Research map / Detection and segmentation / CNN features
Learning to Model Pixel-Embedded Affinity for Homogeneous Instance Segmentation
Read the paperCheck 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 pixel-embedded affinity modeling | Homogeneous instance segmentation · CVPPP A1 test | SBD: 91.1 -> 92.3 | Helped |
| Use pixel-embedded affinity modeling | Homogeneous instance segmentation · BBBC039V1 test | AJI: 0.8477 -> 0.8674 | Helped |
| Use pixel-embedded affinity modeling | Homogeneous instance segmentation · AC3/AC4 test | ARAND: 0.1220 -> 0.0938 | Helped |
| Add CCM to SCM (ablation) | Pixel-embedded affinity model · CVPPP A1 validation | SBD: 87.7 -> 88.1 | Helped |
| Add EPM to SCM (ablation) | Pixel-embedded affinity model · CVPPP A1 validation | SBD: 87.7 -> 88.5 | Helped |
| Use SCM, CCM, and EPM together (ablation) | Pixel-embedded affinity model · CVPPP A1 validation | SBD: 88.5 -> 89.1 | Helped |
| Add flipping and rotation transformations (ablation) | Cross-correlation training · CVPPP A1 validation | SBD: 86.5 -> 89.1 | Helped |
| Add cutout transformation (ablation) | Cross-correlation training · CVPPP A1 validation | SBD: 88.6 -> 89.1 | Helped |
| Add intensity adjustment (ablation) | Cross-correlation training · CVPPP A1 validation | SBD: 87.7 -> 89.1 | Helped |
| Extend inference affinity neighborhood from 4 to 8 (ablation) | Pixel-embedded affinity model · CVPPP A2 validation | SBD: 76.3 -> 77.1 | Helped |
| Extend inference affinity ranges and use 8-neighborhood (ablation) | Pixel-embedded affinity model · CVPPP A2 validation | SBD: 77.1 -> 77.4 | Helped |
Closest papers by meaning
- Sparse Object-Level Supervision for Instance Segmentation With Pixel Embeddings (CVPR 2022 · similarity 0.88)
- Recurrent Pixel Embedding for Instance Grouping (CVPR 2018 · similarity 0.87)
- Open-World Instance Segmentation: Exploiting Pseudo Ground Truth From Learned Pairwise Affinity (CVPR 2022 · similarity 0.87)
- Instance Segmentation of Biological Images Using Harmonic Embeddings (CVPR 2020 · similarity 0.87)
- The Four Color Theorem for Cell Instance Segmentation (ICML 2025 · similarity 0.84)
- WISH: Weakly Supervised Instance Segmentation using Heterogeneous Labels (CVPR 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.