I\'ve trained 1000 images, validate 500 and tested 500. The Convnet is made of 13 layers and once executed I\'ve obtained the relative weights.
So far so good except