#launch docker Set working directory of docker and launch it. Replace ${HOME}/tf_files with path to this repository.
docker run -it --publish 6006:6006 --volume ${HOME}/tf_files:/tf_files --workdir /tf_files gcr.io/tensorflow/tensorflow:latest-devel bash
#retrain inception with your own classes (now you are inside the docker) Here 'cars' is the folder which contains image folders(classes) like beetle,audiq7,scorpio etc
python retrain.py \
--bottleneck_dir=bottlenecks \
--how_many_training_steps=500 \
--model_dir=inception \
--summaries_dir=training_summaries/basic \
--output_graph=retrained_graph.pb \
--output_labels=retrained_labels.txt \
--image_dir=cars
#Evaluation Giving an input image to program for prediction.
python evaluation.py <image-name>.jpg