We have written about this company before – and we are doing so again! Here is another success story about Synclair Vision, which has received support through Aero EDIH via the Test Before Invest service package.
“Synclair Vision with its development of intelligent camera systems were exploring how ML-based vision could be used to strengthen surveillance capabilities. The primary challenge was the training of these models and the company was granted support to develop an environment where synthetic data could be generated to train new models. Overall the project and support allowed Synclair Vision to accelerate the validation process of ML-based computer vision. Aero EDIH together with the delivering company helped continuously with structure and problem solving”, says Carl Wikström, CEO, Synclair Vision.
Challenges
Synclair Vision, a startup specializing in smart camera systems for drones, aimed to move beyond basic imaging and develop autonomous anomaly detection for public safety applications such as surveillance and wildfire detection.
The challenge was to accelerate AI model development without relying on costly and time-consuming real-world data collection and manual annotation. To achieve this, Synclair Vision needed a scalable solution for generating diverse, high-quality training data under varying environmental and lighting conditions.
Solutions
In collaboration with ModulAI and Aero EDIH, Synclair Vision implemented a synthetic data generation pipeline using Unreal Engine 5. ModulAI developed a simulated environment capable of producing large-scale annotated image datasets, including bounding boxes, to train detection algorithms. The tool was delivered to Synclair Vision for further development and was also utilized in a master’s thesis evaluating online learning methods for wildfire detection.
Additionally, practical test flights in Linköping were conducted with RISE to collect real-world imagery using drones equipped with different cameras and IMU sensors, supporting validation and refinement of detection models.
Results and Benefits
The project successfully provided Synclair Vision with both synthetic and real-world datasets, enabling further development of AI-based detection algorithms. By reducing dependency on manual annotation and extensive field data collection, the solution significantly accelerated model training and improved robustness across diverse conditions.
The collaboration strengthened Aero EDIH’s role in fostering innovation and positioned Synclair Vision to advance toward commercial applications in public safety, surveillance, and wildfire monitoring. Future steps include refining detection models and exploring operational deployment in drone-based safety solutions.
Perceived social and economic impact
This project supports the development of autonomous drone-based safety solutions, enhancing capabilities for wildfire detection and surveillance. By enabling faster AI development, Synclair Vision can potentially bring innovative products to market sooner, contributing to public safety, environmental protection, and cost-efficient monitoring.
The approach also demonstrates how synthetic data can democratize AI development for SMEs, reducing barriers to entry in advanced drone applications.
Lessons learned
Do’s: To achieve a robust performance simulated and real-world datasets are combined. Balance leveraging synthetic data to reduce time and cost in AI model development, with practical test flights collecting images. Engage domain experts in an early stage to ensure realistic simultation parameters.
Don’ts: Avoid relying solely on synthetic data, real-world validation still remains essential. Otherwise the complexity of anomaly detection in dynamic outdoor environments might be underestimated.
”The project established a robust link between synthetic simulation environments and real-world flight field data. Simulated environments provide the most cost-effective method for generating the large-scale training data necessary to optimize and calibrate the performance of search and rescue systems.”, says Per Bröms, Project Manager, RISE (Aero EDIH).
