PSVT: End-to-End Multi-Person 3D Pose and Shape Estimation with Progressive Video Transformers.
Zhongwei QiuQiansheng YangJian WangHaocheng FengJunyu HanErrui DingChang XuDongmei FuJingdong WangPublished in: CVPR (2023)
Keyphrases
- end to end
- shape estimation
- scalable video
- shape from shading
- video data
- pose estimation
- video sequences
- congestion control
- shape modeling
- inverse kinematics
- video streams
- multimedia
- video content
- position and orientation
- real time
- inverse problems
- d objects
- structure from motion
- partial occlusion
- real objects
- global optimization
- computational complexity