Install

Envs

conda create -n plant_kaggle python=3.8
conda activate plant_kaggle
pip install -r requirements.txt

If you have GPU, Please install pytorch corresponding to the CUDA version because it could faster your training.

DataSet

链接：https://pan.baidu.com/s/1N10URTnXCaWbBWlRLISAWg

提取码：zvjs

This dataset is provided by the Prof.Liang

Run

Install the extensions for jupyter Notebook And change the work dir path

Change the jupyter kernel to the plant_kaggle and run all.

Result

Histogram.png and PieChart.png show the distribution of the dataset

The figure in the ./Result/ dir are the results I trained.

And Swin-B base model show the best performance, The ACC is 0.8716666666666667