Dogs and cats classification

Sample project with a goal to build and train CNN model and export it to tensorflow lite model to use with the sample android app. It contains:

• jupyter/ai_model.ipynb - dataset preparation, model optimization and training
• jupyter/prepare_data.ipynb - not really needed, just to see train/validation images
• DogsAndCatsAndroid - android sample app. It is missing *.tflite model in data/src/main/ml/ directory. See notes below

ezgif.com-gif-maker.mov

Important notes:

• Due to size constraints dataset, *.h5 and *.tflite models not provided in this repository
• There was no motivation to produce nice UI experience. It should just work
• CNN model was built and trained using tensorflow framework. Validation accuracy reached around 88%

CNN model architecture. Some parameter values were tuned using keras tuner.