Lane detection using a fusion of two different CNN architectures

Lane detection is essential in many applications including autonomous navigation and intelligent vehicles. Recently, the performance of image recognition and detection has been remarkably improved by Convolutional Neural Networks (CNN). In this paper, we present a method for lane detection by combining the results of two CNN architectures. The first CNN detects lane locations on the image via a sliding window, while the second one detects the vanishing point and the lane angle. By combining the results of these two structures, we present a method to improve lane detection results by comparing the lane detection result from each structure.