Research map / Detection and segmentation / CNN features
E2EC: An End-to-End Contour-Based Method for High-Quality High-Speed Instance 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 |
|---|---|---|---|
| Replace handcrafted contour initialization with learnable contour initialization (ablation) | Deep Snake on SBD · SBD val | AP^msk: 54.4 -> 57.5; 3.1 AP^msk | Helped |
| Add learnable contour initialization architecture (ablation) | Deep Snake on SBD · SBD val | AP^bdy: 10.8 -> 17.0; 6.2 AP^bdy | Helped |
| Replace uniform label sampling with multi-direction alignment (ablation) | E2EC architecture on SBD · SBD val | AP^msk: 57.5 -> 58.8; 1.3 AP^msk | Helped |
| Replace uniform label sampling with multi-direction alignment (ablation) | E2EC architecture on SBD · SBD val | AP^bdy: 17.0 -> 18.0; 1.0 AP^bdy | Helped |
| Add dynamic matching loss to the last deformation module (ablation) | E2EC with MDA on SBD · SBD val | AP^msk: 58.8 -> 59.2; 0.4 AP^msk | Helped |
| Add dynamic matching loss to the last deformation module (ablation) | E2EC with MDA on SBD · SBD val | AP^bdy: 18.0 -> 19.1; 1.1 AP^bdy | Helped |
| Set the MDA alignment number to 2 rather than 1 (ablation) | E2EC on SBD · SBD val | AP^msk: 57.5 -> 58.9 | Helped |
| Increase MDA alignment number from 2 to 8 (ablation) | E2EC on SBD · SBD val | AP^msk: 58.9 -> 58.4 | Hurt |
| Set the MDA alignment number to 4 rather than 2 (ablation) | E2EC on SBD · SBD val | AP^bdy: 17.1 -> 18.0 | Helped |
| Replace smooth L1 with Chamfer loss for the final deformation (ablation) | E2EC on SBD · SBD val | AP^msk: 58.8 -> 58.2; 0.6 AP^msk | Hurt |
| Replace smooth L1 with Chamfer loss for the final deformation (ablation) | E2EC on SBD · SBD val | AP^bdy: 18.0 -> 18.4; 0.4 AP^bdy | Helped |
| Remove the two deformation modules (ablation) | E2EC on COCO · COCO test-dev | AP: 33.8 -> 31.7 | Hurt |
Closest papers by meaning
- P2Seg: Pointly-supervised Segmentation via Mutual Distillation (ICLR 2024 · similarity 0.86)
- Training-Free Dataset Pruning for Instance Segmentation (ICLR 2025 · similarity 0.84)
- EFDTR: Learnable Elliptical Fourier Descriptor Transformer for Instance Segmentation (ICML 2025 · similarity 0.84)
- PolarNeXt: Rethink Instance Segmentation with Polar Representation (CVPR 2025 · similarity 0.83)
- The Devil is in the Object Boundary: Towards Annotation-free Instance Segmentation using Foundation Models (ICLR 2024 · similarity 0.83)
- W2P: Switching from Weak Supervision to Partial Supervision for Semantic Segmentation (AAAI 2024 · 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.