news

Check our new paper TimeLens: Rethinking Video Temporal Grounding with Multimodal LLMs. We expose critical quality issues in existing VTG benchmarks and propose quality-assured datasets for both training and evaluation. Building upon reliable data, we explore algorithmic designs and release state-of-the-art open-source grounding models. [Website]

p-MoD is accepted by ICCV 2025!

We present p-MoD, efficient MLLMs built with Mixture-of-Depths mechanism.

I am officially enrolled as a master’s student!