Publication

Abstract : Autonomous system that takes live feed from camera and processes images to detect lanes and collision possibilities such as people/animals coming in front., objects flying towards the vehicle etc. Direction and speed of objects relative to motion of vehicle is the main process being applied. Confusion matrix is used as the base for analysis of the model. First., the images of the road are taken and then lane lines are identified. Emotions from the face maybe used to judge the mentality of the pedestrians. Shape and movement of objects are used as a base classification for rigid and non-rigid objects. Activities on the sidewalk are identified to judge the randomness of motion of objects and its impact to the lane detection system. In autonomous driving., lane detection and collision avoidance are two primary challenges., especially in non-ideal driving environments such as damaged or missing lanes. This paper proposes a robust real-time lane detection system integrated with an advanced object recognition and anomaly detection model., capable of addressing lane ambiguity., recognizing potential collision threats., and understanding surrounding activities. The system leverages a combination of advanced image processing techniques., object detection models like YOLOv5 (You Only Look Once)., and predictive algorithms for decision-making. This algorithm demonstrates superior efficiency., achieving better accuracy and minimizing errors compared to existing technologies.