Project developed for the "Vision and Cognitive Services" course at University of Padova.
The aim of the project is to build a traffic signs recognition system (TSRs) able to recognize different types of traffic signs. The task is similar to the built-in computer of modern cars, in fact those systems are capable of reading traffic signs and to show important information to the driver improving his safety, such as traffic conditions, road rights, forbidden and allowed behaviors. The main focus of this project is, firstly, to analyze the input image to detect potential traffic signs and, secondly, to classify each detected object via a multi-class classifier. In addition to this, we have expanded the functionality of this project to make it works even with video clips, so that the recognition of traffic signs takes place in real time. In order to explore the field of traffic sign recognition we adopt the YOLO object detection framework for detection and feature extraction processes together with a classification system based on Keras.