Torchvision.utils.save_image modifed the input tensor in place

Hi guys,

I found that after invoking save_image(input_tensor), the input_tensor is normalized to [0, 255]. Was it designed this way? it might not be appropriate for this function to modify input in-place.

Did anyone have the same problem?
The source code indicates that a view of input tensor was modified in-place.

def make_grid(tensor, nrow=8, padding=2,
              normalize=False, range=None, scale_each=False, pad_value=0):
    """Make a grid of images.

        tensor (Tensor or list): 4D mini-batch Tensor of shape (B x C x H x W)
            or a list of images all of the same size.
        nrow (int, optional): Number of images displayed in each row of the grid.
            The Final grid size is (B / nrow, nrow). Default is 8.
        padding (int, optional): amount of padding. Default is 2.
        normalize (bool, optional): If True, shift the image to the range (0, 1),
            by subtracting the minimum and dividing by the maximum pixel value.
        range (tuple, optional): tuple (min, max) where min and max are numbers,
            then these numbers are used to normalize the image. By default, min and max
            are computed from the tensor.
        scale_each (bool, optional): If True, scale each image in the batch of
            images separately rather than the (min, max) over all images.
        pad_value (float, optional): Value for the padded pixels.

        See this notebook `here <>`_

    if not (torch.is_tensor(tensor) or
            (isinstance(tensor, list) and all(torch.is_tensor(t) for t in tensor))):
        raise TypeError('tensor or list of tensors expected, got {}'.format(type(tensor)))

    # if list of tensors, convert to a 4D mini-batch Tensor
    if isinstance(tensor, list):
        tensor = torch.stack(tensor, dim=0)

    if tensor.dim() == 2:  # single image H x W
        tensor = tensor.view(1, tensor.size(0), tensor.size(1))
    if tensor.dim() == 3:  # single image
        if tensor.size(0) == 1:  # if single-channel, convert to 3-channel
            tensor =, tensor, tensor), 0)
        tensor = tensor.view(1, tensor.size(0), tensor.size(1), tensor.size(2))

    if tensor.dim() == 4 and tensor.size(1) == 1:  # single-channel images
        tensor =, tensor, tensor), 1)

    if normalize is True:
        tensor = tensor.clone()  # avoid modifying tensor in-place
        if range is not None:
            assert isinstance(range, tuple), \
                "range has to be a tuple (min, max) if specified. min and max are numbers"

        def norm_ip(img, min, max):
            img.clamp_(min=min, max=max)
            img.add_(-min).div_(max - min + 1e-5)

        def norm_range(t, range):
            if range is not None:
                norm_ip(t, range[0], range[1])
                norm_ip(t, float(t.min()), float(t.max()))

        if scale_each is True:
            for t in tensor:  # loop over mini-batch dimension
                norm_range(t, range)
            norm_range(tensor, range)

    if tensor.size(0) == 1:
        return tensor.squeeze()

    # make the mini-batch of images into a grid
    nmaps = tensor.size(0)
    xmaps = min(nrow, nmaps)
    ymaps = int(math.ceil(float(nmaps) / xmaps))
    height, width = int(tensor.size(2) + padding), int(tensor.size(3) + padding)
    grid = tensor.new_full((3, height * ymaps + padding, width * xmaps + padding), pad_value)
    k = 0
    for y in irange(ymaps):
        for x in irange(xmaps):
            if k >= nmaps:
            grid.narrow(1, y * height + padding, height - padding)\
                .narrow(2, x * width + padding, width - padding)\
            k = k + 1
    return grid

def save_image(tensor, filename, nrow=8, padding=2,
               normalize=False, range=None, scale_each=False, pad_value=0):
    """Save a given Tensor into an image file.

        tensor (Tensor or list): Image to be saved. If given a mini-batch tensor,
            saves the tensor as a grid of images by calling ``make_grid``.
        **kwargs: Other arguments are documented in ``make_grid``.
    from PIL import Image
    grid = make_grid(tensor, nrow=nrow, padding=padding, pad_value=pad_value,
                     normalize=normalize, range=range, scale_each=scale_each)
    # Add 0.5 after unnormalizing to [0, 255] to round to nearest integer
    ndarr = grid.mul_(255).add_(0.5).clamp_(0, 255).permute(1, 2, 0).to('cpu', torch.uint8).numpy()
    im = Image.fromarray(ndarr)


Try using normalize=True to avoid that.

Thanks. I know how to avoid this, but I think this function should indicate more clearly about this in-place input modifcaiton.

It is a known issue and was fixed here.
You should get the fix, if you install the nightly binaries or from source.

1 Like