Research map / Detection and segmentation / Open-set and few-shot detection
Harmonizing Base and Novel Classes: A Class-Contrastive Approach for Generalized Few-Shot Segmentation
Weide Liu, Zhonghua Wu, Yang Zhao, Yuming Fang, Chuan-Sheng Foo, Jun Ting Cheng, Guosheng Lin
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 class contrastive loss L_C (ablation) | CAPL baseline · PASCAL VOC GFSSeg 1-shot (mean 4 folds) | average mIoU: 53.89 -> 56.69 | Improved |
| add class contrastive loss L_C (ablation) | CAPL baseline · PASCAL VOC GFSSeg 5-shot (mean 4 folds) | average mIoU: 56.79 -> 59.40 | Improved |
| add self-similarity loss L_W (ablation) | CAPL baseline · PASCAL VOC GFSSeg 1-shot (mean 4 folds) | average mIoU: 53.89 -> 57.79 | Improved |
| add self-similarity loss L_W (ablation) | CAPL baseline · PASCAL VOC GFSSeg 5-shot (mean 4 folds) | average mIoU: 56.79 -> 60.55 | Improved |
| add cross-class similarity loss L_B (ablation) | CAPL baseline · PASCAL VOC GFSSeg 1-shot (mean 4 folds) | average mIoU: 53.89 -> 57.92 | Improved |
| add cross-class similarity loss L_B (ablation) | CAPL baseline · PASCAL VOC GFSSeg 5-shot (mean 4 folds) | average mIoU: 56.79 -> 60.78 | Improved |
| add L_C, L_W and L_B together (ablation) | CAPL baseline · PASCAL VOC GFSSeg 1-shot (mean 4 folds) | average mIoU: 53.89 -> 58.14 | Improved |
| add L_C, L_W and L_B together (ablation) | CAPL baseline · PASCAL VOC GFSSeg 5-shot (mean 4 folds) | average mIoU: 56.79 -> 61.50 | Improved |
| add L_C on top of L_W and L_B (ablation) | class relationship losses · PASCAL VOC GFSSeg 1-shot (mean 4 folds) | average mIoU: 57.96 -> 58.14 | No clear effect |
| learnable edges (L_W + L_B) instead of fixed edges L_h (ablation) | class relationship loss · PASCAL VOC GFSSeg 5-shot (mean 4 folds) | average mIoU: 60.57 -> 60.91 | Improved |
| fixed-edge class relationship loss L_h vs baseline (ablation) | CAPL baseline · PASCAL VOC GFSSeg 1-shot (mean 4 folds) | average mIoU: 53.89 -> 57.86 | Improved |
| add L_C alone for novel class fold 2 (1-shot) (ablation) | CAPL baseline · PASCAL VOC GFSSeg 1-shot fold 2 | novel mIoU: 20.34 -> 20.20 | No clear effect |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 2 of 2 on most shared results.
- PFENet — PFENet++: Boosting Few-Shot Semantic Segmentation With the Noise-Filtered Context-Aware Prior Mask (TPAMI 2023) · better on 2 of 2 shared results
- PANet — PANet: Few-Shot Image Semantic Segmentation With Prototype Alignment (ICCV 2019) · better on 2 of 2 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- A Surprisingly Simple Approach to Generalized Few-Shot Semantic Segmentation (NeurIPS 2024 · similarity 0.88)
- Few-Shot Segmentation via Divide-and-Conquer Proxies (IJCV 2023 · similarity 0.87)
- Evidential Robust Feature Learning for Generalized Few-Shot Segmentation (IJCV 2026 · similarity 0.87)
- A Strong Baseline for Generalized Few-Shot Semantic Segmentation (CVPR 2023 · similarity 0.86)
- Learning What Not To Segment: A New Perspective on Few-Shot Segmentation (CVPR 2022 · similarity 0.86)
- Enhancing Generalized Few-Shot Semantic Segmentation via Effective Knowledge Transfer (AAAI 2025 · similarity 0.86)
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.
