关于RCF算法中数据增加的问题

[Ying-sky999]

作者您好，请问在RCF数据增强中，val文件表示什么意思
train_lst = os.listdir(osp.join(input_dir, 'image', 'train'))
val_lst = os.listdir(osp.join(input_dir, 'image', 'val'))
test_lst = os.listdir(osp.join(input_dir, 'image', 'test'))
还有就是论文中提到的 0.5 1.0 1.5这三种比例的图像处理是如何实现的
希望您不吝赐教，如果可以的话您能否提供相关代码
https://github.com/yun-liu/RCF/issues/124

[yun-liu]

val 表示验证集。三种尺度的图像，用resize实现。数据增强的代码已经在你贴的issue里给了：#124

[RuosiZhou]

Is there any BSDS data extension code? If you can share, I would be very grateful.

[yun-liu]

@RuosiZhou I downloaded the BSDS500 training data from the well-known work HED. You can either use the public BSDS500 data or write the augmentation code following the issue #124.

[RuosiZhou]

Ok, it has been solved, thank you