Research map / Detection and segmentation / Weakly supervised detection and segmentation
Weakly Supervised Object Boundaries
Check your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 |
|---|---|---|---|
| Train object-boundary models using bounding-box-matched weak annotations rather than object boundary labels | SE and HED · VOC12 validation | AP: HED(cons. all methods ∩ BBs): 50; HED(VOC): 61 | Helped |
| Use consensus rather than union of MCG proposal boundaries (ablation) | HED · VOC12 validation | AP: 44 -> 50 | Helped |
| Add object detection at test time (ablation) | HED trained on VOC ground truth · VOC12 validation | AP: 61 -> 58 | Hurt |
| Add object detection at test time (ablation) | HED trained on weak labels · VOC12 validation | AP: 46 -> 47; 50 -> 50 | Mixed |
| Use SeSe rather than F&H segmentation for weak training boundaries (ablation) | SE · VOC12 validation | AP: 29 -> 35 | Helped |
| Add VOC+ images to SeSe-based weak training (ablation) | SE · VOC12 validation | AP: 35 -> 36 | Helped |
| Add VOC+ images to MCG-based weak training (ablation) | SE · VOC12 validation | AP: 34 -> 35 | Helped |
| Use GrabCut or DenseCut/CNN+GraphCut instead of one another for weak labels (ablation) | Weak boundary annotation generation · VOC12 validation | AP | No clear effect |
| Use the union of multiple matched proposals rather than only the highest-overlapping proposal (ablation) | Weak boundary annotation generation · VOC12 validation | AP | Helped |
| Train on noisy F&H boundaries rather than Canny boundaries (ablation) | SE · BSDS500 test | AP: 64 -> 76 | Helped |
| Train HED using SE-generated weak labels rather than F&H labels directly (ablation) | HED · BSDS500 test | AP: 73 -> 75 | Helped |
| Add bounding-box detection at test time (ablation) | SE weak-boundary models · VOC12 validation | AP: 35 -> 39; 35 -> 40 | Helped |
Closest papers by meaning
- Track and Transfer: Watching Videos to Simulate Strong Human Supervision for Weakly-Supervised Object Detection (CVPR 2016 · similarity 0.89)
- Training Object Class Detectors With Click Supervision (CVPR 2017 · similarity 0.88)
- We Don't Need No Bounding-Boxes: Training Object Class Detectors Using Only Human Verification (CVPR 2016 · similarity 0.88)
- Partial Weakly-Supervised Oriented Object Detection (CVPR 2026 · similarity 0.88)
- WeakPCSOD: Overcoming the Bias of Box Annotations for Weakly Supervised Point Cloud Salient Object Detection (AAAI 2024 · similarity 0.86)
- The Devil is in the Object Boundary: Towards Annotation-free Instance Segmentation using Foundation Models (ICLR 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.