Loading video...
Video Failed to Load
DeWorldSG: Depth-Aware 3D Semantic Scene Graph Generation via World-Model Priors ECCV 2026 TL; DR DeWorldSG generates spatio-temporally consistent 3D semantic scene graphs from RGB-D sequences by combining depth-aware probabilistic object modeling with world-model-guided relation reasoning.
31,089 views • 3 months ago •via X (Twitter)
6 Comments

Ryohei Sasaki@engineer3 months ago
Linkedin Gihub io

VAI3 months ago
The tldr is longer than the initial explanation 😆

Matt Radford3 months ago
Realtime?

Pathetic Brook3 months ago
I could’ve used this 2 days ago 😭

qiuwei0013 months ago
可交互吗

Turner NextGen AI3 months ago
Sometimes a lower-level measurement is useful because it helps explain a higher-level organizational problem. But it should be requested because it matters to the function, not collected automatically because the system knows how to measure it. Measure what the reasoning requires. Do not force the reasoning to follow whatever you happened to measure.
