Retina-Unet

来源：https://github.com/orobix/retina-unet 增加dilated_conv、batchnormalization、PReLU、LeakyReLU、resnet34 backbone的实验

此代码已经针对Python3进行了优化，数据集下载：http://www.isi.uu.nl/Research/Databases/DRIVE/
百度网盘数据集下载：https://pan.baidu.com/s/1UpQtbhgtVdlqc0QGbESx3A 密码：4l7v

有关代码内容讲解，请参见CSDN博客：
基于UNet的眼底图像血管分割实例：https://blog.csdn.net/u013063099/article/details/79981097

【注意】run_training.py与run_testing.py的实际作用为了让程序在后台运行，如果运行出现错误，可以运行src目录下的训练与预测文件。

Name		Name	Last commit message	Last commit date
Latest commit History 4 Commits
.idea		.idea
.ipynb_checkpoints		.ipynb_checkpoints
DRIVE		DRIVE
__pycache__		__pycache__
lib		lib
src		src
test_binary_crossentropy_bn_LeakyReLU_lr=0.01, decay=2e-5		test_binary_crossentropy_bn_LeakyReLU_lr=0.01, decay=2e-5
test_binary_crossentropy_bn_PReLU_lr=0.01, decay=2e-5		test_binary_crossentropy_bn_PReLU_lr=0.01, decay=2e-5
test_binary_crossentropy_dilate_bn_LeakyReLU_lr=0.01, decay=2e-5		test_binary_crossentropy_dilate_bn_LeakyReLU_lr=0.01, decay=2e-5
test_binary_crossentropy_dilate_lr=0.01, decay=1e-6		test_binary_crossentropy_dilate_lr=0.01, decay=1e-6
test_binary_crossentropy_dilate_lr=0.01, decay=2e-5		test_binary_crossentropy_dilate_lr=0.01, decay=2e-5
test_binary_crossentropy_lr=0.01, decay=1e-6		test_binary_crossentropy_lr=0.01, decay=1e-6
test_categorical_crossentropy_lr=0.01, decay=1e-6		test_categorical_crossentropy_lr=0.01, decay=1e-6
test_dice_coef_lr=0.01, decay=1e-6		test_dice_coef_lr=0.01, decay=1e-6
README.md		README.md
Untitled.ipynb		Untitled.ipynb
Untitled1.ipynb		Untitled1.ipynb
configuration.txt		configuration.txt
mylr.py		mylr.py
prepare_datasets_DRIVE.py		prepare_datasets_DRIVE.py
retinaNN_training.py		retinaNN_training.py
run_testing.py		run_testing.py
run_training.py		run_training.py

Provide feedback