# Generating All the Roads to Rome: Road Layout Randomization for Improved Road Marking Segmentation

## TL;DR

This paper introduces a domain randomization technique for generating synthetic road marking images to improve segmentation accuracy in real-world driving scenarios, reducing labeling effort and enhancing model robustness.

## Contribution

It presents a novel method for automatic generation of diverse training data through road layout randomization, enhancing segmentation performance without manual labeling.

## Key findings

- Improved mIoU of over 12 percentage points for rare road marking classes in real-world tests.
- Synthetic data generation maintains performance on common classes while boosting rare class segmentation.
- Framework scalable to various domains and conditions for large-scale dataset creation.

## Abstract

Road markings provide guidance to traffic participants and enforce safe driving behaviour, understanding their semantic meaning is therefore paramount in (automated) driving. However, producing the vast quantities of road marking labels required for training state-of-the-art deep networks is costly, time-consuming, and simply infeasible for every domain and condition. In addition, training data retrieved from virtual worlds often lack the richness and complexity of the real world and consequently cannot be used directly. In this paper, we provide an alternative approach in which new road marking training pairs are automatically generated. To this end, we apply principles of domain randomization to the road layout and synthesize new images from altered semantic labels. We demonstrate that training on these synthetic pairs improves mIoU of the segmentation of rare road marking classes during real-world deployment in complex urban environments by more than 12 percentage points, while performance for other classes is retained. This framework can easily be scaled to all domains and conditions to generate large-scale road marking datasets, while avoiding manual labelling effort.