How UpSampling2D layer works in keras with a multichannel input? For example if the input has 3 channels does this layer upsamples each channel independently? Also, does its computational complexity depends on the number of input channels? Continue reading...
