torchvision
transforms
from torchvision import transforms
transform = transforms.Compose([
transforms.ToTensor(),
transforms.Normalize(mean = (0.5, 0.5, 0.5), std = (0.5, 0.5, 0.5)),
])
ToTensor() 將 PIL.Image 或 numpy.ndarray 數(shù)據(jù)進(jìn)轉(zhuǎn)化為 torch.FloadTensor擂送，并歸一化到[0, 1.0]灰粮。
- 形狀為
[H, W, C]的numpy.ndarray隐绵，轉(zhuǎn)換成形狀為[C, H, W]，取值范圍是[0, 1.0]的torch.FloadTensor届搁。 - 取值范圍為
[0, 255]的PIL.Image帕识，轉(zhuǎn)換成形狀為[C, H, W]搔课，取值范圍是[0, 1.0]的torch.FloadTensor
