facexlib is a pytorch-based library for face-related functions, such as detection, alignment, recognition, tracking, utils for face restorations, etc.
It only provides inference (without training).
This repo is based current STOA open-source methods (see more details).
- Python >= 3.7 (Recommend to use Anaconda or Miniconda)
- PyTorch >= 1.3
- NVIDIA GPU + CUDA
| Function | Description | Reference |
|---|---|---|
| Detection | (More details | Pytorch_Retinaface |
| Alignment | (More details | AdaptiveWingLoss |
| Recognition | (More details | InsightFace_Pytorch |
| Tracking | (More details | SORT |
| Utils | (More details) |
This project is released under the MIT license.
If you have any question, open an issue or email [email protected].