Research map / Detection and segmentation / CNN features
End to End Trainable Active Contours via Differentiable Rendering
Shir Gur, Tal Shaharabany, Lior Wolf
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 the full differentiable polygon contour method | Ours · Cityscapes val | mean IoU: 73.70 -> 75.09 | Improved |
| Add ballooning and curvature losses to segmentation loss (ablation) | Ours · Vaihingen | F1-Score: 94.94 -> 95.62 | Improved |
| Add ballooning and curvature losses to segmentation loss (ablation) | Ours · Vaihingen | mIoU: 90.31 -> 91.74 | Improved |
| Add ballooning loss (ablation) | Ours · Vaihingen | F1-Score: 94.94 -> 95.13 | Improved |
| Add ballooning loss (ablation) | Ours · Vaihingen | mIoU: 90.31 -> 90.82 | Improved |
| Add curvature loss without ballooning loss (ablation) | Ours · Vaihingen | F1-Score: 94.94 -> 94.80 | Hurt |
| Add curvature loss without ballooning loss (ablation) | Ours · Vaihingen | mIoU: 90.31 -> 90.20 | Hurt |
| Train at resolution 64 instead of 16 (ablation) | Ours · Vaihingen | F1-Score: 13.11 -> 95.62 | Improved |
| Train at resolution 64 instead of 32 (ablation) | Ours · Vaihingen | mIoU: 39.50 -> 91.74 | Improved |
| Train at resolution 128 instead of 64 (ablation) | Ours · Vaihingen | F1-Score: 95.62 -> 95.13 | Hurt |
| Train at resolution 128 instead of 64 (ablation) | Ours · Vaihingen | mIoU: 91.74 -> 90.85 | Hurt |
Closest papers by meaning
- End-to-End Trainable Deep Active Contour Models for Automated Image Segmentation: Delineating Buildings in Aerial Imagery (ECCV 2020 · similarity 0.85)
- A General Differentiable Mesh Renderer for Image-Based 3D Reasoning (TPAMI 2020 · similarity 0.84)
- Soft Rasterizer: A Differentiable Renderer for Image-Based 3D Reasoning (ICCV 2019 · similarity 0.84)
- Deep Active Surface Models (CVPR 2021 · similarity 0.84)
- Learning Active Contour Models for Medical Image Segmentation (CVPR 2019 · similarity 0.84)
- EFDTR: Learnable Elliptical Fourier Descriptor Transformer for Instance Segmentation (ICML 2025 · similarity 0.83)
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.
