DeWorldSG: depth-aware 3D scene graph generation

Loading video
Loading videoDeWorldSG (ECCV 2026) generates spatio-temporally consistent 3D semantic scene graphs from RGB-D sequences by combining depth-aware probabilistic object modeling with world-model-guided priors.
Category: perception
Author: @rsasaki0109
Date: 2026-08-13T00:00:00
Duration: 75.633s
Reference: https://arxiv.org/abs/2607.00889





