Real Time Road Lane Detection using Computer Vision Techniques in Python

Autonomous driving car is one of the most disruptive Artificial Intelligence developments. Lane detection System (LDS), which is an early phase in the training of a self-driving car, is one of the numerous steps involved. With the growth in urban traffic, road safety becomes more and more essential. The majority of accidents on the roads are caused by drivers leaving the lane without obeying the correct laws. Lane departure warning system and Adaptive Cruise Control (ACC) are sophisticated driver assist systems scenarios that can help the vehicles analyse their present working atmosphere and give the proper input for defensive driving or sound the alert in perilous circumstances. The difficulty in predicting the road traffic environment, however, is a bottleneck in the development of this system. The primary perceptual signals in a complicated traffic condition are street border and carriageway delineation, in addition to pavement colouring separation and roughness identification. The application of machine learning (ML) in intelligent driving systems (IDS) has made lane identification and computer vision a hot issue. The vehicle’s position and trajectory in respect to the lane are precisely predicted by the lane detection system using data collected from lane markers in a complicated environment. This work focuses on a series of images taken from a real-time input feed. After RGB image is converted to grayscale Gaussian filter is used to perform noise removal. The Canny Edge Detector detects the edges with significant variations in intensity. The canny image is masked except for region of interest and Hough Line Transform (HLT) detects the extremes of the detected lines in the image. Thus, the findings from the Hough transform function is a value of pixels that is superimposed into the original image, as detected lane lines.