main process: 1.main.first_step_training.py 2.main.second_step_training.py

bcnn model: 1.bcnn_model_architecture.bcnn_last_layer.py 2.bcnn_model_architecture.bcnn_last_layer.py

utils: 1.utils.create_test_h5_file.py 2.utils.create_train_val_h5_file.py

Step 1: after download images, use utils.create_test_h5_file.py and utils.create_train_val_h5_file.py to create .h5 file.

Step 2: run main.first_step_training.py and save the weights.

Step 3: run main.second_step_training.py to load the weights from step 2

Step 4: output the prediction and upload to Kaggle to evaluate