General Tensorflow Image Classifier

Easily train to classify anything - I classified me, from wirey youth to beardy data scientist.

Requirements

• docker

Usage

A "classifier" directory with a directory "data" inside it with all the images. For example

 [any_path]/the_classifier/
 [any_path]/the_classifier/data
 [any_path]/the_classifier/data/ant_newman

Here you put your images, so if we are making an Ant Newman classifier, it would be photoraphs of me throughout the bearded and non-bearded ages.

This directory has the samples and the trained classifier after execution of "train.sh".

Train process

In your docker environment:

 ./train.sh [any_path]/the_classifier

Guess process

For a single guess:

 ./guess.sh [any_path]/the_classifier /yourfile.jpg

To guess an entire directory

./guessDir.sh [any_path]/classifier [any_path]/srcDir [any_path]/destDir

Requires an absolute file path for classifier and images because the script dos not support relative path (volume mounting)

Credits

Thanks to Xblaster