Inspecting Logos in Military Aircrafts: Classification Tools

This project dealt with generating logos as inputs to a classification neural network model. The goal was to shed light on more difficult tasks that neural networks may face such as different angles, fonts, and shape sizes of logos. Overall results are promising, though this project required making faulty logos by hand to improve the model. We hope that generative algorithms get better to create vast amount of logos to further train the model. Source