Intro

I went on a trip all over Europe with my friend Gabriel. We took several photos and it was amazing. I highly recommend Euro Disney. Although a bit salty, aka costly, it was a good time. Je vraiment manque tout ca! Et parle francais avec les parisiens est tres bon aussi.

Anyways, months later and several emails from Gabriel demanding the photos!!! I was able to install and upload my images on Google Photos and share with him. Only 3 months later
Right after installing it, I saw that the app was classifying my photos and then I was curious to know how was it doing it. Well, it’s a CNN! (I like Quora btw u_u)

CNN

Convolutional neural networks can be called ConvNet or CNN and its composed of an input layer, one or two hidden layers, and one output layer. Each layer might have its own properties as parameters and hyperparameters.

More information can be found here [2] that also has a very interesting and educative demonstration of how the CNN works.

From what I could find until now [blame me flag], Google photo seems to rely on Inception-v4 [3].

REFs

[1] https://www.quora.com/Which-algorithm-is-used-in-Google-Photos-app-for-classification-labeling

[2] http://cs231n.github.io/convolutional-networks/

[3] https://arxiv.org/abs/1602.07261