Research map / Detection and segmentation / Open-set and few-shot detection
Self-Supervised Interactive Object Segmentation through a Singulation-and-Grasping Approach
Houjian Yu, Changhyun Choi
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 optical flow filtering classifier (ablation) | SaG · Toy blocks sim, default | AP50:95: 81.5 -> 87.8 | Improved |
| add optical flow filtering classifier (ablation) | SaG · Toy blocks sim, cluttered | AP50:95: 72.1 -> 73.0 | Improved |
| add optical flow filtering classifier (ablation) | SaG · YCB objects sim, cluttered | AP50:95: 58.7 -> 61.7 | Improved |
| add optical flow filtering classifier (ablation) | SaG · Real robot novel objects, default | AP50:95: 61.1 -> 69.3 | Improved |
| add optical flow filtering classifier (ablation) | SaG · Real robot novel objects, cluttered | AP50:95: 62.5 -> 63.2 | No clear effect |
| add most cluttered mask input m_t (ablation) | SaG · Toy blocks sim, cluttered | AP50:95: 69.1 -> 73.0 | Improved |
| add mask projection map h_t (ablation) | SaG · Toy blocks sim, cluttered | AP50:95: 65.4 -> 73.0 | Improved |
| add grasping interactions to pushing-only data collection (ablation) | SaG · Toy blocks sim, default | AP50:95: 80.1 -> 87.8 | Improved |
| add pushing (singulation) to grasping-only data collection (ablation) | SaG · Toy blocks sim, default | AP50:95: 83.4 -> 87.8 | Improved |
| use m_t as input instead of filtering Q maps by mask (ablation) | SaG · Toy blocks sim, cluttered | AP50:95: 70.5 -> 73.0 | Improved |
| SaG singulation policy versus VPG push-only for data collection | SaG · Toy blocks sim, default | AP50:95: 69.4 -> 87.8 | Improved |
| use SaG fine-tuned segmentation as input to VPG grasping | VPG · Top-down grasping sim, 6 toy blocks | grasp success rate (%): 70.7 -> 85.5 | Improved |
Closest papers by meaning
- Self-Supervised Learning of Object Parts for Semantic Segmentation (CVPR 2022 · similarity 0.87)
- Large-Scale Interactive Object Segmentation With Human Annotators (CVPR 2019 · similarity 0.86)
- Learning to Segment via Cut-and-Paste (ECCV 2018 · similarity 0.86)
- GrabS: Generative Embodied Agent for 3D Object Segmentation without Scene Supervision (ICLR 2025 · similarity 0.86)
- Learning with Free Object Segments for Long-Tailed Instance Segmentation (ECCV 2022 · similarity 0.86)
- AGILE3D: Attention Guided Interactive Multi-object 3D Segmentation (ICLR 2024 · similarity 0.85)
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.
