Research map / Detection and segmentation / Weakly supervised detection and segmentation
Employing Multi-Estimations for Weakly-Supervised Semantic Segmentation
Junsong Fan, Zhaoxiang Zhang, Tieniu Tan
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 |
|---|---|---|---|
| train on a second noisy label set instead of one (noise ratio 0.3) (ablation) | VGG16-LargeFov · VOC 2012 val (hand-crafted noise) | mIoU (%): 64.8 -> 66.2 | Improved |
| add CRF seeds to threshold seeds (multi-type) (ablation) | VGG16 CAM seeds · VOC 2012 val | mIoU (%): 61.4 -> 63.8 | Improved |
| use threshold plus CRF seeds from RES38 CAM (ablation) | RES38 CAM seeds · VOC 2012 val | mIoU (%): 62.3 -> 63.5 | Improved |
| use three CAM scales as separate seeds (ablation) | threshold seeds · VOC 2012 val | mIoU (%): 61.4 -> 62.5 | Improved |
| use three CAM scales as separate seeds (ablation) | both seed types · VOC 2012 val | mIoU (%): 63.8 -> 64.0 | Improved |
| merge multi-scale CAMs into one seed set by max fusion instead of separate seeds (ablation) | threshold seeds · VOC 2012 val | mIoU (%): 62.5 -> 60.2 | Hurt |
| use separate multi-scale seeds instead of average-fused seeds (ablation) | both seed types · VOC 2012 val | mIoU (%): 63.9 -> 64.0 | Improved |
| add RES38 seeds to VGG16 seeds (ablation) | both seed types · VOC 2012 val | mIoU (%): 63.8 -> 64.2 | Improved |
| add RES38 seeds to VGG16 seeds (ablation) | CRF seeds · VOC 2012 val | mIoU (%): 62.9 -> 63.1 | Improved |
| add weighted selective training with all three seed kinds (ablation) | multi-type, multi-scale, multi-arch seeds · VOC 2012 val | mIoU (%): 63.9 -> 64.6 | Improved |
| add weighted selective training with multi-type seeds only (ablation) | multi-type seeds · VOC 2012 val | mIoU (%): 63.9 -> 63.8 | No clear effect |
| add weighted selective training to multi-type and multi-scale seeds (ablation) | multi-type, multi-scale seeds · VOC 2012 val | mIoU (%): 63.8 -> 64.0 | Improved |
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.
- FickleNet — FickleNet: Weakly and Semi-Supervised Semantic Image Segmentation Using Stochastic Inference (CVPR 2019) · better on 4 of 4 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
- Reliability Does Matter: An End-to-End Weakly Supervised Semantic Segmentation Approach (AAAI 2020 · similarity 0.91)
- Modeling the Label Distributions for Weakly-Supervised Semantic Segmentation (TPAMI 2025 · similarity 0.91)
- Threshold Matters in WSSS: Manipulating the Activation for the Robust and Accurate Segmentation Model Against Thresholds (CVPR 2022 · similarity 0.91)
- Progressive Feature Self-Reinforcement for Weakly Supervised Semantic Segmentation (AAAI 2024 · similarity 0.91)
- W2P: Switching from Weak Supervision to Partial Supervision for Semantic Segmentation (AAAI 2024 · similarity 0.90)
- From SAM to CAMs: Exploring Segment Anything Model for Weakly Supervised Semantic Segmentation (CVPR 2024 · similarity 0.90)
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.
