shahjui2000/drive-able-road-detection

This code segments out the drive-able portion of the road from the surrounding.

Drivable-Road-Detection

This code segments out the drivable portion of the road from the surrounding with the help of inbuilt open-CV library.

Installations required:

1.Numpy library

2.OpenCV library(cv2)

3.Matplotlib library

4.Python 3.6

Now road-lane detection techniques exists but they do not work for Indian-roads due to the fading of lanes and unsmooth road surfaces. So, this project directly takes use of the pixel-values of the road and applies different opencv functions to segment out the drivable portion of the road.

Now, we can deal with pixel values in 2 different ways- OpenCV library for image processing or using Neural Networks.

Output images of particular frames of 2 videos are here below:

Frame 1:

Frame 2:

Since, the output shows great acuracy , using openCV for segmentation becomes a better way since, Neural Networks come with a lot of computations and costs.

drive-able-road-detection's Issues

Night mode

Since this code deals with pixel-values, night mode creates a problem since the sky and other objects all appear black and close to the road-pixel value. If the night-mode can be converted to daylight mode, issue will be solved or else, a different method for night mode needed.