Training a Convolutional Neural Network for Appearance-Invariant Place Recognition | Read Paper on Bytez