Sule Bai | 白苏乐

Hi! I'm a second-year master student in Tsinghua University Shenzhen International Graduate School, supervised by Prof. Yansong Tang. Before that, I received my bachelor's degree from the School of Computer Science and Technology in the Beijing University of Posts and Telecommunications in 2023.

My currect research interest lies in Multimodal LLM and its applications.

Email  /  Github  /  Scholar

* indicates equal contribution

We propose UniVG-R1, a reasoning-guided MLLM for universal visual grounding, which leverages reinforcement learning to enhance reasoning across complex multi-image and multi-modal scenarios.

We propose a training-free method that enhances CLIP's dense representation through self-calibration without introducing new parameters or relying on additional backbones.

We propose an open-vocabulary segmentation (OVS) method by calibrating in-vocabulary and domain-biased embedding space with generalized contextual prior of CLIP.

We investigate a new problem called narrative action evaluation (NAE) and propose a prompt-guided multimodal interaction framework.

We present a new benchmark that differs significantly from the training semantics. And we propose OVSNet to improve the segmentation performance for diverse and open scenarios.