Research map / 3D vision / 3D generation
Diffusion Models are Open-World Affordance Learners: Leveraging Generative Priors for 3D Affordance Learning
Hanqing Wang, Zhenhao Zhang, Kaiyang Ji, Mingyu Liu, Wenti Yin, yuchao chen, Zhirui Liu, Xiangyu Zeng, Tianxiang Gui, Hangxing Zhang, Jiahao Yuan, Zhiqing Cui and 3 more
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 |
|---|---|---|---|
| use diffusion U-Net features instead of ResNet/CLIP/DINOv2 as affordance extractor (ablation) | DAG · PIAD1 Seen | mIoU: 22.3 (DINOv2) -> 24.84 | Improved |
| add Affordance Block (ablation) | DAG · PIAD1 Seen | mIoU: 10.02 -> 18.56 | Improved |
| add [CLS] token (ablation) | DAG · PIAD1 Seen | mIoU: 18.56 -> 24.84 | Improved |
| add Affordance Block (ablation) | DAG · PIAD1 Unseen | mIoU: 8.45 -> 9.02 | Improved |
| use implicit ASE caption instead of empty, BLIP or verb text (ablation) | DAG · PIAD1 Seen | mIoU: 20.2 (Verb) -> 24.84 | Improved |
| use diffusion timestep 0 instead of 50 or 100 (ablation) | DAG · PIAD1 Seen | mIoU: 21.93 (t=100) -> 24.85 (t=0) | Improved |
| use DAG instead of GREAT on unseen objects | DAG · PIAD2 Unseen Obj | mIoU: 20.16 -> 28.93 | Improved |
| train DAG with 5 shots (2% data) vs full-data PFU baseline | DAG · PIAD1 Seen | mIoU: 12.31 (PFU all data) -> 15.74 | Improved |
| reduce training data from all to 1 shot | DAG · PIAD1 Seen | mIoU: 24.84 -> 12.98 | Hurt |
Closest papers by meaning
- GEAL: Generalizable 3D Affordance Learning with Cross-Modal Consistency (CVPR 2025 · similarity 0.88)
- Move as You Say Interact as You Can: Language-guided Human Motion Generation with Scene Affordance (CVPR 2024 · similarity 0.88)
- Learning 2D Invariant Affordance Knowledge for 3D Affordance Grounding (AAAI 2025 · similarity 0.88)
- 3D-AffordanceLLM: Harnessing Large Language Models for Open-Vocabulary Affordance Detection in 3D Worlds (ICLR 2025 · similarity 0.88)
- SeqAfford: Sequential 3D Affordance Reasoning via Multimodal Large Language Model (CVPR 2025 · similarity 0.87)
- Grounding 3D Object Affordance with Language Instructions, Visual Observations and Interactions (CVPR 2025 · similarity 0.87)
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.
