Research map / Detection and segmentation / CNN features
Region Mutual Information Loss for 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 RMI loss to CE | DeepLabv3 · PASCAL VOC 2012 val | mIoU: 77.14 -> 78.71 | Helped |
| add RMI loss to CE | DeepLabv3+ · PASCAL VOC 2012 val | mIoU: 78.17 -> 79.66 | Helped |
| add RMI loss to CE | DeepLabv3 · PASCAL VOC 2012 test | mIoU: 76.58 -> 78.58 | Helped |
| add RMI loss to CE | DeepLabv3+ · CamVid test | mIoU: 62.82 -> 65.06 | Helped |
| CRF post-processing vs CE | DeepLabv3+ · CamVid test | mIoU: 62.82 -> 61.68 | Hurt |
| affinity field loss vs CE | DeepLabv3 · PASCAL VOC 2012 val | mIoU: 77.14 -> 76.24 | Hurt |
| RMI vs CRF-5 | DeepLabv3 · PASCAL VOC 2012 val | mIoU / inference time: 78.40 -> 78.71; 0.71s -> 0.12s | Helped |
| use average pooling instead of interpolation for downsampling (ablation) | RMI · PASCAL VOC 2012 val | mIoU: 77.58 -> 78.71 | Helped |
| use average pooling instead of max pooling (ablation) | RMI · PASCAL VOC 2012 val | mIoU: 78.62 -> 78.71 | Helped |
| smaller downsampling factor (6 -> 2) (ablation) | RMI · PASCAL VOC 2012 val | mIoU: 78.28 -> 79.23 | Helped |
| region size R=3 instead of R=1 (ablation) | RMI · PASCAL VOC 2012 val | mIoU: 78.25 -> 78.71 | Helped |
| region size 4 instead of 3 (DF 4) (ablation) | RMI · PASCAL VOC 2012 val | mIoU: 78.71 -> 78.65 | No clear effect |
Closest papers by meaning
- Complex Wavelet Mutual Information Loss: A Multi-Scale Loss Function for Semantic Segmentation (ICML 2025 · similarity 0.88)
- Semi-Supervised Semantic Segmentation With Directional Context-Aware Consistency (CVPR 2021 · similarity 0.87)
- On the Relaxation of Conditional Independence Assumption for Image Segmentation (NeurIPS 2026 · similarity 0.86)
- Integrating Low-Level Visual Cues for Enhanced Unsupervised Semantic Segmentation (AAAI 2025 · similarity 0.86)
- Hierarchy-Agnostic Unsupervised Segmentation: Parsing Semantic Image Structure (NeurIPS 2024 · similarity 0.85)
- RankSEG-RMA: An Efficient Segmentation Algorithm via Reciprocal Moment Approximation (NeurIPS 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.