👋 About Me

I am a second-year master student at Shenzhen International Graduate School, Tsinghua University. I am fortunate to be supervised by Prof. Yansong Tang in IVG@SZ group. Before that, I got B.S. in Electric and Electronic Engineering from the University of Electronic Science and Technology of China (UESTC) in 2024.

My research interests lie in Computer Vision, such as Lagre Vision-Language Model, Tool-calling, Multimodal Learning, Segmentation, and Tracking.

Email / Google Schoolar

✨ News

2025-12: One paper on Tool-Refined Visual Grounding is available on arXiv
2025-02: One paper on Triple Modalality Referring Segmentation is accepted to CVPR 2025
2024-12: One paper on Referring Image Segmentation is accepted to AAAI 2025
2024-07: One paper on Multimodal Learning is accepted to ECCV 2024

🔬 Research

	VG-Refiner: Towards Tool-Refined Referring Grounded Reasoning via Agentic Reinforcement Learning Yuji Wang, Wenlong Liu, Jingxuan Niu, Haoji Zhang, Yansong Tang arXiv preprint, 2025 [PDF] [Project Page] We propose VG-Refiner, the first framework for tool-refined referring grounded reasoning with a two-stage think-rethink mechanism and refinement reward to handle unreliable tool outputs.
	SAM2-LOVE: Segment Anything Model 2 in Language-aided Audio-Visual Scenes Yuji Wang, Haoran Xu, Yong Liu, Jiaze Li, Yansong Tang IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2025 [PDF] [Project Page] We propose a novel framework called SAM2-LOVE to effectively segment the video objects referred by the audio and text and achieve significant improvement in Ref-AVS tasks.
	IteRPrimE: Zero-shot Referring Image Segmentation with Iterative Grad-CAM Refinement and Primary Word Emphasis Yuji Wang, Jingchen Ni, Yong Liu, Chun Yuan, Yansong Tang AAAI Conference on Artificial Intelligence (AAAI), 2025 [PDF] [Project Page] We propose the novel IteRPrimE network to leverage the Grad-CAM for zero-shot referring image segmentation, which addresses the previous CLIP-based methods' low robustness of positional phrases.
	Robust Multimodal Learning via Representation Decoupling Shicai Wei, Yang Luo, Yuji Wang, Chunbo Luo European Conference on Computer Vision (ECCV), 2024 [PDF] [Project Page] We propose DMRNet improves multimodal learning with missing modalities by modeling inputs as probabilistic distributions to capture modality-specific information, outperforming state-of-the-art methods.