This is an example on how to use pre-existing TensorFlow models to retrain on your data. In this example, we use Inception which is an image classifier. You can train it on the existing images of different noodles type and then use images in the test directory to get the predictions.

• Install TensorFlow
• Put directories with images in categories directory. The name of the directories will be the classifications

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=categories

python label_image.py <insert_file_path_to_predict_here>

See the model and progress in TensorBoard

tensorboard --logdir training_summaries

• Here is a presentation of this sample in action. Demo starts at 42:00 - (link)[https://www.youtube.com/watch?v=xlxSVWxeZdM]

• Albert Padin
• Based on Image Retraining Example