TOWARDS A LOW-COST VISION SYSTEM FOR REAL-TIME PAVEMENT CONDITION ASSESSMENT

Abstract

Pavement condition assessment is typically performed through manual inspections and specialized hardware and software. Although advances in camera and sensing technology in the last decade helped propel the automation of pavement distress detection and characterization, increased equipment acquisition and running costs limit access to the most effective solutions. Furthermore, some of these advanced techniques require substantial human involvement to process and analyze data correctly. This thesis proposes a cost-effective, end-to-end automated approach to pavement condition assessment that employs a neural object detector to identify and measure instances of pavement distress in real time from oblique 2D imagery acquired using a UAV. To promote ease of implementation and scale, the associated modeling process is simplified by using Google Street View data as a proxy for data collected via a UAV-mounted camera. A state-of-the-art object detector architecture is applied to identify and localize pavement distress instances in monocular images. Camera data, information about Street View image acquisition conditions, and the principles of photogrammetry and planar homography are combined to construct a mapping for translating pixel distances to real-world distances. This capability is then integrated into the neural network inference process to derive an end-to-end system for real-time distress identification and measurement.