Research map / Detection and segmentation / Weakly supervised detection and segmentation
Mix-and-Match Tuning for Self-Supervised Semantic Segmentation
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 M&M class-wise graph tuning after colorization pre-training | VGG-16 · PASCAL VOC2012 val | mIoU: 60.2 -> 64.5 | Helped |
| Add M&M class-wise graph tuning after colorization pre-training | AlexNet · PASCAL VOC2012 val | mIoU: 38.4 -> 42.8 | Helped |
| Add M&M tuning after jigsaw-puzzles pre-training | AlexNet · PASCAL VOC2012 val | mIoU: 36.5 -> 44.5 | Helped |
| Add M&M tuning with random initialization | AlexNet · PASCAL VOC2012 val | mIoU: 19.8 -> 43.6 | Helped |
| Add M&M tuning with random initialization | VGG-16 · PASCAL VOC2012 val | mIoU: 35.0 -> 56.7 | Helped |
| Add M&M tuning after colorization pre-training | VGG-16 · CityScapes val | mIoU: 57.5 -> 66.4 | Helped |
| Add M&M tuning with random initialization | VGG-16 · CityScapes val | mIoU: 42.5 -> 49.1 | Helped |
| Add a class-wise connected graph instead of random triplet selection (ablation) | M&M with colorization pre-training and AlexNet · PASCAL VOC2012 val | mIoU: 40.9 -> 42.8 | Helped |
| Replace colorization pre-training with jigsaw-puzzles pre-training (ablation) | M&M with AlexNet · PASCAL VOC2012 val | mIoU: 42.8 -> 44.5 | Helped |
| Increase graph size from 100 to 200 or 400 nodes (ablation) | M&M with colorization pre-training and AlexNet · PASCAL VOC2012 val | mIoU | Helped |
| Combine ImageNet and M&M predictions by averaging probabilities | VGG-16 · PASCAL VOC2012 val | mIoU: 67.4 | Helped |
Closest papers by meaning
- Colorization as a Proxy Task for Visual Understanding (CVPR 2017 · similarity 0.88)
- Three Ways To Improve Semantic Segmentation With Self-Supervised Depth Estimation (CVPR 2021 · similarity 0.87)
- Learning Pixel-Level Semantic Affinity With Image-Level Supervision for Weakly Supervised Semantic Segmentation (CVPR 2018 · similarity 0.87)
- Learning from Future: A Novel Self-Training Framework for Semantic Segmentation (NeurIPS 2022 · similarity 0.87)
- $\mathbf{D^{3}S^{2}}$: Diffusion-Guided Dataset Distillation for Semantic Segmentation (NeurIPS 2026 · similarity 0.85)
- Integrating Low-Level Visual Cues for Enhanced Unsupervised Semantic Segmentation (AAAI 2025 · similarity 0.85)
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.