Research map / Detection and segmentation / CNN features
Eigencontours: Novel Contour Descriptors Based on Low-Rank Approximation
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 |
|---|---|---|---|
| represent contours with an SVD-derived eigencontour basis rather than PolarMask descriptors | Contour description · KINS | AUC-F: 75.47 -> 89.17 | Helped |
| represent contours with an SVD-derived eigencontour basis rather than ESE-Seg descriptors | Contour description · SBD | AUC-F: 76.86 -> 86.51 | Helped |
| represent contours with an SVD-derived eigencontour basis rather than PolarMask descriptors | Contour description · COCO2017 | AUC-F: 74.05 -> 76.92 | Helped |
| cluster contours using eigencontour coefficients rather than ESE-Seg descriptors | Contour clustering · COCO2017 (M=16, K=500) | F: 26.82 -> 27.40 | Helped |
| regress eigencontour coefficients for instance masks rather than ESE-Seg contour descriptors | YOLOv3 instance segmentation · SBD validation | AP50: 52.14 -> 56.47 | Helped |
| use M=10 instead of M=20 eigencontours (ablation) | Eigencontours with YOLOv3 · SBD validation | AP50: 56.47 -> 49.96 | Hurt |
| increase the eigencontour dimension from M=20 to M=40 (ablation) | Eigencontours with YOLOv3 · SBD validation | AP75: 29.35 -> 22.42 | Hurt |
| use category-specific eigencontour spaces instead of one universal space (ablation) | Eigencontours · KINS | AUC-F: 64.65 -> 67.67 | Helped |
| use category-specific eigencontour spaces instead of one universal space (ablation) | Eigencontours · COCO2017 | AUC-F: 47.19 -> 48.77 | Helped |
Closest papers by meaning
- EFDTR: Learnable Elliptical Fourier Descriptor Transformer for Instance Segmentation (ICML 2025 · similarity 0.87)
- E2EC: An End-to-End Contour-Based Method for High-Quality High-Speed Instance Segmentation (CVPR 2022 · similarity 0.85)
- ShapeEmbed: a self-supervised learning framework for 2D contour quantification (NeurIPS 2025 · similarity 0.84)
- EAGLE: Eigen Aggregation Learning for Object-Centric Unsupervised Semantic Segmentation (CVPR 2024 · similarity 0.83)
- The Devil is in the Object Boundary: Towards Annotation-free Instance Segmentation using Foundation Models (ICLR 2024 · similarity 0.82)
- Rethinking Decoders for Transformer-based Semantic Segmentation: A Compression Perspective (NeurIPS 2024 · similarity 0.82)
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.