Crack-pot: Autonomous Road Crack and Pothole Detection

TL;DR

This paper presents a real-time, GPU-compatible deep learning system for autonomous detection of road cracks and potholes, utilizing texture features to improve robustness under various challenging conditions.

Contribution

It introduces a novel texture-based feature approach within a deep neural network for accurate, real-time road impairment detection suitable for autonomous vehicles.

Findings

Effective detection under large viewpoint changes

Robust performance amidst shadows and occlusion

Compatible with standard GPU processing boards

Abstract

With the advent of self-driving cars and autonomous robots, it is imperative to detect road impairments like cracks and potholes and to perform necessary evading maneuvers to ensure fluid journey for on-board passengers or equipment. We propose a fully autonomous robust real-time road crack and pothole detection algorithm which can be deployed on any GPU based conventional processing boards with an associated camera. The approach is based on a deep neural net architecture which detects cracks and potholes using texture and spatial features. We also propose pre-processing methods which ensure real-time performance. The novelty of the approach lies in using texture- based features to differentiate between crack surfaces and sound roads. The approach performs well in large viewpoint changes, background noise, shadows, and occlusion. The efficacy of the system is shown on standard road crack datasets.