CODEWITHSUNDEEP
deep-learningvgg-net
User2010
User2010

Reputation: 11

Why does VGG trained from scratch throw an error when input channes are changed to more than three?

Using this guide on how to implemenent VGG19 from scratch, I changed the input shape but gives error. How do I modify this layer to accept images with more than 3 channels?

Changed from:

input_shape = (224,224,3)

to:

input_shape = (32, 32, 6)

error:

ValueError: Input 0 of layer "vgg19" is incompatible with the layer: expected shape=(None, 32, 32, 6), found shape=(None, 32, 32, 3)

Upvotes: 0

Views: 39

Answers (1)

Buoy Rina
Buoy Rina

Reputation: 568

You need to update the first convolutional layer to accept an input channel of 6 in this case.

Upvotes: 0

Related Questions