Research map / Detection and segmentation / Weakly supervised detection and segmentation
Seminar Learning for Click-Level Weakly Supervised Semantic Segmentation
Hongjun Chen, Jinbao Wang, Hong Cai Chen, Xiantong Zhen, Feng Zheng, Rongrong Ji, Ling Shao
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 L_CRF to L_pCE (ablation) | click-supervised DeepLab-v3+ ResNet101 · Pascal VOC 2012 val | mIOU (%): 54.70 -> 55.63 | No clear effect |
| add L_CRF to L_pCE (ablation) | click-supervised DeepLab-v3+ MobileNet · Pascal VOC 2012 val | mIOU (%): 50.85 -> 48.75 | Hurt |
| add pixel consistency loss L_pCons (EMA teacher) to L_pCE (ablation) | click-supervised DeepLab-v3+ ResNet101 · Pascal VOC 2012 val | mIOU (%): 54.70 -> 61.47 | Improved |
| add L_pCons to L_pCE + L_CRF (ablation) | click-supervised DeepLab-v3+ ResNet101 · Pascal VOC 2012 val | mIOU (%): 55.63 -> 70.29, increasing by 14.66% | Improved |
| add L_CRF to L_pCE + L_pCons (ablation) | click-supervised DeepLab-v3+ ResNet101 · Pascal VOC 2012 val | mIOU (%): 61.47 -> 70.29, exceeds by 8.82% | Improved |
| add L_pCons to L_pCE + L_CRF (ablation) | click-supervised DeepLab-v3+ MobileNet · Pascal VOC 2012 val | mIOU (%): 48.75 -> 62.03 | Improved |
| add heterogeneous pseudo-label loss L_pseudo (student-student module) (ablation) | seminar learning ResNet101 · Pascal VOC 2012 val | mIOU (%): 70.29 -> 72.51, extra 2.22% | Improved |
| add heterogeneous pseudo-label loss L_pseudo (student-student module) (ablation) | seminar learning MobileNet · Pascal VOC 2012 val | mIOU (%): 62.03 -> 64.44 | Improved |
| generate pseudo-labels from the same network (self-generation) instead of a separate ancillary network (ablation) | seminar learning · Pascal VOC 2012 val | mIOU (%) | Hurt |
| add a first student-student module (ablation) | seminar learning ResNet101 · Pascal VOC 2012 val | mIOU (%): 70.29 -> 72.51 | Improved |
| add a first student-student module (ablation) | seminar learning MobileNet · Pascal VOC 2012 val | mIOU (%): 60.59 -> 62.59 | Improved |
| chain more than one student-student module (ablation) | seminar learning · Pascal VOC 2012 val | mIOU (%) | No clear effect |
Closest papers by meaning
- Learning from Future: A Novel Self-Training Framework for Semantic Segmentation (NeurIPS 2022 · similarity 0.89)
- Single-Stage Semantic Segmentation From Image Labels (CVPR 2020 · similarity 0.88)
- Reliability Does Matter: An End-to-End Weakly Supervised Semantic Segmentation Approach (AAAI 2020 · similarity 0.88)
- Weakly Supervised Semantic Segmentation Using Web-Crawled Videos (CVPR 2017 · similarity 0.88)
- Semantic-aware SAM for Point-Prompted Instance Segmentation (CVPR 2024 · similarity 0.87)
- Separate and Conquer: Decoupling Co-occurrence via Decomposition and Representation for Weakly Supervised Semantic Segmentation (CVPR 2024 · similarity 0.87)
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.
