This is the official code repository for "UCM-NetV2 and BNN-UCM-NetV2: Efficient and Accurate Deep Learning Models for Skin Lesion Segmentation "
Prepare the dataset.
- The ISIC2017 and ISIC2018 datasets,which follows the same as UltraLight-VM-UNet , can be found here {[One Drive]}.
- {ISIC2017}
- {ISIC2018}
- (PS: you can use above link to directly download the dataset and put them in the data folder or use UltraLight-VM-UNet code to generate the division of dataset)
Train the UCMV2-UNet.
cd UCMV2-UNet
# ISIC2017
python trainucm.py --data 'ISIC2017'
# ISIC2018
python trainucm.py --data 'ISIC2018'
** Citation **
@article{yuan2024ucm,
title={UCM-NetV2 and BNN-UCM-NetV2: Efficient and Accurate Deep Learning Models for Skin Lesion Segmentation on Mobile Devices},
author={Yuan, Chunyu and Zhao, Dongfang and Agaian, Sos S},
year={2024},
publisher={Preprints}
}
Acknowledgement.
Some code,dataset generaation, data loading, model trainning , are cited from UltraLight-VM-UNet {UltraLight-VM-UNet }.
We thank their wonderful works!