Dimension Error when feeding 3d tensor into CNN and keras network

Question

My Xtrain is a RGB image of dimension (numofsamples,height,width,channel). Where numofsamples = 1047, height = 128, width = 128, channels = 3.

This is done by running the