Research map / Detection and segmentation / CNN features
RepSNet: A Nucleus Instance Segmentation Model Based on Boundary Regression and Structural Re-Parameterization
Shengchun Xiong, Xiangru Li, Yunpeng Zhong, Wanfen Peng
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 |
|---|---|---|---|
| add RepVgg units in encoder (ablation) | NBRIA(RepSNet) without RepUpsample and L_nb · Lizard/CoNIC local test | mPQ: 0.5324 -> 0.5537 | Improved |
| add RepUpsample in decoder (ablation) | NBRIA(RepSNet) with RepVgg · Lizard/CoNIC local test | mPQ: 0.5537 -> 0.5612 | Improved |
| add nucleus boundary loss L_nb (ablation) | NBRIA(RepSNet) with RepVgg and RepUpsample · Lizard/CoNIC local test | mPQ: 0.5612 -> 0.5633 | Improved |
| replace Hover-Net post-processing with BVM (ablation) | Hover-Net · Lizard/CoNIC local test | mPQ: 0.5314 -> 0.5410 | Improved |
| replace StarDist post-processing with BVM (ablation) | StarDist · Lizard/CoNIC local test | mPQ: 0.5472 -> 0.5543 | Improved |
| add L_nb loss (ablation) | StarDist · Lizard/CoNIC local test | mPQ: 0.5472 -> 0.5512 | Improved |
| add L_nb loss (ablation) | DCAN · Lizard/CoNIC local test | mPQ: 0.4799 -> 0.4844 | Improved |
| use RepSNet encoder and decoder (ablation) | Hover-Net · Lizard/CoNIC local test | mPQ: 0.5314 -> 0.5355 | Improved |
| use RepSNet encoder and decoder (ablation) | StarDist · Lizard/CoNIC local test | mPQ: 0.5472 -> 0.5541 | Improved |
| deepen RepSNet to n=5 blocks [2,3,3,6,3] (ablation) | RepSNet basic · CoNIC online test set | mPQ+: 0.428 -> 0.440 | Improved |
| add oversampling and H&E augmentation for minority classes (ablation) | RepSNet · CoNIC online test set | mPQ+: 0.440 -> 0.455 | Improved |
| adjust loss weights toward neu and eos classes (ablation) | RepSNet · CoNIC online test set | mPQ+: 0.455 -> 0.461 | Improved |
Closest papers by meaning
- Unleashing the Power of Prompt-driven Nucleus Instance Segmentation (ECCV 2024 · similarity 0.87)
- CDNet: Centripetal Direction Network for Nuclear Instance Segmentation (ICCV 2021 · similarity 0.85)
- Minimizing Labeling Cost for Nuclei Instance Segmentation and Classification with Cross-domain Images and Weak Labels (AAAI 2021 · similarity 0.85)
- CiNuSeg: Class Incremental Nuclei Segmentation via Anchor-driven Consistency Learning with Dual Region Regularization (AAAI 2026 · similarity 0.85)
- Incremental Nuclei Segmentation from Histopathological Images via Future-class Awareness and Compatibility-inspired Distillation (CVPR 2024 · similarity 0.84)
- Mudslide: A Universal Nuclear Instance Segmentation Method (CVPR 2024 · 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.
