AI Paper Atlas

Research map / Vision-language and multimodal / Multimodal LLMs

GeoProg3D: Compositional Visual Reasoning for City-Scale 3D Language Fields

Shunsuke Yasuki, Taiki Miyanishi, Nakamasa Inoue, Shuhei Kurita, Koya Sakamoto, Daichi Azuma, Masato Taki, Yutaka Matsuo

ICCV 2025 · Vision-language and multimodal › Multimodal LLMs · approach: Spatial reasoning in VLMs · 3 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
add GV-APIs and visual programming on top of the GCLF language fieldGCLF · GeoEval3D GoogleEarth GRDlocalization accuracy (%): 20.09 -> 45.20Improved
add GV-APIs and visual programming on top of the GCLF language fieldGCLF · GeoEval3D UrbanScene3D GRDlocalization accuracy (%): 6.98 -> 30.23Improved
use GCLF (tree-structured, georeferenced language field) instead of LangSplat (ablation)LangSplat · GeoEval3D GoogleEarth GRDlocalization accuracy (%): 14.15 -> 20.09Improved
use GCLF instead of LangSplat on the 5.0x10^6 m^2 scene (ablation)LangSplat · GeoEval3D UrbanScene3Dlocalization: OOM -> 6.98Improved
add GCLF's tree structure (more Gaussians) versus LangSplat (ablation)GCLF · rendering speedrendering time (ms): 2.73 -> 14.46Hurt
add GetLandmarkSeg module (ablation)GeoProg3D · GeoEval3D GoogleEarth GRDaccuracy (%): 6.26 -> 45.20Improved
add GetLandmarkSeg module (ablation)GeoProg3D · GeoEval3D GoogleEarth CMPaccuracy (%): 0.00 -> 59.73Improved
add SegDirection module (ablation)GeoProg3D · GeoEval3D GoogleEarth GRDaccuracy (%): 26.01 -> 45.20Improved
add SegAround module (ablation)GeoProg3D · GeoEval3D GoogleEarth SPRaccuracy (%): 36.76 -> 64.00Improved
add SegBetween module (ablation)GeoProg3D · GeoEval3D GoogleEarth CMPaccuracy (%): 60.53 -> 59.73Mixed
add LargestSeg module (ablation)GeoProg3D · GeoEval3D GoogleEarth CMPaccuracy (%): 44.74 -> 59.73Improved
add LargestSeg module (ablation)GeoProg3D · GeoEval3D GoogleEarth GRDaccuracy (%): 45.20 -> 45.20No clear effect

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 3 of 3 on most shared results.

Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy