Quantized=false2021.04 | 89.6 | 86.13 | 71.93 | |
| Quantized RetinaFaceBackbone=MobileNetV1-0.25, Training=Scratch, Context Module (CM)=true, Depthwise and Dilated Convolutions (DDC)=true, Quantized=true2021.04 | 89.56 | 86.02 | 71.75 | |
| RetinaFaceBackbone=MobileNetV1-0.25, Training=Pre-trained, Context Module (CM)=false, Depthwise and Dilated Convolutions (DDC)=false, Quantized=false2021.04 | 89.55 | 86.21 | 68.83 | |
| RetinaFaceBackbone=MobileNetV1-0.25, Training=Scratch, Context Module (CM)=false, Depthwise and Dilated Convolutions (DDC)=false, Quantized=false2021.04 | 88.68 | 84.98 | 68.56 |