Xiaoqian Shen
About Me
I am currently a PhD student of Computer Science at King Abdullah University of Science and Technology (KAUST) supervised by Mohamed Elhoseiny. Before that, I received BSc in Computer Science from Jilin University, China. I have been fortunate to develop research and industry experience through internships at Meta and Nvidia.
My publications have received 8,389 citations on Google Scholar. I have 6 first-author papers at top-tier venues, including 2× CVPR, 1× ECCV, 1× ICML, 1× NeurIPS (Spotlight), and 1× TMLR.
Research Interests
-
Vision-Language
Multimodal understanding and long video reasoning
VisReflect (ECCV ’26) Zoom-Zero (TMLR) Vgent (NeurIPS ’25 · Spotlight) LongVU (ICML ’25) MiniGPT-4 (ICLR ’24)
-
Generative Models
Image and video generation
- Jun 2026Two papers (VisReflect, HyGAE) get accepted to ECCV’26 🎉
- Sep 2025One paper (Vgent) gets accepted to NeurIPS’25 (Spotlight) 🎉
- May 2025One paper (LongVU) gets accepted to ICML’25 🎉
- Feb 2025One paper (StoryGPT-V) gets accepted to CVPR’25 🎉
- Jul 2024Two papers (GoldFish, AffectVisDial) get accepted to ECCV’24
- Mar 2024One paper (HyperCGAN) gets accepted to CVPR’24
- Jan 2024One paper (MiniGPT-4) gets accepted to ICLR’24
- Nov 2023Successfully defended my Master thesis
- Jul 2023One paper (HRS-Bench) gets accepted to ICCV’23
- Feb 2023One paper (MoStGAN-V) gets accepted to CVPR’23
- Sep 2022Started my Master journey at KAUST
- Jul 2022One paper (HGR-Net) gets accepted to ECCV’22
- Dec 2021Joined Vision-CAIR at KAUST as a visiting research student
Experience
-
–
Tsinghua UniversityResearch Assistant -
–
KAUSTVisiting Research Student -
–
MetaResearch Scientist Intern -
–
NvidiaResearch Intern -
–
Meta Most recentResearch Scientist Intern
Selected Publications [Google Scholar]
-
TMLRZoom-Zero: Reinforced Coarse-to-Fine Video Understanding via Temporal Zoom-in
TMLR & ICLR 2026 Workshop Work done at NVIDIA -
NeurIPSVgent: Graph-based Retrieval-Reasoning-Augmented Generation For Long Video Understanding
NeurIPS 2025 Spotlight (Top 3%) -
ICMLLongVU: Spatiotemporal Adaptive Compression for Long Video-Language Understanding
ICML 2025 Work done at Meta -
ICLRMiniGPT-4: Enhancing Vision-language Understanding with Advanced Large Language Models
ICLR 2024 Github 25k+ Stars, 4k+ Cites
Services
Conference Reviewers
- CVPR, ECCV, AAAI, ICLR, ICCV
- SIGGRAPH Asia, NeurIPSW
Journal Reviewers
- IJCV, CVIU, TMM, TPAMI
Teaching Assistant
- KAUST CS 283 Deep Generative Modeling
Powered by Jekyll and Minimal Light theme.