Prologue

Welcome to DeepTrafficSigns: a peek into real-time road sign assistance.

In the 1st quarter of 2018, I participated in an intensive data science program in Berlin: Data Science Retreat. In the latter stages, we were tasked with developing a portfolio project to showcase some of the skills we had developed over the preceding months. My interest in computer vision and deep learning pushed me towards object detection with neural networks, and road sign assistance held out the promise of an interesting and topical use case. From the standpoint of autonomous transport, environment comprehension is of utmost importance and, given the time constraints, road sign detection and recognition seemed feasible in some form. Ultimately, I was hoping to gain some insight into the topic in general, and to pick up some cool technical skills along the way.

If this piques you’re interest, please don’t hesitate to read further. The following pages chart the development of the project from inception all the way through to public presentation. (Yikes!)

Timeline

1. Introduction
2. Environment
3. Data