All pretrain models and supply materials can be accessed by my share link:
BaiDuPan pwd: ibgg
- cv2box: A gather of tools or functions frequently using in my work.
- AI_power: Convenience API for your AI power.
- apstone: Base stone of AI_power, maintain all inference of AI_Power models.
'*' means self-trained model
Supported face related models&methods:
more details refer face_lib
(click to collapse)
- EyeOpenDetect (eye close detect)
- IrisLandmark
- Deep3DFace (face 3d)
- *MobileNetV3 (face attitrbute)
- MTCNN (face detect)
- FAN
- S3FD
- SCRFD
- ArcFace (face embedding)
- CurricularFace
- InsightFace
- PFPLD (face landmark)
- *XsegNet (face parsing)
- face-parsing.PyTorch
- GPEN (face restore)
- DFDNet
- GFPGAN
- RestoreFormer
- CodeFormer
- *HifiFace (face swap)
- InSwapper
- FaceFusion
Supported body related models&methods:
more details refer body_lib
(click to collapse)
- yolox_tiny/yolox_s (mmdetection) (body bbox)
- hrnetv2w32 (modelscope) (body keypoints)
- BlazePose
- lightweight-human-pose-estimation
- MoveNet
- kapao
Supported hand related models&methods:
more details refer hand_lib
(click to collapse)
- yolox_tiny (mmdetection) (hand 21 keypoints detect)
- hand_detector.d2 (hand bbox)
- mediapipe hands
- *yolox
- minimal-hand (hand mesh)
- frankmocap hand regressor (hand pose)
Supported mocap related models&methods:
more details refer mocap_lib
(click to collapse)
- SPIN (body regress)
- r50/vipnas_mbv3_dark/hrnet_w48_384_dark etc. (mmpose) (whole body keypoints)
- mediapipe holistic
- Calibration
- Smooth Filter
- Triangulate
Supported segmentation related models&methods:
more details refer seg_lib
(click to collapse)
- carvekit
- cihp_pgn
- ppmattingv2
- RAFT
- u2net (object segmentation)
- cv_u2net_salient-detection
Supported AIGC related models&methods:
more details refer art_lib