Semantic segmentation problem to detect road/lane pixels on images in KITTI vision benchmark data - View it on GitHub