Research map / Label-efficient and robust learning / Recognition and long-tail
Zero Pixel Directional Boundary by Vector Transform
Edoardo Mello Rella, Ajad Chhatkuli, Yun Liu, Ender Konukoglu, Luc Van Gool
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (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 vector-field boundary representation instead of binary boundary classification | Boundary detection with the shared network architecture · Cityscapes | assd: 4.71 -> 4.33 | Improved |
| use vector-field boundary representation instead of binary boundary classification | Boundary detection with the shared network architecture · Cityscapes | ODS F: 0.764 -> 0.845 | Improved |
| use vector-field boundary representation instead of binary boundary classification | Boundary detection with the shared network architecture · Synthia | assd: 1.91 -> 1.67 | Improved |
| use vector-field boundary representation instead of binary boundary classification | Boundary detection with the shared network architecture · Synthia | ODS F: 0.900 -> 0.819 | Hurt |
| use vector-field boundary representation instead of binary boundary classification | Boundary detection with the shared network architecture · Mapillary Vistas | assd: 4.35 -> 3.60 | Improved |
| use vector-field boundary representation instead of binary boundary classification | Boundary detection with the shared network architecture · Mapillary Vistas | ODS F: 0.760 -> 0.806 | Improved |
| use vector-field boundary representation instead of binary boundary classification | Boundary detection with the shared network architecture · BSDS500 | assd: 5.66 -> 5.83 | Hurt |
| use vector-field boundary representation instead of binary boundary classification | Boundary detection with the shared network architecture · BSDS500 | ODS F: 0.688 -> 0.676 | Hurt |
| replace vector-field prediction with distance-transform prediction trained using L1 loss (ablation) | VT boundary detection · Cityscapes | assd: 4.33 -> 5.63 | Hurt |
| replace vector-field prediction with distance-transform prediction trained using L1 loss (ablation) | VT boundary detection · Synthia | assd: 1.67 -> 3.74 | Hurt |
| replace vector-field prediction with distance-transform prediction trained using L1 loss (ablation) | VT boundary detection · Mapillary Vistas | assd: 3.60 -> 7.13 | Hurt |
| replace vector-field prediction with distance-transform prediction trained using L1 loss (ablation) | VT boundary detection · BSDS500 | assd: 5.83 -> 5.87 | No clear effect |
Closest papers by meaning
- Learning to Predict Crisp Boundaries (ECCV 2018 · similarity 0.89)
- Learning on the Edge: Investigating Boundary Filters in CNNs (IJCV 2019 · similarity 0.87)
- Convolutional Oriented Boundaries (ECCV 2016 · similarity 0.85)
- Instance Segmentation With Mask-Supervised Polygonal Boundary Transformers (CVPR 2022 · similarity 0.85)
- Rethinking Boundary Discontinuity Problem for Oriented Object Detection (CVPR 2024 · similarity 0.84)
- EFDTR: Learnable Elliptical Fourier Descriptor Transformer for Instance Segmentation (ICML 2025 · similarity 0.84)
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.
