dognotdog: Developed by Roshan Noronha*

*With help from friends and the internet

The goal of this project is to train a convolutional neural network (CNN) to accurately determine if an object is a dog or not. Google Colab was used to run and train this model via a Jupyter notebook. This project will be done in 3 stages.

1) Implementing a basic CNN from scratch that can recognize dogs. The dataset is fairly large so this should not be a limitation. The code for this step is in the "dognotdog-basic" folder.

2) Using the code implemented in step 1 as a starting point, use transfer learning to increase the accuracy of the dognotdog CNN. The code for this step is in the "dognotdog-transferlearning" folder.

3) Using R-shiny, create a webapp that uses the model trained in step 2). This webapp will let the user upload an image and then classify it as a "dog" or "not dog". The code for this step is in the "dognotdog-shinyapp" folder.