EgoTools: 100-hour egocentric video dataset teaches AI tool-centric reasoning
liuziwei7 · x · 2026-10-03
Researchers from NTU's S-Lab, ASTAR, KAIST and others released EgoTools, a project asking whether AI can understand why, which, and how humans use tools in real-world first-person video.
- Dataset: 100.37 hours of real egocentric video across 7 everyday domains (kitchens, labs, workshops, classrooms).
- Benchmark: 1,000 questions across four tracks, structured as ground → track → infer.
- Reference model: EgoTools-8B lifts overall accuracy from 50.0% to 60.9% (+10.9 pp) on the same 8B backbone with 64 frames.
Paper, code, data and model are all open-sourced, with a leaderboard and 3D playground.
Related event: EgoTools: 100-Hour Egocentric Dataset Tests AI Tool Reasoning(3 posts)→
More from Multimodal
- Runway-Generated Short Film 'Do You Ever Miss a Life You Never Lived?' Resonates — tlakomy · 2026-10-03
- ComfyUI v0.38 speeds up MiniMax H3 I2V by at least 16% on AMD, test shows — mwhjose · 2026-10-03
- LTX 2.3 is fast but often ignores prompts, and LoRA support is drying up — Fit_Satisfaction2953 · 2026-10-03
- Reddit user crafts anime-girl AI wingman video with Opus 5, Seedance and Nano Banana — EndlessMendless · 2026-10-03
- How real has AI video gotten? Reddit debates realism with demo footage — Solar-Guardian · 2026-10-03
- KAIST's World Observer gives world models movable panoramic eyes to track unseen regions — Scobleizer · 2026-10-03