📝 Publications

TPAMI 2025 Fine-Grained Visual Text Prompting
Lingfeng Yang, Xiang Li, Yueze Wang, Xinlong Wang, Jian Yang

Paper | Code

• Proposes fine-grained multimodal prompting to enhance large multimodal models’ localization and grounding capability, thereby boosting referring comprehension performance.
• Our work has been adopted by the research group of Prof. Philip H. S. Torr (Oxford University, Marr Prize laureate), who employed the proposed Fine-Grained Visual Prompting (FGVTP) as the core target extractor in their weakly supervised referring segmentation framework.
• Our work has inspired subsequent studies and has been applied to multiple domains, including Egocentric Action Recognition and Compositional Action Recognition for embodied intelligence perception.

NeurIPS 2023 Fine-Grained Visual Prompting
Lingfeng Yang, Yueze Wang, Xiang Li, Xinlong Wang, Jian Yang

Paper | Code | 中文解读 | 中文视频

• Propose a specific visual prompting technique that enhances referring expression comprehension by highlighting regions of interest through background blurring based on fine-grained segmentation.
• Maintains faster inference speed in the trade-off while achieving more than a 5-point improvement over state-of-the-art methods.

NeurIPS 2022 RecursiveMix: Mixed Learning with History (Spotlight, Top 12.8%)
Lingfeng Yang, Xiang Li, Borui Zhao, Renjie Song, Jian Yang

Paper | Code

• Propose a simple yet effective mixed-data augmentation technique for image classification.
• Enhance model pretraining performance for object detection and semantic segmentation tasks.

CVPR 2022 Dynamic MLP for Fine-Grained Image Classification by Leveraging Geographical and Temporal Information (Oral, Top 3.3%)
Lingfeng Yang, Xiang Li, Renjie Song, Borui Zhao, Juntian Tao, Shihao Zhou, Jiajun Liang, Jian Yang

Paper | Code

• Proposed a dynamic MLP fusion framework for fine-grained image classification by incorporating geo-temporal information.
• Improved classification accuracy on multiple fine-grained datasets.