This project was to build an end to end algorithm that takes and image of a human or dog face, determines which it is, then gives a prediction for the most likely dog breed. The model for the dog breed classification was able to obtain an accuracy rate of 93.6% on 133 breeds.
An overview of the project is described in this blog post.
The actual project submission can be viewed here.
A more organized format with in depth explainations and extensions beyond the original project can be viewed here.