FSSD-12 dataset
We integrate a novel few-shot segmentation dataset (FSSD-12)of pure strip steel surface defect, including 12 different surface defects of strip steel.
CPANet
We propose a few-shot semantic segmentation method named cross position aggregation network (CPANet) to solve the number of strip steel surface defect is deficient and the ground truth is difficultly labeled.
Download the dataset and code
The dataset and code are available at:https://pan.baidu.com/s/1_BORNJrO4msD0OPEcVSc-Q?pwd=9m10
Paper
https://ieeexplore.ieee.org/document/10049179
2023-Cross Position Aggregation Network for Few-Shot Strip Steel Surface Defect Segmentation.pdf
Citation
H. Feng, K. Song, W. Cui, Y. Zhang and Y. Yan, "Cross Position Aggregation Network for Few-shot Strip Steel Surface Defect Segmentation," in IEEE Transactions on Instrumentation and Measurement, doi: 10.1109/TIM.2023.3246519.
Related work of few-shot surface defect segmentation
[1] TGRNet: Triplet-graph reasoning network for few-shot metal generic surface defect segmentation
The dataset and code are available at:https://pan.baidu.com/s/1dEai3yXrFOsuWcQ5mkE7_A?pwd=qzo6
Few-shot classification for surface defects
[1] FSC-20 dataset & FaNet The dataset and code are available at:https://github.com/VDT-2048/FSC-20
[2] MSD-Cls dataset & GTnet The dataset and code are available at:https://github.com/successhaha/GTnet