Jaemin Cho @ ECCV 2026 🇸🇪

@jmin__cho

Assistant Prof @JHUCompSci | Prev @Allen_AI PhD @UNCCS | multimodal and embodied AI

Baltimore, MD

j-min.io

Replies Reposts Media

Pinned
Jaemin Cho @ ECCV 2026 🇸🇪
@jmin__cho
Sep 6
I'm attending #ECCV2026 🇸🇪 Sep 08-11! Let me know if you know of a fun event or wanna chat in Malmö / Copenhagen. You can also find me 1) giving keynotes at workshops - Workshop on Multimodal LLMs for Unified Comprehension and Generation (Sep 8) mllm-mucg.github.io/ECCV2026/ -
2
Jaemin Cho @ ECCV 2026 🇸🇪
@jmin__cho
Jun 29
Attending #ICML2026 in Seoul!🇰🇷 Looking forward to catching up and meeting new friends. DM me if you want to chat or know of fun events!
2
Jaemin Cho @ ECCV 2026 🇸🇪
@jmin__cho
Jun 25
Look closely at modern AI video generators, and you'll notice subtle (and not-so-subtle) physics slip-ups that standard metrics completely miss. To pinpoint these errors, we introduce our #ECCV2026 work, Physics Question Scene Graph (PQSG), a framework that evaluates
Yue Zhang
@zhan1624
Jun 25
AI video looks incredible until something moves. Paper towels dissolve instead of soaking up water. Balls bounce off pillows like they're on a trampoline. Gravity stops being a law and starts being a suggestion. Physics is the thing AI video still struggles with. We introduce
00:00
1
Jaemin Cho @ ECCV 2026 🇸🇪
@jmin__cho
Jun 2
Can LLMs predict GPU kernel runtimes instead of measuring them on actual hardware? We find that: - LLMs act as great selective surrogates (deferring to GPUs when unsure) - RL improves LLM accuracy & calibration - Kernel search becomes much more efficient We're releasing 12K
Zaid Khan
@codezakh
Jun 2
Can an LLM act as a selective model of a GPU during evolutionary search, by reasoning + forecasting a kernel’s runtime but deferring to a GPU when unsure? We produced 12k kernels + runtimes from evolutionary search, costing 400M reasoning tokens + 600 GPU-hours to answer this.
2
Jaemin Cho @ ECCV 2026 🇸🇪
@jmin__cho
May 15
Introducing PhyMotion! You can now RL-tune your video generator with Real2Sim2Real reward. PhyMotion is our structured reward for human video generation. It lifts SMPL human motion from video and calculates physical feasibility scores across fine-grained dimensions within
Yidong Huang
@owenhuang117
May 15
🚨 Excited to introduce PhyMotion🤸: Structured 3D Motion Reward for Physics-Grounded Human Video Generation! ❌ Existing 2D video rewards misleadingly assign high scores to videos with floating feet, self-penetrating limbs, and physics-violating motions. ✅ PhyMotion lifts