PulseJournal ClubResearchersJournalsExplore
Instagram
HomeJournal ClubExplore
Synapse
⌘+K
Synapse
July 12, 2026ISPRS annals of the photogrammetry, remote sensing and spatial information sciencesOpen Access

GeoSceneGraph: Geometric Scene Graph Diffusion Model for Text-guided 3D Indoor Scene Synthesis

View Full Paper
Ask AI
Bookmark
Share

Authors

ARAntonio RuizTWTao WuAMAndrew Melnik

Discussion

Loading...

Member takes

Overview

Randomized trial synthesizes 3D indoor scenes from text prompts, indicating improved coherence and realism.

Key Points

  • The aim is to synthesize coherent 3D indoor scenes from text prompts while leveraging geometric structures of scenes.
  • Developed GeoSceneGraph, utilizing the graph structure and geometric symmetries of 3D scenes.
  • Incorporated Equivariant Graph Neural Networks for enhanced scene representation.
  • Conducted ablation studies to validate the effectiveness of conditioning EGNNs on text features.
  • Achieved performance comparable to existing methods that use ground-truth relationships.
  • Improved scene coherence and realism without needing user-provided semantic graphs.

Cite This Study

Ruiz et al. (2026) studied this question.

synapsesocial.com/papers/6a5331ce4f7abc118adeda29https://doi.org/10.5194/isprs-annals-xi-4-2026-293-2026
View Full Paper
Ask AI
Bookmark
Share