Inspectiondrone is an Italian company at the forefront of infrastructure inspection and Digital Twin technology. With a network of over 1,500 ENAC-certified drone pilots and deep experience in structural monitoring, the company delivers inspection services for bridges, viaducts, and buildings across the country. What sets them apart is their ability to combine high-quality drone data with civil engineering expertise, turning complex site conditions into actionable insights.
To further scale their operations and bring even more automation and consistency to the inspection process, Inspectiondrone partnered with Focoos AI. The collaboration was born from a shared vision: enabling technical teams to build and iterate on computer vision models quickly and independently, while keeping full control over the entire development workflow.

Challenge and Opportunities

Although Inspectiondrone already had an internal AI infrastructure running on Google Cloud Platform and some models in place, they needed a more integrated and user-friendly solution. Their goal was to streamline the training and deployment of models, simplify experimentation, and reduce the dependency on fragmented tools or opaque systems. The need was for a solution that would allow them to manage everything through a single, intuitive interface, accessible even to non-ML specialists.

The collaboration

As one of the early adopters of the Focoos AI platform, Inspectiondrone played a key role in shaping its evolution. Their real-world needs helped validate product decisions and inspired features specifically designed to support industrial-grade AI development, where data is complex, and precision matters.
Through the platform, the Inspectiondrone team developed a new computer vision model for the automatic detection of critical defects in public infrastructure. The model was trained on their proprietary datasets, carefully annotated by in-house civil engineers. Their domain knowledge proved essential, not just for the quality of annotations, but for guiding model behavior toward practical, interpretable results in high-stakes environments.
The Focoos platform provided efficient ready-to-use models and a structured, fast experimentation environment, with built-in tools for tracking metrics, adjusting training parameters, and applying data augmentation strategies to enhance robustness. The transparency and modularity of the system also enabled them to maintain full ownership of the process, adapting and improving the model over time without relying on external consultants or locked-in solutions.

Results

Today, the model is fully integrated into InspectionDrone’s operational workflows, enabling faster and more consistent inspections directly from drone imagery. This has significantly reduced turnaround times and improved the accuracy and reliability of defect detection in the field. Thanks to the high performance of Focoos’ models and platform, combined with InspectionDrone’s strong technical expertise, the company is now fully equipped to scale and evolve its AI capabilities independently, with complete control over the entire process.