advanced_lane_detection

Why we included this project

This is a compact, self-contained take on the classic computer-vision lane detection pipeline, built for the Udacity Self-Driving Car Nanodegree. It runs the whole sequence a perception engineer would recognize: calibrating the camera from chessboard images, correcting distortion, building a thresholded binary image from color and gradient transforms, warping to a bird's-eye view, fitting polynomials to the lane pixels, and estimating curvature plus the car's offset from center. That makes it a solid reference for anyone learning or teaching classical lane finding without deep learning, and a readable baseline to hold up against modern neural approaches. The code is short and clearly commented, so you can trace each step and point it at your own video by editing a few lines. It is not a production self-driving system, but as a study resource and a starting point for classical vision work it is hard to beat.