Compute Engine

CrossTech Cuts High-Risk Rail Faults 96% with Computer Vision Predictive Maintenance on Google Cloud

UK startup CrossTech built Hubble, an AI computer vision platform that analyzes video captured from trains to detect hazards such as overgrown vegetation, signal obstructions and track ballast issues, running on Google Cloud's Compute Engine, Cloud Run and Vertex AI. Using Vertex AI, CrossTech halved the time to build a first model version to six weeks and cut deployment cycles by 70%. The predictive maintenance has delivered roughly £20 million in annual net maintenance efficiency for customers, reduced unplanned service interruptions by 30%, cut high-risk faults on key lines by 96%, and enabled frontline teams to remediate 6,000 faults.