Research map / Detection and segmentation / Weakly supervised detection and segmentation
Regional Semantic Contrast and Aggregation for Weakly Supervised Semantic Segmentation
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 |
|---|---|---|---|
| Add RCA to OAA++ | OAA++ with VGG16 classifier · VOC 2012 val | mIoU (%): 67.7 -> 70.6 | Helped |
| Add RCA to EPS | EPS with ResNet38 classifier · VOC 2012 val | mIoU (%): 70.9 -> 72.2 | Helped |
| Add RCA to EPS | EPS with VGG16 · COCO 2014 val | mIoU (%): 35.7 -> 36.8 | Helped |
| Add RSC to OAA++ without RSA (ablation) | OAA++ with VGG16 · VOC 2012 train pseudo labels | mIoU (%): 68.2 -> 69.5 | Helped |
| Add RSC to OAA++ without RSA (ablation) | OAA++ with VGG16 · VOC 2012 val segmentation | mIoU (%): 67.7 -> 69.3 | Helped |
| Add RSA without RSC (ablation) | OAA++ with VGG16 · VOC 2012 train pseudo labels | mIoU (%): 68.2 -> 68.5 | Helped |
| Add RSC and RSA together rather than RSC alone (ablation) | OAA++ with VGG16 · VOC 2012 train pseudo labels | mIoU (%): 69.5 -> 71.4 | Helped |
| Add region mixup to the contrastive objective (ablation) | RCA · VOC 2012 | mIoU (%): 70.6 -> 71.4 | Helped |
| Use memory momentum with gamma 0.99 instead of no momentum (ablation) | RCA · VOC 2012 pseudo labels | mIoU (%): 69.9 -> 71.4 | Helped |
| Increase gamma from 0.99 to 0.999 (ablation) | RCA · VOC 2012 pseudo labels | mIoU (%): 71.4 -> 70.9 | Hurt |
| Use one prototype per class instead of 10 (ablation) | RCA semantic aggregation · VOC 2012 | mIoU (%): 71.4 -> 70.4 | Hurt |
| Use all memory embeddings as prototypes instead of clustered prototypes (ablation) | RCA semantic aggregation · VOC 2012 | mIoU (%): 71.4 -> 70.0 | Hurt |
Closest papers by meaning
- Universal Weakly Supervised Segmentation by Pixel-to-Segment Contrastive Learning (ICLR 2021 · similarity 0.90)
- Weakly-Supervised Semantic Segmentation via Sub-Category Exploration (CVPR 2020 · similarity 0.90)
- Weakly-Supervised Semantic Segmentation by Iteratively Mining Common Object Features (CVPR 2018 · similarity 0.89)
- CIAN: Cross-Image Affinity Net for Weakly Supervised Semantic Segmentation (AAAI 2020 · similarity 0.89)
- Separate and Conquer: Decoupling Co-occurrence via Decomposition and Representation for Weakly Supervised Semantic Segmentation (CVPR 2024 · similarity 0.89)
- Boosting Weakly Supervised Referring Image Segmentation via Progressive Comprehension (NeurIPS 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.