Lane Detection for ADAS & Self-Driving Cars

Overview

This a real-time lane detection using computer vision techniques and deep learning.

The system takes a single (road) image or a recorded local video from the road and perform the following tasks:

1. Convert the color frame/image to grayscale for easy and quick treatment
2. Reduce noise and smooth with Gaussian Blur filter
3. Identify edges with Canny Functions : 
    - This allows to detect edges in the frame/image. By checking the intensity changes in brightness of adjecent pixels
4. Calculate the Region of interest of the lines 
5. Hough transform
6. Optimizing

Application

The system has several situations and can be used in many situations:

• Lane departure warning systems
• Lane-keeping assistant technology (vehicle longitudinal & lateral control)
• Hand-free highway driving systems (during steering takeover)
• Kodiak Robotics uses lane lines detection for its self-driving truck (KodiakDriver) stack to navigate in the world

Requirements & dependency packages

python=3.7.*
numpy
matplotlib
opencv
opencv-contrib-python

To install all package from the requirements.txt file via conda:

conda install --file requirements.txt

or

pip install -r requirements.txt

Usage

To run the system:

python main.py

Upcoming features & Upgrades

@TODO-List :

- add a logger for debugging and logging results
- add camera config module for real-time applications handling
- add a SegNet or a CNN model interface to increase system accuracy  
- add green marker between two road lanes for urban/highway roads
- add different color between lane departure and inside lane
- Add Unit testing
- Test with camera in a dynamic environment (car or robot)
- CI/CD actions

Contributing

If you want to help this project grow, feel free to submmit a PR.