How does the convolution work for an input which has multiple channels convolving with filters that also have depth.

For example, assume somewhere in the network we have a 5x5x3being an input to a convolution layer. The layer has 2 filters which have size 3x3, the depth of the filters would therefore be 3 because of the input, so each filter is actually 3x3x3. Now how does the convolution work in this case.

My guess is that a convolution happens with each specific channel. E.g. Filter 1 channel 1 convolved with input channel 1, Filter 1 channel 2 convolved with input channel 2, and Filter 1 channel 3 convolved with input channel 3. Is this what happens?

I have read tutorials but they mostly show convolution with 1 channel only. If you could link any tutorials which explains this it would quite helpful.

  • $\begingroup$ your guess is correct $\endgroup$ – shimao Jun 29 '19 at 16:15

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.