Self-Driving Car

Overview

Following sections show videos for applications of computer vision and machine learning for self-driving cars. Links to code can be find within the corresonponding sections if available.

1 Computer Vision
- 1.1 Finding Lane Lines
2 Machine Learning
More added in the next weeks.

Computer Vision

Finding Lane Lines

Simple python code for detecting lane lines of a road can be found here: Simple Find Lane Lines

The Pipeline contains following steps:

1. Conversion to grayscale image
2. Masking (Reducing processed image to region of interest)
3. Removing noise using gaussian blur
4. Canny edge detection
5. Overpaint edges introduced by masking
6. Hough line detection
7. Process lines ( removing non lane lines )
8. Overlay the line image onto the original road image

Video for finding lane lines on a motorway:

Possible improvements:

Add Equalizing of the gray-scale image as step 2b. After Removing noise using the gaussian blur a threshold via (cv2.adaptiveThreshold) might make the lane detection more robust against brightness changes (daylight variations, shadows, etc.).

The resulting pipeline would be as follows:

1a. Using different color space HSV (Hue, Saturation, Value)
1b. Conversion to grayscale image
2a. Masking (Reducing processed image to region of interest)
2b. Equalized the grayscaled image (e.g. cv2.equalizeHist)
3a. Removing noise using gaussian blur
3b. Threshold (e.g. cv2.adaptiveThreshold)
4. Skeletonize (skimage.morphology.skeletonize)
5. Overpaint edges introduced by masking
6. Hough line detection
7. Process lines ( removing non lane lines )
8. Overlay the line image onto the original road image

Machine Learning

Vehicle Detection with machine Learning

A linear support vector machine is used as a classifier. The feature extraction is carried out by performing a Histogram of oriented Gradients on a labeled training set of images. The classifier is then trained with the labeled HOG feature vectors.

Short Video:

Long Video: