Pytorch Notes

This notes mostly comes from source code from pytorch just to summarize the most import part of those most frequently used functions.

torchvision

transforms

..ToTensor()

torchvision.tansforms.functional(pic): Convert a PIL Image or numpy.ndarray to tensor.

# This will divide image pixel value by 255.0 to normalize it into [0, 1.0]
img = img.transpose(0, 1).transpose(0, 2).contiguous()
if isinstance(img, torch.ByteTensor):
    return img.float().div(255)
else:
    return img

..Normalize()

Using *.*.functional.normalize() function: Normalize a tensor image with mean and standare deviation:

1
2
3

mean = torch.as_tensor(mean, dtype=dtype, device=tensor.device)
std = torch.as_tensor(std, dtype=dtype, device=tensor.device)
tensor.sub_(mean[:, None, None]).div_(std[:, None, None])