Graph Embedding and Optimal Transport for Few-Shot Classification of Metal Surface Defect https://ieeexplore.ieee.org/document/9761830?source=authoralert
Requirements
To install requirements:
pip install -r requirements.txt
Dataset:
link：https://pan.baidu.com/s/14-x_blzNvtY7N5Ue1U2skw password：z584
Code：
link：https://pan.baidu.com/s/1H9ohxDf2qwxKkHgj9UQa2A password：aoi3
Datasets split:
Move the datafile to dataset/
Run 'python write_dataset_filelist.py'
Training
To train the feature extractors in the paper, run this command:
python train.py --dataset [miniImagenet/CUB] --method [S2M2_R/rotation] --model [WideResNet28_10/ResNet18] --train_aug
Evaluation
To evaluate my model on miniImageNet/CUB/cifar/cross, run: For miniImageNet/CUB
python save_plk.py
python test.py
Hyperparameter setting
common setting:
1-shot: k=10 kappa=9 beta=0.5 5-shot: k=4 kappa=1 beta=0.75
cross-domain setting1 :
1-shot: k=10 kappa=1 beta=0.7 5-shot: k=4 kappa=1 beta=0.6
cross-domain setting2 :
1-shot: k=10 kappa=1 beta=0.7 5-shot: k=4 kappa=1 beta=0.5
Contact the author e-mail：1900412@neu.edu.cn or 2878570391@qq.com