Google brings 41 accepted papers to ECCV 2026, from video spatial understanding to TIPSv2
ymatias · x · 2026-09-11
Google Research and Google DeepMind are at ECCV 2026 in Malmö (Sept 8–12) as a Diamond Sponsor, with 41 accepted papers, participation in 47 workshops and tutorials, and live booth demos.
Highlights include:
- A booth session on "Perception, Not Reasoning, Limits Video Spatial Understanding," probing whether spatial perception or reasoning is the bottleneck for video foundation models
- TIPSv2: advancing vision-language pretraining with enhanced patch-text alignment for faster multimodal training
More from Multimodal
- G.I. Joe scene remade with MiniMax H3, now with 100% more bullet holes — darthfurbyyoutube · 2026-09-11
- Nari Labs launches 50ms TTS endpoint, 10x cheaper than ElevenLabs on Qwen3-TTS — iamaliveix · 2026-09-11
- MiniMax H3 text-to-video test unexpectedly nails 90s anime meets old Disney style — Mr__Earthling · 2026-09-11
- Midjourney srefs behave differently by aspect ratio — 1:3 nails subtle painterly styles — Kyrannio · 2026-09-11
- "I gave my computer synesthesia" — a delightful real-time demo — floguo · 2026-09-11
- AI movie studio builds 3D camera previs tool to save generation credits — purecharisma2020 · 2026-09-11