While the basic lane-finding project only considers lane lines to be straight lines (a very simplistic approach), here an effort is made to model and measure lane curvature. Furthermore, more advanced image processing operations are applied to improve detection of image pixels which may be part of lane lines.
Important sources are in /src. For a description of the project, a writeup describing the steps involved with a number of
supplementary images, see the Jupyter Notebook advanded_lane_finding.ipynb. This is the primary entrypoint to the project,
all else is supporting material.
This experiment was run on a MacBook Pro 11'3, with Ubuntu 16.10, running Python 3.5.2 and Anaconda 4.2.0. The sources depend on OpenCV, numpy, scipy, pandas, and matplotlib.