Research map / 3D vision / 3D generation
AffIn-Space: Learning Affine-Invariant Representations for 3D Spatial Understanding with MLLMs
Zhenyu Lu, Liupeng Li, Jinpeng Wang, Haoqian Kang, Manyuan Zhang, Yan Feng, Ke Chen, Yaowei Wang
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 |
|---|---|---|---|
| add explicit geometric affine resampling (ablation) | AffIn-Space-8B · ScanQA | CIDEr: 90.7 -> 99.4 | Improved |
| add explicit geometric affine resampling (ablation) | AffIn-Space-4B · ScanQA | CIDEr: 86.6 -> 93.8 | Improved |
| add explicit geometric affine resampling (ablation) | AffIn-Space-8B · Scan2Cap | CIDEr: 76.4 -> 87.4 | Improved |
| add affine-invariant orthogonal projection constraints (ablation) | AffIn-Space-8B · ScanQA | CIDEr: 94.1 -> 99.4 | Improved |
| add affine-invariant orthogonal projection constraints (ablation) | AffIn-Space-4B · ScanQA | CIDEr: 90.7 -> 93.8 | Improved |
| add reconstruction loss (coef 1.0 vs 0.0) (ablation) | AffIn-Space-8B · ScanQA | CIDEr: 92.8 -> 99.4 | Improved |
| add reconstruction loss (coef 1.0 vs 0.0) (ablation) | AffIn-Space-8B · SQA3D | EM: 55.2 -> 59.2 | Improved |
| raise reconstruction coefficient above 1.0 (2.0) (ablation) | AffIn-Space-8B · ScanQA | CIDEr: 99.4 -> 89.9 | Hurt |
| raise reconstruction coefficient to 5.0 (ablation) | AffIn-Space-8B · ScanQA | CIDEr: 99.4 -> 97.8 | Hurt |
| use AffIn-Space instead of VG-LLM-8B | AffIn-Space-8B · EmbodiedScan 6-frame | F1: 40.8 -> 44.5 | Improved |
| use AffIn-Space-8B vs Video-3D LLM (uses 3D inputs) | AffIn-Space-8B · ScanQA | CIDEr: 102.1 -> 99.4 | Hurt |
Closest papers by meaning
- ImageNet3D: Towards General-Purpose Object-Level 3D Understanding (NeurIPS 2024 · similarity 0.87)
- Multi-view Relational Distillation for Spatial Reasoning with Vision-Language Models (NeurIPS 2026 · similarity 0.87)
- Imagine3D-LLM: Teaching MLLMs to Imagine 3D Scenes Before Answering (NeurIPS 2026 · similarity 0.87)
- Spatial-MLLM: Boosting MLLM Capabilities in Visual-based Spatial Intelligence (NeurIPS 2025 · similarity 0.86)
- Less Decoder is More Encoder: Geometric Representation Learning from Novel View Synthesis (NeurIPS 2026 · similarity 0.86)
- InfiniBench: Infinite Benchmarking for Visual Spatial Reasoning with Customizable Scene Complexity (CVPR 2026 · similarity 0.86)
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.
