Novel lane and object detection technique using visual camera

The project aims to implement a real-time lane, object and pedestrian detection system using visual input froma camera mounted on the front of the vehicle. The project is developed using OpenCV based YOLO libraries and Tensorflow. Various objects that appear in the feed of the camera are identified, numbered and labelled so that the self-driving vehicle can maneuver accordingly. To ensure that proposed system is unique in its approach and performs better than existing methodologies, a novel approach was considered wherein the detection of each parameter is segmented, cascaded and simultaneously executed. This novel approach ensures that detection of each parameter remains independent of the other, and the performance and efficiency is at maximum making the system ideal for autonomous vehicles and semi autonomous vehicles.