Research map / Detection and segmentation / Open-set and few-shot detection
PANet: Few-Shot Image Semantic Segmentation With Prototype Alignment
Kaixin Wang, Jun Hao Liew, Yingtian Zou, Daquan Zhou, Jiashi Feng
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 prototype alignment regularization (PAR) (ablation) | PANet · PASCAL-5i 1-shot | mean-IoU: 47.2 -> 48.1 | Improved |
| add prototype alignment regularization (PAR) (ablation) | PANet · PASCAL-5i 5-shot | mean-IoU: 54.9 -> 55.7 | Improved |
| add PAR: distance between query and support prototypes (ablation) | PANet · PASCAL-5i split-1 5-shot | prototype Euclidean distance (lower is better): 42.6 -> 32.2 | Improved |
| train the network on PASCAL-5i instead of ImageNet-init only | PANet-init to PANet · PASCAL-5i 1-shot | mean-IoU: 35.3 -> 48.1 | Improved |
| train the network on PASCAL-5i instead of ImageNet-init only | PANet-init to PANet · PASCAL-5i 5-shot | mean-IoU: 46.7 -> 55.7 | Improved |
| use scribble instead of dense support annotation | PANet · PASCAL-5i 1-shot | mean-IoU: 48.1 -> 44.8 | Hurt |
| use scribble instead of dense support annotation | PANet · PASCAL-5i 5-shot | mean-IoU: 55.7 -> 54.6 | Hurt |
| use bounding box instead of dense support annotation | PANet · PASCAL-5i 1-shot | mean-IoU: 48.1 -> 45.1 | Hurt |
| use bounding box instead of dense support annotation | PANet · PASCAL-5i 5-shot | mean-IoU: 55.7 -> 52.8 | Hurt |
| use scribbles rather than bounding boxes with more support | PANet · PASCAL-5i 5-shot | mean-IoU: 54.6 -> 52.8 | Improved |
| increase support set from 1 to 5 shots | PANet · PASCAL-5i | mean-IoU: 48.1 -> 55.7 | Improved |
| increase support set from 1 to 5 shots | co-FCN · PASCAL-5i | mean-IoU: 41.1 -> 41.4 | No clear effect |
Later papers that compared with this method
41 later accepted papers report PANet as a baseline in their results tables; 33 of them report a better number on most of the results they share with it.
- Seeing Through the Noise: Improving Infrared Small Target Detection and Segmentation from Noise Suppression Perspective (CVPR 2026 · better on 1 of 1 shared result)
- WISNet: Pseudo Label Generation on Unbalanced and Patch Annotated Waste Images (CVPR 2025 · better on 1 of 1 shared result)
- UniverSeg: Universal Medical Image Segmentation (ICCV 2023 · better on 2 of 2 shared results)
- Harmonizing Base and Novel Classes: A Class-Contrastive Approach for Generalized Few-Shot Segmentation (IJCV 2023 · better on 2 of 2 shared results)
- Occlusion-Aware Instance Segmentation Via BiLayer Network Architectures (TPAMI 2023 · better on 1 of 1 shared result)
- Generalized Few-Shot Semantic Segmentation (CVPR 2022 · better on 2 of 2 shared results)
- Dual Contrastive Learning with Anatomical Auxiliary Supervision for Few-Shot Medical Image Segmentation (ECCV 2022 · better on 2 of 2 shared results)
- Cross-Domain Few-Shot Semantic Segmentation (ECCV 2022 · better on 2 of 2 shared results)
33 more not listed.
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
- Part-aware Prototype Network for Few-shot Semantic Segmentation (ECCV 2020 · similarity 0.90)
- Simpler Is Better: Few-Shot Semantic Segmentation With Classifier Weight Transformer (ICCV 2021 · similarity 0.90)
- Cross-Domain Few-Shot Semantic Segmentation (ECCV 2022 · similarity 0.89)
- Learning from the Target: Dual Prototype Network for Few Shot Semantic Segmentation (AAAI 2022 · similarity 0.89)
- Unlocking the Potential of Pre-trained Vision Transformers for Few-Shot Semantic Segmentation through Relationship Descriptors (CVPR 2024 · similarity 0.88)
- A Surprisingly Simple Approach to Generalized Few-Shot Semantic Segmentation (NeurIPS 2024 · similarity 0.87)
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.
