Research map / Detection and segmentation / Weakly supervised detection and segmentation
Bootstrapping Semantic Segmentation with Regional Contrast
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 |
|---|---|---|---|
| Add regional contrast with adaptive hard-negative keys and low-confidence queries | ClassMix · Pascal VOC, CityScapes, and SUN RGB-D, few-label full-label setting | mean IoU: 49.06 -> 53.31; 45.61 -> 49.86; 28.42 -> 29.65 | Helped |
| Add regional contrast | supervised segmentation · Pascal VOC, CityScapes, and SUN RGB-D, all labels | mean IoU: 77.79 -> 78.39; 70.48 -> 71.45; 51.06 -> 52.01 | Helped |
| Add regional contrast | ClassMix · Pascal VOC, PseudoSeg benchmark | mean IoU: 57.60 -> 64.78 | Helped |
| Add regional contrast | ClassMix · Pascal VOC and CityScapes, partial-label settings | mean IoU: 63.69 -> 66.11; 71.04 -> 72.67; 72.90 -> 74.09; 75.79 -> 75.96; 47.42 -> 49.66; 56.68 -> 58.97; 60.96 -> 62.32; 66.46 -> | Helped |
| Replace active query and key selection with random queries and keys (ablation) | ReCo + ClassMix · CityScapes, 20 labels | mean IoU: 49.86 -> 46.56 | Hurt |
| Replace active key selection with random key selection (ablation) | ReCo + ClassMix · CityScapes, 20 labels | mean IoU: 49.86 -> 46.38 | Hurt |
| Sample easy rather than hard queries (ablation) | ReCo + ClassMix · CityScapes, 20 labels | mean IoU: 49.86 -> 45.81 | Hurt |
| Increase the number of queries and keys (ablation) | ReCo + ClassMix · CityScapes, 20 labels | mean IoU | Helped |
| Train with only 10% of the unlabelled data (ablation) | ReCo + ClassMix · CityScapes, 20 labels | mean IoU | Helped |
| Replace batch-wise sampling with feature-bank sampling (ablation) | ReCo + ClassMix · CityScapes, 20 labels | mean IoU: 49.86 -> 49.34 | Hurt |
| Use ReCo with different semi-supervised methods instead of only ClassMix (ablation) | semi-supervised segmentation · CityScapes, 20 labels | mean IoU | Helped |
Closest papers by meaning
- Looking Beyond Single Images for Contrastive Semantic Segmentation Learning (NeurIPS 2021 · similarity 0.89)
- Learn to Rectify the Bias of CLIP for Unsupervised Semantic Segmentation (CVPR 2024 · similarity 0.87)
- Diffusion-Driven Two-Stage Active Learning for Low-Budget Semantic Segmentation (NeurIPS 2025 · similarity 0.87)
- Learning Hierarchical Image Segmentation For Recognition and By Recognition (ICLR 2024 · similarity 0.87)
- Contextrast: Contextual Contrastive Learning for Semantic Segmentation (CVPR 2024 · similarity 0.86)
- On the Relaxation of Conditional Independence Assumption for Image Segmentation (NeurIPS 2026 · similarity 0.86)
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.