TensorFlow face recognition experiment. You can train your model using tensorflow docker images with python and use the trained model with java later.

• Create "tests" folder to test the model.
• Add "good" and "bad" folders with jpg images inside "tests" folder.
• Run: mvn test

• Create "images" folder to train the model.
• Add one folder inside "images" for each category you want to classify with jpg images on it.
• Run docker image: docker run -it -p 8888:8888 -p 6006:6006 -v ${PWD}:/tf_files --workdir /tf_files floydhub/dl-docker:cpu bash
• Now you can train the model typing: python retrain.py --bottleneck_dir=bottlenecks --model_dir=inception --summaries_dir=training_summaries/long --output_graph=model/retrained_graph.pb --output_labels=model/retrained_labels.txt --image_dir=images
• After training the model you can test it with: python label_image.py test.jpg