UPenn Open-Sources HARMONY for Single-Image 3D Scene Reconstruction

UPenn's Lingjie Liu team open-sourced HARMONY, which combines hierarchical agentic reasoning with vision-geometry foundation models to reconstruct compositional 3D scenes from a single indoor image, released alongside the HARMONY300 benchmark and claimed to rival GPT-6 Astra.

2026-09-24 ~ 2026-09-24 · 3 related posts

• Open-source HARMONY uses VLM agents to reconstruct 3D scenes, matching GPT-6 Astra quality — thoma_gu · 2026-09-24
• Penn's open-source HARMONY rebuilds 3D scenes from one image, matching GPT-6 Astra — thoma_gu · 2026-09-24
• UPenn's HARMONY reconstructs 3D indoor scenes from a single image via hierarchical agentic reasoning — upenn · 2026-09-24