Seungwoo (Simon) Kim

I am a graduate student researcher at the Stanford NeuroAILab, advised by Professor Dan Yamins. I am pursuing a concurrent BS and MS at Stanford University, with a focus in Computer Systems and AI.

My research centers on computer vision and machine learning, particularly in developing models with complex visual structure and scene understanding.

Email  /  Github  /  Resume

profile photo

Research

Taming generative video models for zero-shot optical flow extraction
Seungwoo Kim*, Khai Loong Aw*, Klemen Kotar*, Cristobal Eyzaguirre, Wanhee Lee, Yunong Liu, Jared Watrous, Stefan Stojanov, Juan Carlos Niebles, Jiajun Wu, Daniel LK Yaminsa
arXiv, 2025
project page / arXiv

A novel test-time procedure that uses the Kullback-Leibler (KL) divergence of prediction logits for zero-shot extraction of optical flow from a generative video model without any additional task-specific fine-tuning.

BountyBench: Dollar Impact of AI Agent Attackers and Defenders on Real-World Cybersecurity Systems
Andy K. Zhang*, Joey Ji*, Celeste Menders*, Riya Dulepet*, Thomas Qin*, Ron Y. Wang*, Junrong Wu*, Kyleen Liao*, Jiliang Li*, Jinghan Hu, Sara Hong, Nardos Demilew, Shivatmica Murgai, Jason Tran, Nishka Kacheria, Ethan Ho, Denis Liu, Lauren McLane, Olivia Bruvik, Dai_rong Han, Seungwoo Kim, Akhil Vyas, Cuiyuanxiu Chen, Ryan Li, Weiran Xu, Jonathan Z. Ye, Prerit Choudhary, Siddharth M. Bhatia, Vikram Sivashankar, Yuxuan Bao, Dawn Song, Dan Boneh, Daniel E. Ho, Percy Liang
arXiv, 2025
project page / arXiv

A benchmark with 25 systems with complex, real-world codebases, and include 40 bug bounties that cover 9 of the OWASP Top 10 Risks. The first framework to capture offensive & defensive cyber-capabilities in evolvigin real-world systems.

Self-Supervised Learning of Motion Concepts by Optimizing Counterfactuals
Stefan Stojanov*, David Wendt*, Seungwoo Kim*, Rahul Venkatesh*,
Kevin Feigelis, Jiajun Wu, Daniel LK Yamins
arXiv, 2025
project page / arXiv

State-of-the-art self-supervised point tracking and occlusion estimation with optimized Counterfactual World Models (Opt-CWM).

* => (Equal Contribution)

Source: Jon Barron -> template