Research map / Detection and segmentation / CNN features
Convolutional Random Walk Networks for Semantic Image Segmentation
Check 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 the random walk layer and jointly learned affinities | DeepLab-LargeFOV, DeepLab-attention, and DeepLab-v2 · SBD | mean IOU: 73.8 -> 75.8; 79.0 -> 79.9; 81.4 -> 82.1 | Helped |
| Add RWN to DeepLab-LargeFOV | DeepLab-LargeFOV · Stanford Background | overall IOU: 65.74 -> 68.31 | Helped |
| Add RWN to DeepLab-LargeFOV | DeepLab-LargeFOV · Sift Flow | overall IOU: 67.31 -> 69.85 | Helped |
| Use RWN instead of CRF post-processing | DeepLab-v2 · SBD | overall IOU: 84.2 -> 84.3 | Helped |
| Use RWN instead of CRF post-processing | DeepLab-v2 · SBD predicted boundaries | MF: 0.763 -> 0.773 | Helped |
| Use RWN instead of CRF post-processing | DeepLab-v2 · SBD predicted boundaries | AP: 0.584 -> 0.595 | Helped |
| Apply more random walk steps until convergence rather than stopping early (ablation) | RWN · SBD | IOU | Helped |
| Use R=5 and random walk until convergence instead of R=40 and one step (ablation) | RWN · SBD | mean IOU: 75.5 -> 75.8 | Helped |
| Use R=5 and random walk until convergence instead of R=40 and one step (ablation) | RWN · SBD | overall IOU: 77.6 -> 77.9 | Helped |
| Vary the radius within R>3 and R<20 while running to convergence (ablation) | RWN · SBD | IOU | No clear effect |
| Use multiple random walk steps with R=40 rather than one step (ablation) | RWN · SBD | IOU | No clear effect |
| Add features from deeper FCN layers such as fc6 and fc7 to the affinity branch (ablation) | RWN affinity branch · SBD | segmentation performance | No clear effect |
Closest papers by meaning
- Semantic Segmentation With Boundary Neural Fields (CVPR 2016 · similarity 0.89)
- DifNet: Semantic Segmentation by Diffusion Networks (NeurIPS 2018 · similarity 0.89)
- 3D Semantic Segmentation With Submanifold Sparse Convolutional Networks (CVPR 2018 · similarity 0.88)
- Unsupervised Semantic Segmentation Through Depth-Guided Feature Correlation and Sampling (CVPR 2024 · similarity 0.87)
- Diffusion-Driven Two-Stage Active Learning for Low-Budget Semantic Segmentation (NeurIPS 2025 · similarity 0.87)
- DiffCut: Catalyzing Zero-Shot Semantic Segmentation with Diffusion Features and Recursive Normalized Cut (NeurIPS 2024 · 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.