Road Surface Translation Under Snow-covered and Semantic Segmentation for Snow Hazard Index

TL;DR

This paper presents a deep learning system that automatically assesses snow hazard levels on roads using live images, combining image translation and semantic segmentation for real-time snow hazard index calculation.

Contribution

It introduces a novel combination of GAN-based road surface translation and semantic segmentation to automatically compute snow hazard indices from live images.

Findings

Effective snow hazard index calculation demonstrated on 1,155 images.

Robustness of the method in real-world cold region conditions.

Potential for real-time road safety alerts based on snow coverage.

Abstract

In 2020, there was a record heavy snowfall owing to climate change. In reality, 2,000 vehicles were stuck on the highway for three days. Because of the freezing of the road surface, 10 vehicles had a billiard accident. Road managers are required to provide indicators to alert drivers regarding snow cover at hazardous locations. This study proposes a deep learning application with live image post-processing to automatically calculate a snow hazard ratio indicator. First, the road surface hidden under snow is translated using a generative adversarial network, pix2pix. Second, snow-covered and road surface classes are detected by semantic segmentation using DeepLabv3+ with MobileNet as a backbone. Based on these trained networks, we automatically compute the road to snow rate hazard index, indicating the amount of snow covered on the road surface. We demonstrate the applied results to 1,155 live snow images of the cold region in Japan. We mention the usefulness and the practical robustness of our study.