KlubJagiellonski/pola-ai

Install Tensorflow

Based on: https://www.tensorflow.org/install/pip

$ cd data
$ sudo pip3 install -U virtualenv
$ virtualenv --system-site-packages -p python3 ./venv
$ source ./venv/bin/activate
$ pip install --upgrade pip
$ pip install --upgrade tensorflow==1.12.0
$ pip install --upgrade tensorflow-hub==0.1.1
$ pip install --upgrade requests
$ pip install --upgrade pillow
$ pip install --upgrade opencv-python

$ deactivate # run when finished

Download pictures from Pola

$ cd data
$ python get_ai_pics.py $SHARED_SECRET Pola_ai

Retrain MobileNet v2 model

$ rm -rf model bottleneck pola_retrained.pb pola_retrained_labels.txt pola_retrained.tflite

# https://raw.githubusercontent.com/tensorflow/hub/master/examples/image_retraining/retrain.py
python retrain.py \
  --bottleneck_dir=bottleneck \
  --how_many_training_steps=20000 \
  --model_dir=model \
  --output_graph=pola_retrained.pb \
  --output_labels=pola_retrained_labels.txt \
  --tfhub_module=https://tfhub.dev/google/imagenet/mobilenet_v2_100_224/feature_vector/2 \
  --image_dir=Pola_ai \
  --validation_percentage=15 \
  --misclassified_test_images_output_file=misclassified_images.txt```

# print urls to misclassified images
$ python show_misclassified_urls.py

# test retrained model  
$ python label_image.py \
  --graph=pola_retrained.pb \
  --labels=pola_retrained_labels.txt \
  --image=ludwik.jpg

Convert model to tflite

toco \
  --graph_def_file=pola_retrained.pb \
  --input_format=TENSORFLOW_GRAPHDEF \
  --output_format=TFLITE \
  --output_file=pola_retrained.tflite \
  --inference_type=FLOAT \
  --input_type=FLOAT \
  --input_arrays=Placeholder \
  --output_arrays=final_result \
  --input_shapes=1,224,224,3

$ python lite_label_image.py \
  -m pola_retrained.tflite \
  -l pola_retrained_labels.txt \
  -i ludwik.jpg

Copy retrained model and labels to iOS and Android app

$ cd data
$ cp pola_retrained.tflite pola_retrained_labels.txt ../ios_camera/data
$ cp pola_retrained.tflite pola_retrained_labels.txt ../android_camera/app/src/main/assets