Research map / Point clouds / Segmentation and detection
PourIt!: Weakly-Supervised Liquid Perception from a Single Image for Visual Closed-Loop Robotic Pouring
Haitao Lin, Yanwei Fu, Xiangyang Xue
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 feature contrast losses (L_pos + L_neg) to the CAM classifier (ablation) | PourIt! · PourIt! dataset (seen) | mIoU (%): 61.6 -> 65.9 | Improved |
| add feature contrast losses (L_pos + L_neg) to the CAM classifier (ablation) | PourIt! · PourIt! dataset (unseen) | mIoU (%): 59.2 -> 63.6 | Improved |
| add L_neg to L_pos (versus L_pos alone) (ablation) | PourIt! · PourIt! dataset (unseen) | mIoU (%): 36.7 -> 63.6 | Improved |
| use L_pos alone without L_neg (ablation) | PourIt! · PourIt! dataset (seen) | mIoU (%): 61.6 -> 46.7 | Hurt |
| add L_pos on top of L_neg (ablation) | PourIt! · PourIt! dataset (unseen) | mIoU (%): 63.2 -> 63.6 | No clear effect |
| weakly supervised CAM with feature contrast versus AFA | PourIt! · PourIt! dataset (unseen) | mIoU (%): 52.8 -> 63.6 | Improved |
| use PourIt! liquid feedback instead of open-loop pouring point from Lin et al. | Franka robot pouring · Robotic pouring, static target | success rate (%): 66.7 -> 93.3 | Improved |
| use AFA mask as visual feedback instead of Lin et al. | Franka robot pouring · Robotic pouring, static target | success rate (%): 66.7 -> 60.0 | Hurt |
| weak image-level supervision versus full supervision | PourIt! · Liquid dataset (seen) | mIoU (%): 69.0 -> 54.0 | Hurt |
Closest papers by meaning
- ZeroFlow: Scalable Scene Flow via Distillation (ICLR 2024 · similarity 0.81)
- DynaFLIP: Rethinking Robotics Perception via Tri-Modal-Dynamics Guided Representation (NeurIPS 2026 · similarity 0.81)
- FlowRAM: Grounding Flow Matching Policy with Region-Aware Mamba Framework for Robotic Manipulation (CVPR 2025 · similarity 0.80)
- Dataset Distillation for Pre-Trained Self-Supervised Vision Models (NeurIPS 2025 · similarity 0.80)
- Interactive 4D Volumetric Liquid Forecasting under Moving-Solid Interaction (NeurIPS 2026 · similarity 0.80)
- Floxels: Fast Unsupervised Voxel Based Scene Flow Estimation (CVPR 2025 · similarity 0.80)
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.
