As Australia’s leading telecommunications provider, Telstra sought to automate previously high-risk, manual assessments of its widespread wireless network to improve safety and optimize management of its cellular towers. The company retained SiteSee to generate condition and as-built assessment reports by using machine learning and object recognition technology on reality meshes. Given the slim, vertical nature of the towers, SiteSee needed a comprehensive reality modeling application to accurately capture the infrastructure and extract point clouds for hybrid processing to implement and refine its artificial intelligence (AI) processes for eventual roll-out across Telstra’s entire portfolio of 8,000 towers.
SiteSee used unmanned aerial vehicles (UAVs) to capture the images and cloud-based processing to manage all the the data, selecting ContextCapture as the photogrammetric application to generate the highly accurate reality meshes. Using ContextCapture provided the technological foundation for premium, web-accessible 3D models necessary for SiteSee to manipulate and render the models intelligently through its proprietary AI 3D analysis. The integrated reality modeling plus AI solution delivered an accurate record of the site and tower equipment for immediate use by Telstra engineers for inventory and auditing purposes.
Using ContextCapture to model the communication towers helped SiteSee optimize its AI process and achieve over 98% recognition accuracy of all installed equipment with a dimensional accuracy of +/- 10 millimeters for structures over 50 meters. Compared to standard inspections, using reality modeling and AI provided Telstra with access to transparent, accurate, as-built information to make better informed decisions at 25% less cost, while reducing site visits and tower climbs by 60%. SiteSee’s web-accessible 3D reality mesh and object recognition technology reduced travel and design time for wireless network engineering teams by a minimum of eight hours.
SiteSee evaluated numerous photogrammetric processing applications and determined that ContextCapture was the best application available to deliver very high-resolution 3D reality meshes, which were necessary for the foundation for their AI telecommunications inspection solution. Bentley’s reality modeling software efficiently processed point clouds into fully rendered, geometrically accurate 3D models that were digital twins of the actual towers. The aerotriangulation capability in ContextCapture optimized image processing of the slim, vertical towers by de-emphasizing ground control and focusing more on the tower equipment for point cloud registration.
Project Playbook: ContextCapture