Image_manipulation_detection
Image_manipulation_detection copied to clipboard
How to pre-train noise stream's vgg weights on ImageNet?
hello, can you talk about the pre-train noise stream's vgg weights on ImageNet more details? thanks. for accurate prediction, please pre-train noise stream's vgg weights on ImageNet and overwrite the trainable setting of noise stream after SRM conv layer
Yes,i encounteered the problem too. In Image_manipulation_detection-master/data/imagenet_weights/ the file vgg16.ckpt how to pre-train to generate my own model parameters?Thanks a lot!
I met the same question
Did anyone solve it?
"for accurate prediction, please pre-train noise stream's vgg weights on ImageNet and overwrite..." read it over and over but absolutely no clue what that means. What ImageNet? Overwrite how?