Research map / Detection and segmentation / Weakly supervised detection and segmentation
Adversarial Erasing Framework via Triplet with Gated Pyramid Pooling Layer for Weakly Supervised Semantic Segmentation
Sung-Hoon Yoon, Hyeokjun Kweon, Jegyeong Cho, Shinjeong Kim, Kuk-Jin Yoon
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 |
|---|---|---|---|
| replace GAP with Gated Pyramid Pooling (coarse-to-fine) (ablation) | classifier · VOC 2012 train (CAM pseudo-labels) | mIoU: 48.4 -> 54.2 | Improved |
| use larger pooling bin size in GPP (ablation) | GPP layer · VOC 2012 train (CAM pseudo-labels) | mIoU: 49.9 -> 53.1 | Improved |
| aggregate pyramid with coarse-to-fine gated conv instead of averaging (ablation) | GPP layer · VOC 2012 train (CAM pseudo-labels) | mIoU: 53.3 -> 54.2 | Improved |
| aggregate with fine-to-coarse gating instead of coarse-to-fine (ablation) | GPP layer · VOC 2012 train (CAM pseudo-labels) | mIoU: 54.2 -> 51.3 | Hurt |
| add triplet adversarial erasing (soft attraction) to GPP (ablation) | AEFT · VOC 2012 train (CAM pseudo-labels) | mIoU: 54.2 -> 55.0 | Improved |
| add repelling loss to attraction loss (ablation) | AEFT · VOC 2012 train (CAM pseudo-labels) | mIoU: 55.0 -> 56.0 | Improved |
| use rigid classification labels for the erased image instead of triplet (ablation) | AEFT · VOC 2012 train (CAM pseudo-labels) | mIoU: 54.2 -> 53.4 | Hurt |
| use GPP feature as embedding instead of CAMs (ablation) | AEFT · VOC 2012 train (CAM pseudo-labels) | mIoU: 54.6 -> 56.0 | Improved |
| maximize distance between eA and eN instead of eAL and eN (ablation) | AEFT · VOC 2012 train (CAM pseudo-labels) | mIoU: 56.0 -> 54.1 | Hurt |
| apply CRF to AEFT CAMs | AEFT · VOC 2012 train (CAM pseudo-labels) | mIoU: 56.0 -> 63.5 | Improved |
| apply random walk after CRF | AEFT · VOC 2012 train (CAM pseudo-labels) | mIoU: 63.5 -> 71.0 | Improved |
Closest papers by meaning
- Unlocking the Potential of Ordinary Classifier: Class-Specific Adversarial Erasing Framework for Weakly Supervised Semantic Segmentation (ICCV 2021 · similarity 0.94)
- Object Region Mining With Adversarial Erasing: A Simple Classification to Semantic Segmentation Approach (CVPR 2017 · similarity 0.90)
- Knowledge Transfer with Simulated Inter-Image Erasing for Weakly Supervised Semantic Segmentation (ECCV 2024 · similarity 0.90)
- Weakly Supervised Semantic Segmentation via Adversarial Learning of Classifier and Reconstructor (CVPR 2023 · similarity 0.89)
- ECS-Net: Improving Weakly Supervised Semantic Segmentation by Using Connections Between Class Activation Maps (ICCV 2021 · similarity 0.89)
- Erasing Integrated Learning: A Simple Yet Effective Approach for Weakly Supervised Object Localization (CVPR 2020 · similarity 0.89)
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.
