'Convolutional Neural Network : Convolutinal layer
I have some question about the process in the convolutional layer. If I have a 48*48*3 image and I do the 2d convolution with kernel size to be 5*5 without padding. Suppose that I want to get a result of 44*44*6, how many kernels do I need？ What's the depth of each kernel? How many weights do I have to initialize? As far as I understand, I need to use 6 kernels with the size of 5*5*3 and that is to say I need to initialize 5*5*3*6=450 weights. Am I right? Please help me.
Sources
This article follows the attribution requirements of Stack Overflow and is licensed under CC BY-SA 3.0.
Source: Stack Overflow
| Solution | Source |
|---|