Research map / Vision-language and multimodal / Robot action and navigation
RAGNet: Large-scale Reasoning-based Affordance Segmentation Benchmark towards General Grasping
Dongming Wu, Yanping Fu, Saike Huang, Yingfei Liu, Fan Jia, Nian Liu, Feng Dai, Tiancai Wang, Rao Muhammad Anwer, Fahad Shahbaz Khan, Jianbing Shen
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 HANDAL data to LISA training (ablation) | LISA · HANDAL subset | gIoU: 16.2 -> 56.3 | Improved |
| add GraspNet data to training (ablation) | LISA · GraspNet novel (zero-shot) | gIoU: 8.0 -> 51.5 | Improved |
| add EgoObjects data to training (ablation) | LISA · GraspNet novel (zero-shot) | gIoU: 19.2 -> 8.0 | Hurt |
| add EgoObjects data to training (ablation) | LISA · 3DOI (zero-shot) | gIoU: 24.5 -> 35.5 | Improved |
| add reasoning instruction data (ablation) | LISA · GraspNet novel (zero-shot) | gIoU: 51.5 -> 43.0 | Hurt |
| add reasoning instruction data (ablation) | LISA · HANDAL subset | gIoU: 61.7 -> 56.5 | Hurt |
| add RLBench simulation data (ablation) | LISA · GraspNet novel (zero-shot) | gIoU: 43.0 -> 42.8 | No clear effect |
| add specialized system prompt and <AFF> token (ablation) | AffordanceVLM · GraspNet novel (zero-shot) | gIoU: 42.8 -> 45.6 | Improved |
| add specialized system prompt and <AFF> token (ablation) | AffordanceVLM · HANDAL subset | gIoU: 56.7 -> 60.5 | Improved |
| use AffordanceNet instead of VLPart for affordance prediction (ablation) | Real-robot grasping · 5 real-robot tasks | success rate: 34% -> 70% | Improved |
| use AffordanceNet instead of LISA for affordance prediction (ablation) | Real-robot grasping · 5 real-robot tasks | success rate: 26% -> 70% | Improved |
| use hard reasoning instructions instead of template instructions (ablation) | AffordanceNet · 5 real-robot tasks | success rate: 70% -> 48% | Hurt |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- GLaMM — GLaMM: Pixel Grounding Large Multimodal Model (CVPR 2024) · better on 3 of 3 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
- SeqAfford: Sequential 3D Affordance Reasoning via Multimodal Large Language Model (CVPR 2025 · similarity 0.90)
- BiAssemble: Learning Collaborative Affordance for Bimanual Geometric Assembly (ICML 2025 · similarity 0.89)
- 3D-AffordanceLLM: Harnessing Large Language Models for Open-Vocabulary Affordance Detection in 3D Worlds (ICLR 2025 · similarity 0.89)
- AffordDP: Generalizable Diffusion Policy with Transferable Affordance (CVPR 2025 · similarity 0.89)
- Affordance-R1: Reinforcement Learning for Generalizable Affordance Reasoning in Multimodal Large Language Models (AAAI 2026 · similarity 0.88)
- RealVLG-R1: A Large-Scale Real-World Visual-Language Grounding Benchmark for Robotic Perception and Manipulation (CVPR 2026 · 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.
