XNet icon indicating copy to clipboard operation
XNet copied to clipboard

About mask input

Open sleepy1231 opened this issue 1 year ago • 1 comments

作者您好!请问mask的输出应该是什么样的?也就是在dataset_2d.py中180行左右的: augment_1 = self.augmentation_1(image=img_1, image2=img_2, mask=mask) img_1 = augment_1['image'] img_2 = augment_1['image2'] mask = augment_1['mask'] normalize_1 = self.normalize_1(image=img_1, mask=mask) img_1 = normalize_1['image'] mask = normalize_1['mask'] mask = mask.long() ..... sampel = {'image': img_1, 'image_2': img_2, 'mask': mask, 'ID': os.path.split(mask_path)[1]} return sampel

也就是最后一行的输出中,经过处理后的mask是前景=1,背景=0吗? 另一个问题是:为什么mask标准化采用的是L的均值和方差?

sleepy1231 avatar Dec 10 '23 12:12 sleepy1231