This is a project created in Python using Tensorflow and Keras that can identify the breed of a dog within an image.
It uses a Convolutional Neural Network and predicts the dog with an 89% accuracy rating.
Originally I wanted this to work on 200 breeds of dog, and I definatley think I have the ability to scale this project up so it works with that many dogs, but as of right now this only works on 5 different breeds of dog.
This is due to constraints with the training time and hardware. Convolutional Neural Networks take a very long time to train and to create a network with that many outputs I was looking at 18 hours of waiting for this network and I would probably have to deepen the network for it to work so it was just too taxing on my computer to be able to get it working with 200 breeds.
Note, this code is not on github yet as I am (probably overly) paranoid about the university thinking I've stolen the code from my own website, so after the coursework is all handed in and wrapped up, then I intend to upload the code.