Research map / Detection and segmentation / Weakly supervised detection and segmentation
Splitting vs. Merging: Mining Object Regions with Discrepancy and Intersection Loss for Weakly Supervised Semantic Segmentation
Tianyi Zhang, Guosheng Lin, Weide Liu, Jianfei Cai, Alex Kot
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 Discrepancy loss with weight 20 (beta 0 -> 20) (ablation) | region mining, pseudo-masks · PASCAL VOC 2012 train | pseudo-mask mIoU: 59.65 -> 61.27 | Improved |
| raise Discrepancy loss weight from 20 to 50 (ablation) | region mining, pseudo-masks · PASCAL VOC 2012 train | pseudo-mask mIoU: 61.27 -> 57.22 | Hurt |
| increase Discrepancy loss weight from 0 to 50 (ablation) | S_A and S_B maps · map similarity | similarity score: 93.54 -> 64.31 | Improved |
| add Intersection loss at beta=50 (ablation) | union map U(SA,SB) · PASCAL VOC 2012 val | mSoftoverlap: 24.5 -> 27.35 | Improved |
| add Intersection loss at beta=50 (ablation) | union map U(SA,SB) · PASCAL VOC 2012 val | mSoftrec: 34.00 -> 49.15 | Improved |
| add Intersection loss at beta=20 (ablation) | union map U(SA,SB) · PASCAL VOC 2012 val | mSoftoverlap: 20.0 -> 24.17 | Improved |
| add Intersection loss at beta=10 (ablation) | union map U(SA,SB) · PASCAL VOC 2012 val | mSoftoverlap: 24.0 -> 23.7 | No clear effect |
| increase Discrepancy loss weight from 10 to 50 (with Inter loss) (ablation) | union map U(SA,SB) · PASCAL VOC 2012 val | mSoftpre: 50.03 -> 36.82 | Hurt |
| increase Discrepancy loss weight from 10 to 50 (with Inter loss) (ablation) | union map U(SA,SB) · PASCAL VOC 2012 val | mSoftrec: 30.77 -> 49.15 | Improved |
| add reference-stream map S_C to union of S_A and S_B (ablation) | union localization map, beta=50 · PASCAL VOC 2012 val | mSoftoverlap: 27.35 -> 28.57 | Improved |
| add reference-stream map S_C to union of S_A and S_B (ablation) | union localization map, beta=0 · PASCAL VOC 2012 val | mSoftoverlap: 21.22 -> 24.57 | Improved |
| use union of S_A and S_B instead of single map S_A (ablation) | localization map, beta=50 · PASCAL VOC 2012 val | mSoftoverlap: 24.47 -> 27.35 | 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 2 of 2 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
- Non-Salient Region Object Mining for Weakly Supervised Semantic Segmentation (CVPR 2021 · similarity 0.91)
- Weakly-Supervised Semantic Segmentation by Iteratively Mining Common Object Features (CVPR 2018 · similarity 0.91)
- Reliability Does Matter: An End-to-End Weakly Supervised Semantic Segmentation Approach (AAAI 2020 · similarity 0.89)
- Weakly-Supervised Semantic Segmentation via Sub-Category Exploration (CVPR 2020 · similarity 0.89)
- Mining Cross-Image Semantics for Weakly Supervised Semantic Segmentation (ECCV 2020 · similarity 0.89)
- W2P: Switching from Weak Supervision to Partial Supervision for Semantic Segmentation (AAAI 2024 · similarity 0.88)
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.
