What is the difference between training a model with RGB images and using only the color channels separately (like only the red channel, green channel, etc.)? Would the model also learn patterns between the different colors in the first case?
If for me the single-channel results are relevant but also the patterns between different channels are relevant, it would be beneficial to use them together?
I am asking this because I want to apply this to signals of an accelerometer that has x, y, z-axis data. And I want to increase the resolution of the data. Will the model learn to combine all features from different axis if I input (1024, 3) length, channels of a one-dimensional signal into my one-dimensional CNN?