unet

Introduction

Official Repo

Code Snippet

UNet (MICCAI'2016/Nat. Methods'2019)

@inproceedings{ronneberger2015u,
    title={U-net: Convolutional networks for biomedical image segmentation},
    author={Ronneberger, Olaf and Fischer, Philipp and Brox, Thomas},
    booktitle={International Conference on Medical image computing and computer-assisted intervention},
    pages={234--241},
    year={2015},
    organization={Springer}
}

Results

HRF

Segmentor	Pretrain	Backbone	Crop Size	Schedule	Train/Eval Set	Dice	Download
FCN	-	UNet-S5-D16	256x256	LR/POLICY/BS/EPOCH: 0.01/poly/16/1	train/val	79.88%	cfg | model | log
PSPNet	-	UNet-S5-D16	256x256	LR/POLICY/BS/EPOCH: 0.01/poly/16/1	train/val	80.26%	cfg | model | log
DeepLabV3	-	UNet-S5-D16	256x256	LR/POLICY/BS/EPOCH: 0.01/poly/16/1	train/val	80.29%	cfg | model | log

DRIVE

Segmentor	Pretrain	Backbone	Crop Size	Schedule	Train/Eval Set	Dice	Download
FCN	-	UNet-S5-D16	64x64	LR/POLICY/BS/EPOCH: 0.01/poly/16/1	train/val	78.67%	cfg | model | log
PSPNet	-	UNet-S5-D16	64x64	LR/POLICY/BS/EPOCH: 0.01/poly/16/1	train/val	78.77%	cfg | model | log
DeepLabV3	-	UNet-S5-D16	64x64	LR/POLICY/BS/EPOCH: 0.01/poly/16/1	train/val	78.96%	cfg | model | log

STARE

Segmentor	Pretrain	Backbone	Crop Size	Schedule	Train/Eval Set	Dice	Download
FCN	-	UNet-S5-D16	128x128	LR/POLICY/BS/EPOCH: 0.01/poly/16/1	train/val	81.03%	cfg | model | log
PSPNet	-	UNet-S5-D16	128x128	LR/POLICY/BS/EPOCH: 0.01/poly/16/1	train/val	81.24%	cfg | model | log
DeepLabV3	-	UNet-S5-D16	128x128	LR/POLICY/BS/EPOCH: 0.01/poly/16/1	train/val	81.19%	cfg | model | log

CHASE DB1

Segmentor	Pretrain	Backbone	Crop Size	Schedule	Train/Eval Set	Dice	Download
FCN	-	UNet-S5-D16	128x128	LR/POLICY/BS/EPOCH: 0.01/poly/16/1	train/val	80.50%	cfg | model | log
PSPNet	-	UNet-S5-D16	128x128	LR/POLICY/BS/EPOCH: 0.01/poly/16/1	train/val	80.50%	cfg | model | log
DeepLabV3	-	UNet-S5-D16	128x128	LR/POLICY/BS/EPOCH: 0.01/poly/16/1	train/val	80.54%	cfg | model | log

More

You can also download the model weights from following sources:

• BaiduNetdisk: https://pan.baidu.com/s/1gD-NJJWOtaHCtB0qHE79rA with access code s757