Is using MobileNet always better than normal conv networks?
Professor said mobilenet reduces computation and make the model has richer functions. So i thought that if we use depthwise separable convolution to the neural network we studied before, the precision of the models increase??
Thank you.