Hi, I discover that when I use plt.savefig (cmap=“gray”)and torchvision.save_image to save a identical element but different dtype (ndarray and torch.Tensor) in gray scale, but the result in brightness is very different, how can I tune or is there any different algorithms behind two functions? Thanks
Below is my output feature map(above right one) and below( I know it looks too dark, but it indeed has the pattern when you look zoom in)
