Yibing Song宋奕兵

About me

I am developing algorithmic applications around Multi-modal AI. Previously, I worked for Tencent, Alibaba, and Fudan University. I got my PhD/MPhil degrees from City University of Hong Kong and a bachelor degree from University of Science and Technology of China.

I was active in academic research, serving as (senior) area chairs 20+ times in CVPR/ICCV/ECCV/NeurIPS/ICML/ICLR, publishing 70+ premier papers from above venues, with 17K+ citations accumulated.

Some outdated papers that I still treasure

CoT-lized Diffusion: Let’s Reinforce T2I Generation Step-by-Step NeurIPS 2025

Using VLM to reward T2I model in each diffusion step

LLaVA-CoT: Let Vision Language Models Reason Step-by-Step ICCV 2025

First structured CoT for vision-language data with online retracing search

Re-Aligning Language to Visual Objects with an Agentic Workflow ICLR 2025

First agent for vision-language data engineering

DiffusionDET: Diffusion Model for Object Detection ICCV 2023 ★ Best Paper Nominee ★

First diffusion model for visual recognition

VideoMAE: Masked Autoencoders are Data-Efficient Learners for Self-Supervised Video Pre-Training NeurIPS 2022

Self-supervised video representation learning stuffRanked 8th in most influential NeurIPS 2022 papersRanked 39th in most cited 2022 AI papers