Research
AIGC
This research area is about generate diverse and high-fidelity data. Some methods focus on the generation quality, while others further focus on the recognition performance based on the generated data.
- 2D Human ICCV’17, ECCV’18, TIP’18, CVPR’19, TMM’22
- 2D to 3D Human npj AI’26
- Text to Human ACM MM’23
- Text to 3D TOMM’26
- Face IJCAI’20, TCYB’20, CVPR’22
- Sign Language TOMM’24
- Hand-Object Interaction ICRA’26
- Painting ACM MM’25 BNI
Uncertainty Estimation
- Large Models arXiv’24, ICLR’26
- Domain Adaptation IJCV’21
- Depth Completion PR’24
- Composed Retrieval ICLR’24
- Active Learning CVPR’23
- 3D Bbox Detection ICCV’25
- Conditional Generation ICLR’25
Object Re-identification
This research area is to retrieval the relevant multi-media data including cross-view, cross-modality or other challenging cases.
- 2D Human TOMM’18, TCSVT’19, PR’19, AAAI’21, TPAMI’22, TCYB’22, TMM’23
- 2D Vehicle TMM’20, CVPRW’21
- 3D Human / Objects TNNLS’22, TMM’22
- Cross Modality (Image,Text, Video) TOMM’20, CVPRW’21, TMM’22, ACM MM’23, ICLR’24
- Sign Language TOMM’24
- Adversarial Examples IJCV’22
Domain Adaptation
This research area is to minimize the gap between the generated data / simulation system with the real-world data / scenario during deployment.