Satellite Imaging
for Remote Monitoring

Overview

• Japan faces a critical need for maintaining road markings as part of its aging infrastructure. Collaboration with police in Hyogo prefecture emphasizes disaster prevention and community safety, leveraging AI for cost-effective and efficient maintenance strategies. 

• In this project, we work with Mind Foundry to explore the potential of satellite imagery and AI for detecting zebra crossings and assessing road marking deterioration to streamline infrastructure maintenance. These technologies lay the groundwork for practical applications in road safety and operational efficiency. 
   

Detail

Background

• Aging road infrastructure and workforce constraints globally necessitate innovative technologies for cost-effective maintenance. 

• Advances in satellite data and AI provide unprecedented opportunities for periodic monitoring and assessment of road assets. 

​

Hypothesis

• Combining satellite imagery with AI models can automate detection and analysis of road markings, enabling quicker and more accurate maintenance prioritization. 

• A successful implementation in one region can serve as a scalable blueprint for widespread adoption, optimizing resource use across varying infrastructures. 

 

R&D

• Investigated AI models for detecting and classifying zebra crossings and road markings, achieving initial feasibility in image annotation and modeling processes. 

• Evaluated the time and cost associated with AI model training and satellite data processing to ensure economic scalability. 

​​​

Next step

• Focus on improving AI detection accuracy and robustness while expanding capabilities to assess road marking conditions more comprehensively. 

• Overcome challenges in aligning satellite and on-ground datasets to maximize actionable insights.