Ayush12062000/Road_lane_Detection

Task is to identify Road either by Image processing technique or deep learning. There are multiple ways we can detect roads. We can use the learning-based approaches, such as training a deep learning model on an annotated video dataset or use a pre-trained model. However, I used simpler methods to detect road with the popular OpenCV library in Python. As we can observe there are several different things apart from the road in the frames/images. So, before processing those images I used the Frame mask technique to ignore unwanted objects, and then we are just left with roads in the image. After this, applied image thresholding for changing the pixel values in the image if they are greater then the threshold value and Hough line Transformation to detect lane but unfortunately in the images lanes drawn on the road were not that clear so, as I thought if I'll be able to detect lanes I can easily take their reference and mark the road with a polygon but this approach failed. Then I simply used Frame mask and image thresholding technique for further processing. I have processed the video in three parts as I wasn't able to process the whole video at once due to OUTOFMEMORYERROR (the system wasn't able to handle 3336 images at once) then after the images/frames have been processed and in each frame/image road is marked. Finally, I combined the frames into a video. Since I processed in three parts, I got three output videos. Then to get a final output video of the given input I used Moviepy library to merge these three videos into one.