Use vgg16 and SAFF for small sample classification from the paper.

• Extract dataset features using pretrained vgg16

• SAFF converts features into 1D tensor

conda create -n zh python=3.9
conda activate zh
python3 -m pip install --upgrade pip
pip3 install -r requirements.txt

If your dataset is at path /hy-tmp/data Suppose you want to train on the UC dataset.

• Feature extraction

python run.py 
--data_path /hy-tmp/data 
--extract
--dataset UC

• Train & verify

python run.py 
--data_path /hy-tmp/data
--train
--dataset UC
--ratio 0.8