- Customer
- KlaraView
- Industry
- PropTech
- Location
- Malmö, Sweden
- Platform
- Drones, Computer Vision
Customer
KlaraView is a Swedish PropTech company building a platform for data-driven facade management. By combining drone-based inspections with structured facade data, the platform enables inspection information to remain accessible, comparable, and valuable over time.
Based in Malmö, KlaraView combines AI, software development, and drone technology in a platform that transforms facade inspection data into structured, long-term decision support. It enables property owners and managers to track facade conditions over time and make better-informed decisions about maintenance, investments, and long-term planning.
Challenge
Developing accurate computer vision models for facade inspection requires large, diverse, and representative datasets. For KlaraView, relying solely on real-world drone imagery made it difficult to obtain sufficient training data, particularly for facade conditions that are rare or underrepresented.
Limited data diversity can restrict how well models generalize across different buildings, materials, environments, and visual conditions. KlaraView needed a scalable way to introduce greater variation and control into its datasets while improving model accuracy and making validation more efficient.
Solution
syntheticAIdata developed custom synthetic image datasets designed to reflect the visual characteristics of KlaraView's real-world drone inspection imagery. Photorealistic virtual building replicas were created to simulate a wide range of facade conditions in a controlled environment.
The datasets incorporate variations in facade materials, textures, lighting, weather, camera angles, distances, and surrounding environments. Each image is delivered with bounding box annotations, providing ready-to-use data for computer vision training and validation. Through an ongoing, iterative collaboration, the datasets are continuously refined based on KlaraView's feedback, model performance, and evolving data requirements.
Result
By integrating synthetic data into its computer vision development workflow, KlaraView has improved model accuracy while accelerating training and validation. Access to diverse, purpose-built datasets has also reduced reliance on collecting and manually annotating large volumes of real-world drone imagery, enabling the team to develop and iterate more efficiently.
The ongoing collaboration allows new and refined synthetic datasets to support KlaraView's evolving requirements, helping the team continuously strengthen its computer vision models and advance the capabilities of its AI-assisted facade intelligence platform.