Project 2: Car Classification

This project was done by myself and fellow classmates Liam Jogel and Freddy Boelter while studying at DIS in Copenhagen, Denmark. This project was the first hands on experience using a neural network, more specifically a CNN, but we had a lot of fun doing it!

This project takes a dataset offered by Stanford that contains thousands of images of car. Our goal of the project was to use these images to build a car classifier that could identify the make, model, and year of the car. This was quite difficult considering there was 196 different classes of the cars in the training set and different models of cars can vary slightly. Ultimately, through an iterative learning process we improved our test accuracy to 40%, which was a success for our team.

You can find more information about this project on my Github page, as well as a medium article I wrote details the steps we took in our approach detailed . Hope you enjoy!