Highway Asset Inspection Using Panoramic Imagery

Overview

Highway assets are widely distributed, and the existing asset inventory mainly relies on manpower, which is time-consuming and labor-intensive, and cannot realize high-frequency inspection. To address the above problems, this study develops automatic inspection methods for different highway asset types based on multi-source data (vehicle-mounted panoramic images, inclined photogrammetry and LIDAR point clouds) and constructs an asset sample database, which realizes automatic inspection of highway assets and rapid updating of information.

My Role

I am responsible for cleaning the collected panoramic images and then labeling the highway traffic signage in the images using a semi-automatic approach.

Research Content

1. A semi-automatic annotation of highway traffic signage was conducted, resulting in a panoramic image dataset
2. A detection model for highway traffic signage was developed.