All projects
Perception

Road Lane Detection

Classical vs. CNN lane detection for autonomous guidance, validated on highway footage.

Overview

My bachelor's thesis compared classical lane-detection pipelines (ROI masking, edge detection, Hough transform and lane fitting) with CNN-based approaches for autonomous guidance. I tested them on highway footage in good visibility, at night and on worn roads with changing light.

The same work powered the University of Seville team's lane detector in the SEAT × CUPRA × Volkswagen Autonomous Driving Challenge, where I led that module.

Highlights

  • 01Lateral offset estimation from the detected lanes.
  • 02Robustness tests at night and under changing light.
  • 03Classical vs. learned methods compared side by side.

Command menu

Jump to a section or project