Keras Dog-Cat Image Classification

Project Summary

• This project is to provide a detailed explanation of applying convolutional neural network (CNN) for a classification task of classifying dogs vs cats by using the Kaggle dataset.
• There are two CNN models including manually built CNN model (Model 1) and Customised pre-trained CNN model using the VGG16 architecture (Model 2).
• Model 1 has an accuracy of 76.48%, meanwhile, Model 2 has an accuracy of 93.12%.
• Full report is available via this link

Technology

• Python
• Keras

Getting Started

1. Clone this repository (for help see this tutorial).
2. Use Jupyter Notebook Environment then run CNN1.ipynb for Model 1 or CNN2.ipynb for Model 2.
3. Use saved models' files including cnn1.h5 for Model 1 or cnn2.h5 for Model 2 to use the models directly and to save training time which could take several hours especially for Model 2.

License

This project is under the MIT license.