I have been working on road lane detection using LaneNet by Tensorflow 2.x for the past two months. In the beginning, I chose to build a road lane detection program by using the Canny Edge method utilizing the contrast feature of images as the base of the detection method. The result was not quite good. It was not adaptable to various road conditions, such as curvature, terrain, brightness, etc. In order to improve it, I decided to do further research about methods that are more robust. Later, I found that using Image Segmentation could make the system more adaptable to road conditions. One of the research papers that I found was “Towards End-to-End Lane Detection: an Instance Segmentation Approach” by Davy Neven et. al.
GGPlot Multiple Label for Swimmer Lane
Background Story: One day, my manager asked me if I could have labels for Censor, Response, and Dose groups on one graph. The trick…