End to End Video Segmentation for Driving : Lane Detection For Autonomous Car

TL;DR

This paper presents a real-time lane detection system for autonomous vehicles using a Global Convolution Networks model, color segmentation, and edge server training to improve safety and reduce accidents.

Contribution

It introduces a framework for real-time lane detection in autonomous cars utilizing GCN, color segmentation, and edge server training for practical deployment.

Findings

Achieved state-of-the-art performance with boundary refinement and Adam optimization.

Developed a real-time video transfer system for edge training.

Demonstrated effectiveness of the GCN model in lane segmentation.

Abstract

Safety and decline of road traffic accidents remain important issues of autonomous driving. Statistics show that unintended lane departure is a leading cause of worldwide motor vehicle collisions, making lane detection the most promising and challenge task for self-driving. Today, numerous groups are combining deep learning techniques with computer vision problems to solve self-driving problems. In this paper, a Global Convolution Networks (GCN) model is used to address both classification and localization issues for semantic segmentation of lane. We are using color-based segmentation is presented and the usability of the model is evaluated. A residual-based boundary refinement and Adam optimization is also used to achieve state-of-art performance. As normal cars could not afford GPUs on the car, and training session for a particular road could be shared by several cars. We propose a framework to get it work in real world. We build a real time video transfer system to get video from the car, get the model trained in edge server (which is equipped with GPUs), and send the trained model back to the car.