Research map / Detection and segmentation / CNN features
CASENet: Deep Category-Aware Semantic Edge Detection
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 |
|---|---|---|---|
| use CASENet instead of DSN | ResNet-101 semantic edge detection · SBD test | MF: 70.3 -> 71.4 | Helped |
| use CASENet instead of DSN | ResNet-101 semantic edge detection · Cityscapes validation | MF: 68.5 -> 71.3 | Helped |
| use CASENet instead of DSN | ResNet-101 semantic edge detection · Cityscapes validation | AP: 66.4 -> 70.8 | Helped |
| add Side-5 supervision (ablation) | CASENet · SBD test | MF: 70.7 -> 71.4 | Helped |
| replace multi-label loss with reweighted softmax loss (ablation) | Basic architecture with ResNet · SBD test | MF: 70.6 -> 60.2 | Hurt |
| replace multi-label loss with reweighted softmax loss (ablation) | Basic architecture with VGG · SBD test | MF: 55.1 -> 51.1 | Hurt |
| use CASENet architecture instead of Basic (ablation) | ResNet-101 semantic edge detection · SBD test | MF: 70.6 -> 71.4 | Helped |
| use CASENet architecture instead of Basic (ablation) | VGG semantic edge detection · SBD test | MF: 55.1 -> 56.8 | Helped |
| supervise lower sides with class-specific edge labels (ablation) | CASENet side features · SBD test | MF | Hurt |
| keep stride 2 in res1 (ablation) | CASENet · SBD at lower input resolution | MF | Hurt |
| add binary edge losses to Side-1–3 (ablation) | CASENet-edge · SBD | MF | Hurt |
Closest papers by meaning
- Coupled Segmentation and Edge Learning via Dynamic Graph Propagation (NeurIPS 2021 · similarity 0.87)
- Devil Is in the Edges: Learning Semantic Boundaries From Noisy Annotations (CVPR 2019 · similarity 0.87)
- Semantic Image Segmentation With Task-Specific Edge Detection Using CNNs and a Discriminatively Trained Domain Transform (CVPR 2016 · similarity 0.87)
- Unsupervised Learning of Edges (CVPR 2016 · similarity 0.87)
- Visual Boundary Knowledge Translation for Foreground Segmentation (AAAI 2021 · similarity 0.87)
- Edge-Aware 3D Instance Segmentation Network with Intelligent Semantic Prior (CVPR 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.