I am designing a CNN that will take 2000 X 2000 input images. They are grey images where the CNN need to detect edges, scratches. I would like to know whether the pre-trained networks for coloured images will work or have to design a CNN from scratch ?
there is already a thread about that topic:
I received the better results by using a DenseNet from scratch.
hello… can i get ur contact? email? im beginner in cnn… i wanna ask something to you if you dont mind