4 papers
DexFormer: Cross-Embodied Dexterous Manipulation via History-Conditioned Transformer
Ke Zhang, Lixin Xu, Chengyi Song +4
Dexterous manipulation remains one of the most challenging problems in robotics, requiring coherent control of high-DoF hands and arms under complex, contact-rich dynamics. A major…
TBAC-UniImage: Unified Understanding and Generation by Ladder-Side Diffusion Tuning
Junzhe Xu, Yuyang Yin, Xi Chen
This paper introduces TBAC-UniImage, a novel unified model for multimodal understanding and generation. We achieve this by deeply integrating a pre-trained Diffusion Model, acting…
Simple Semantic-Aided Few-Shot Learning
Hai Zhang, Junzhe Xu, Shanlin Jiang +1
Learning from a limited amount of data, namely Few-Shot Learning, stands out as a challenging computer vision task. Several works exploit semantics and design complicated semantic…
Attribute Localization and Revision Network for Zero-Shot Learning
Junzhe Xu, Suling Duan, Chenwei Tang +2
Zero-shot learning enables the model to recognize unseen categories with the aid of auxiliary semantic information such as attributes. Current works proposed to detect attributes f…