A collection of face detection models pre-trained on the Widerface dataset. In the table below you can see each model detailed information including:
- a model name
- a download link to a
tar.gzfile containing the model and configuration files - TODO model speed
- detector performance measured on the FDDB benchmark
| Model | [email protected] | cfg/weights |
|---|---|---|
| YOLOv2 | 84.90 | link |
| Tiny YOLO | 80.04 | link |
| SSD mobilenet v1 | ? | link |
| Faster RCNN inception resnet v2 atrous | 94.39 | link |
| R-FCN resnet101 | 94.73 | link |
Morghulis was used to download and convert it to either Darknet or Tensorflow Object Detection API format.
There are 2 models trained with Darknet: one based on YOLOv2 and other on Tiny YOLO. Both used convolutional weights that are pre-trained on Imagenet: darknet19_448.conv.23.
The remaining models were trained with Tensorflow Object Detection API on Google Cloud ML Engine.
