Research map / 3D vision / 3D generation
Rel3D: A Minimally Contrastive Benchmark for Grounding Spatial Relations in 3D
Ankit Goyal, Kaiyu Yang, Dawei Yang, Jia Deng
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 changed | Where | Measured | Effect |
|---|---|---|---|
| Train on minimally contrastive pairs rather than non-contrastive examples (ablation) | DRNet, VTransE, and 3D-feature models · Rel3D | sample efficiency: only about $1/4 training data | Improved |
| Use a simple 2D bounding-box baseline instead of sophisticated visual relationship models | Rel3D spatial relation recognition · Rel3D | average class accuracy | Improved |
| Use language-only input instead of visual or geometric input | Rel3D spatial relation recognition · Rel3D | accuracy: 50% | No clear effect |
| Use aligned 3D features instead of raw 3D features (ablation) | MLP spatial relation classifier · Rel3D | accuracy: 0.803 vs 0.682 (behind); 0.698 vs 0.708 (passing through) | Mixed |
| Use 3D position, pose, and scale features | MLP spatial relation classifier · Rel3D | spatial relation recognition | Improved |
| Rely on 3D features alone as a complete solution | MLP spatial relation classifier · Rel3D | spatial relation recognition | Hurt |
Closest papers by meaning
- Surprise3D: A Dataset for Spatial Understanding and Reasoning in Complex 3D Scenes (NeurIPS 2025 · similarity 0.90)
- From Flatland to Space: Teaching Vision-Language Models to Perceive and Reason in 3D (NeurIPS 2025 · similarity 0.89)
- Grounded 3D-Aware Spatial Vision-Language Modeling (CVPR 2026 · similarity 0.89)
- MMScan: A Multi-Modal 3D Scene Dataset with Hierarchical Grounded Language Annotations (NeurIPS 2024 · similarity 0.89)
- SpatialLLM: A Compound 3D-Informed Design towards Spatially-Intelligent Large Multimodal Models (CVPR 2025 · similarity 0.88)
- Can Transformers Capture Spatial Relations between Objects? (ICLR 2024 · similarity 0.88)
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.
