Fine Tunning VGG16

Script

data_prep.py

• info:
  • caches and preproces dataset to be used in training and testing the model
• arguments:
  • --input_path: path to dataset
  • --output_path: full path(filename includede) to save the preprocessed and cached dataset
  • --dimension: dimension for any given feature sample of the dataset
  • --batch_size: number of feature and label to be preprocessed and cached the same time

fine_tunning.py

• info:
  • loads the preprocessed/cached dataset and train a fine tunned VGG16 model
• arguments:
  • --input_path[required]: path to cached/preprocessed dataset
  • --output_path[optional] : full path(filename includede) to save the train model
  • --epochs[optional]: number of training epochs
  • --train_size[optional] : fraction of dataset to be used to training
  • --learning_rate[optional] : learning rate of optimizer used in training model
  • --batch_size[optional] : number of feature and label samples to train model per each steps

re_train.py

• info:
  • loads the fine tunned model and retrains it
• arguments:
  • --input_path[required]: path to the existing model
  • --output_path[optional] : full path(filename includede) to save the re-trained model
  • --epochs[optional]: number of training epochs
  • --train_size[optional] : fraction of dataset to be used to training
  • --learning_rate[optional] : learning rate of optimizer used in training model
  • --batch_size[optional] : number of feature and label samples to train model per each steps