1.cv2.imread():
shape: (H,W,C)
dtype: uint8
channel: BGR
2. Image.open()
size: (w,h)
channel: RGB
3. T.ToTensor(),
把PIL.Image或ndarray从 (H x W x C)形状转换为 (C x H x W) 的tensor
还会将数值从 [0, 255] 归一化到[0,1]
1.cv2.imread():
shape: (H,W,C)
dtype: uint8
channel: BGR
2. Image.open()
size: (w,h)
channel: RGB
3. T.ToTensor(),
把PIL.Image或ndarray从 (H x W x C)形状转换为 (C x H x W) 的tensor
还会将数值从 [0, 255] 归一化到[0,1]