Research map / Detection and segmentation / Open-set and few-shot detection
Act the Part: Learning Interaction Strategies for Articulated Object Part Discovery
Samir Yitzhak Gadre, Kiana Ehsani, Shuran Song
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 hold action (bimanual hold+push instead of push only) (ablation) | Act the Part · PartNet-Mobility Multilink | mIoU (%): 19.9 -> 54.7 | Improved |
| add hold action (bimanual hold+push instead of push only) (ablation) | Act the Part · PartNet-Mobility Scissors | mIoU (%): 28.5 -> 65.6 | Improved |
| add part-aware interaction network (part memory input) vs Act-NoPart (ablation) | Act the Part · PartNet-Mobility USB | mIoU (%): 47.6 -> 75.3 | Improved |
| add part-aware interaction network (part memory input) vs Act-NoPart (ablation) | Act the Part · PartNet-Mobility Eyeglasses | mIoU (%): 34.4 -> 43.0 | Improved |
| add touch feedback reward (ablation) | Act the Part · PartNet-Mobility Scissors | mIoU (%): 58.0 -> 65.6 | Improved |
| add touch feedback reward (ablation) | Act the Part · PartNet-Mobility Multilink | mIoU (%): 46.6 -> 54.7 | Improved |
| add touch feedback reward (ablation) | Act the Part · PartNet-Mobility Lighter | mIoU (%): 64.4 -> 62.8 | Hurt |
| add touch feedback reward (ablation) | Act the Part · PartNet-Mobility Knife | mIoU (%): 66.2 -> 65.9 | No clear effect |
| learned interaction policy instead of random actions | Act the Part · PartNet-Mobility Multilink | mIoU (%): 10.7 -> 54.7 | Improved |
| train on unseen categories: test on categories not seen in training | Act the Part (Ours-Touch) · PartNet-Mobility Pliers vs Scissors | mIoU (%): 65.6 -> 64.9 | No clear effect |
| evaluate objects with three links when trained only on two-link objects | Act the Part (Ours-Touch) · PartNet-Mobility Multilink | MAPE: 0.22 | Mixed |
| use ground-truth optical-flow motion masks and optimal actions (oracle) | Act the Part · PartNet-Mobility Multilink | mIoU (%): 54.7 -> 86.9 | Improved |
Closest papers by meaning
- Where2Act: From Pixels to Actions for Articulated 3D Objects (ICCV 2021 · similarity 0.89)
- Command-Driven Articulated Object Understanding and Manipulation (CVPR 2023 · similarity 0.88)
- GAPartNet: Cross-Category Domain-Generalizable Object Perception and Manipulation via Generalizable and Actionable Parts (CVPR 2023 · similarity 0.88)
- IAAO: Interactive Affordance Learning for Articulated Objects in 3D Environments (CVPR 2025 · similarity 0.88)
- BiAssemble: Learning Collaborative Affordance for Bimanual Geometric Assembly (ICML 2025 · similarity 0.87)
- Articulate your NeRF: Unsupervised articulated object modeling via conditional view synthesis (NeurIPS 2024 · 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.
