Closed wilile26811249 closed 2 years ago
@wilile26811249 Yeah, that's because of the saliency map. At points in which no activation occurred, it's a dark layer. Also, for better visualization, a normalization operation is performed, resulting in a darker image.
Thank you for your explain. I understand.
When I try to specify the img-size to 1280, the output image is darker than the input image.