.npy as input of ResNet

Hy guys, i have a problem, i would to pass many distributions that are storage like .npy as input of a ResNet. It’s a new thing for me the type of input. How can do this? Is it possible or i can transformate my input first? The code that i wrote is ready for images…

You can pass whatever you want, for reading numpy files you will have to replace the reading function to adapt it to npy files.
What’s the exact question about?

1 Like

The problem in another, i have the .npy as images of 9 channels.
I can’t change them
How can i change the resnet50 model pretrained?

When you load the model model.load(state_dict) use the arg strict=False

Anyway you wiol have to fije tune

1 Like