Research map / Detection and segmentation / Weakly supervised detection and segmentation
Universal Weakly Supervised Segmentation by Pixel-to-Segment Contrastive Learning
Read the paperCheck 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 |
|---|---|---|---|
| Use pixel-to-segment contrastive learning with the SPML framework | Weakly supervised segmentation on Pascal VOC 2012 image tags · Pascal VOC 2012 val, image tags | mIoU: 67.1 -> 69.5 | Helped |
| Use pixel-to-segment contrastive learning with the SPML framework | Weakly supervised segmentation on Pascal VOC 2012 bounding boxes · Pascal VOC 2012 val, bounding boxes | mIoU: 70.2 -> 73.5 | Helped |
| Use SPML instead of the strongest listed prior method | Pascal VOC 2012 scribble supervision, without CRF · Pascal VOC 2012 val, scribbles | mIoU: 75.6 -> 74.2 | Hurt |
| Apply CRF post-processing | SPML on Pascal VOC 2012 scribbles · Pascal VOC 2012 val, scribbles | mIoU: 74.2 -> 76.1 | Helped |
| Use SPML instead of the prior baseline | DensePose point-supervised segmentation · DensePose minival 2014 | mIoU: 31.3 -> 44.2 | Helped |
| Add low-level image similarity regularization (ablation) | SPML on Pascal VOC 2012 scribbles · Pascal VOC 2012 val | mIoU: 71.2 -> 72.8 | Helped |
| Add semantic co-occurrence regularization (ablation) | SPML on Pascal VOC 2012 scribbles · Pascal VOC 2012 val | mIoU: 72.8 -> 74.2 | Helped |
| Add feature affinity regularization (ablation) | SPML on Pascal VOC 2012 scribbles · Pascal VOC 2012 val | mIoU: 74.2 -> 73.8 | Hurt |
| Increase low-level similarity weight from λI=0.1 to λI=0.3 (ablation) | SPML on Pascal VOC 2012 scribbles with λO=0.5 · Pascal VOC 2012 val | mIoU: 74.2 -> 73.7 | Hurt |
| Reduce low-level similarity weight from λI=0.1 to λI=0.05 (ablation) | SPML on Pascal VOC 2012 scribbles with λO=0.5 · Pascal VOC 2012 val | mIoU: 74.2 -> 73.5 | Hurt |
| Remove low-level image similarity regularization (ablation) | SPML on Pascal VOC 2012 scribbles with λO=0.5 · Pascal VOC 2012 val | mIoU: 74.2 -> 71.7 | Hurt |
| Remove semantic co-occurrence regularization (ablation) | SPML on Pascal VOC 2012 scribbles with λI=0.1 · Pascal VOC 2012 val | mIoU: 74.2 -> 72.8 | Hurt |
Closest papers by meaning
- Combining Bottom-Up, Top-Down, and Smoothness Cues for Weakly Supervised Image Segmentation (CVPR 2017 · similarity 0.92)
- Reliability Does Matter: An End-to-End Weakly Supervised Semantic Segmentation Approach (AAAI 2020 · similarity 0.91)
- Weakly-Supervised Semantic Segmentation via Sub-Category Exploration (CVPR 2020 · similarity 0.90)
- Separate and Conquer: Decoupling Co-occurrence via Decomposition and Representation for Weakly Supervised Semantic Segmentation (CVPR 2024 · similarity 0.90)
- WISH: Weakly Supervised Instance Segmentation using Heterogeneous Labels (CVPR 2025 · similarity 0.89)
- Scribble Hides Class: Promoting Scribble-Based Weakly-Supervised Semantic Segmentation with Its Class Label (AAAI 2024 · 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.