Tensorflow Image Classifier

This repo contains the code and data for image classification tutorial in tensorflow. It is a fork of this repo and is based on this CodeLab by Google as well as this tutorial is quite helpful. For a quick video walkthrough of the process check out this information-packed YouTube video.

Requirements

• docker

Usage

Training data is availabe in tf_files/data directory. There are 2 sub-directories for 2 categories pre-populated. If you need additional categories, create additional sub-directories, like, for example

 tf_files/data/iron_man
 tf_files/data/wonder_woman

and then put your images in them.

These will be used for training. After training the tf_files directory will have the classifier.

Training process

Just type

 bash ./train.sh $PWD/tf_files

And it will do everything for you !

Testing the classifier

Just type for a single guess

 bash ./guess.sh $PWD/tf_files $PWD/test_data/images00.jpg 

To guess an entire directory

bash ./guessDir.sh $PWD/tf_files $PWD/test_data $PWD/classified

Ps. Make sure the directory $PWD/classified exists. It can be empty.

Example of result

$ bash guess.sh $PWD/tf_files $PWD/test_data/images00.jpg 
elsa (score = 0.99636)
darth vader (score = 0.00364)

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

The Challenge

Make your own classifier for scientists, then post a clone of this repo with your retrained model in it. (you can name it retrained_graph.pb and it will be around 80 MB. If it's too big for GitHub, upload it to DropBox and post the link to it in your README)

Credits

Credit goes to Xblaster & @sirajology for the majority of this code. I've merely customized it for the training data.