Separate from AI_power, consists methods/onnx-models for face related missions.
Examples can be found in examples.
Models can be found in huggingface.
- eye open detect from Open-eye-closed-eye-classification
- iris detect from iris-detection
- based on mmclassification, already convert to onnx.
- supply pretrained model trained by private dataset, mind the face should be aligned first.
- ffhq align method
- face alignment from 1adrianb
- conform multi similarity align methods
- mtcnn from mtcnn-pytorch
- scrfd from insightface
- s3fd from DeepFaceLab
- Arcface from SimSwap (mod to jit model)
- CurricularFace (mod to jit model)
- pfpld from nniefacelib
- face-parsing.PyTorch, onnx converted
- DFL Xseg, based on opensource, self trained (private data)
- Gpen
- DFDNet (add batch parallel support)
- GFPGAN onnx converted
- RestoreFormer
- CodeFormer
- Mouth Open/Close Detect, private model