Zeqian Li | Shanghai Jiao Tong University

About Me

I am a third-year PhD candidate at Shanghai Jiao Tong University (SJTU), advised by Prof. Weidi Xie. I received my Bachelor degree in Information Engineering, also from Shanghai Jiao Tong University.

My research focuses on long video understanding and multi-modal representation learning.

News

2026.07 Two papers are accepted by ACM MM 2026.
2025.09 UniTime is accepted by NeurIPS 2025.
2025.06 StreamFormer is accepted by ICCV 2025.
2024.08 HowToStep is accepted by ECCV 2024.

Publications

ACM MM (Oral)

SpotSound: Enhancing Large Audio-Language Models with Fine-Grained Temporal Grounding

Luoyi Sun, Xiao Zhou, Zeqian Li, Ya Zhang, Yanfeng Wang, Weidi Xie

ACM International Conference on Multimedia (ACM MM) Oral, 2026.

PDF Code Project Page

Utilize large audio-language models for audio temporal grounding and construct a challenging needle-in-a-haystack benchmark.
ACM MM (Oral)

DenseStep2M: A Scalable, Training-Free Pipeline for Dense Instructional Video Annotation

Mingji Ge, Qirui Chen, Zeqian Li, Weidi Xie

ACM International Conference on Multimedia (ACM MM) Oral, 2026.

PDF

An automated, training-free pipeline to extract high-quality procedural annotations from in-the-wild instructional videos.
NeurIPS

Universal Video Temporal Grounding with Generative Multi-modal Large Language Models

Zeqian Li, Shangzhe Di, Zhonghua Zhai, Weilin Huang, Yanfeng Wang, Weidi Xie

Advances in Neural Information Processing Systems (NeurIPS), 2025.

PDF Code Project Page

Towards universal video grounding with superior accuracy, generalizability, and robustness.
ICCV

Learning Streaming Video Representation via Multitask Training

Yibin Yan*, Jilan Xu*, Shangzhe Di, Yikun Liu, Yudi Shi, Qirui Chen, Zeqian Li, Yifei Huang, Weidi Xie

IEEE/CVF International Conference on Computer Vision (ICCV), 2025.

PDF Code Project Page

Learn streaming video representations of various granularities through multitask training, including retrieval, action recognition, temporal grounding, and segmentation.
ECCV

Multi-Sentence Grounding for Long-term Instructional Video

Zeqian Li*, Qirui Chen*, Tengda Han, Ya Zhang, Yanfeng Wang, Weidi Xie

European Conference on Computer Vision (ECCV), 2024.

PDF Code Project Page

An automatic, scalable pipeline for denoising the large-scale instructional dataset and construct a high-quality video-text dataset with multiple descriptive steps supervision.