3D Spatial Reasoning Comparison Hub
3 papers - avg viability 5.7
Top Papers
- Map2Thought: Explicit 3D Spatial Reasoning via Metric Cognitive Maps(7.0)
Map2Thought enhances 3D spatial reasoning in vision-language models, significantly outperforming state-of-the-art methods with reduced training data.
- Boosting MLLM Spatial Reasoning with Geometrically Referenced 3D Scene Representations(7.0)
Enhance MLLM spatial reasoning by encoding 3D geometric attributes as textual references, enabling zero-shot application and improved performance on spatial understanding tasks.
- Context-Nav: Context-Driven Exploration and Viewpoint-Aware 3D Spatial Reasoning for Instance Navigation(3.0)
Context-Nav enhances instance navigation by integrating contextual captions and 3D spatial reasoning for improved object identification.