AI Verse and Soloma Avionics are Finalists for DAIC Partnership of the Year Award
Recognizing joint innovation in thermal UAV detection for frontline defense
When AI Verse and Soloma Avionics began working together, our shared goal was clear: improve thermal detection performance where it matters most – saving lives in Ukraine.
Our partnership has now been recognized by the Defence AI Council as the Finalists for Partnership of the Year award. It honors projects that show measurable operational impact through effective collaboration between AI developers and defense technology companies.
Why Detection Matters
Shahed drones are a persistent and lethal threat. With thousands launched every month, detecting them early, especially under low‑contrast night‑time conditions, is essential.
Soloma’s existing detection system had solid benchmark scores but struggled against the unpredictable conditions of the field. To strengthen it, AI Verse used our procedural synthetic data engine to generate an infrared dataset that mirrored real mission environments.
The Synthetic Edge
By fine‑tuning Soloma’s models with our synthetic infrared Shahed‑136 dataset, the team achieved:
- Detection recall boosts from 9.4% to 51%, later stabilizing at 42% in field conditions
- Zero false positives
- 100% detection precision
- 1.13 s faster first detections, giving operators up to 23 m more response distance
These improvements mean earlier awareness, faster reaction, and reduced operator fatigue under pressure.
Shared Impact
This project shows what’s possible when synthetic data and field experience meet with purpose. We are thankful to Soloma’s engineers for their deep commitment and to DAIC for recognizing the effort.
More Content
How to Build Better Computer Vision Models
Computer vision (CV) is revolutionizing industries such as smart home, security, and defense. From enabling fall detection to powering detection of weapons, CV applications are reshaping the way we interact with technology. However, achieving high-performing CV models remains a challenging task due to the dependency on high-quality, diverse datasets. Explore how synthetic images can address […]
Discover how synthetic data revolutionized our tank detection model training.
Training a tank detection model using conventional data presents several challenges. One of the biggest obstacles is the scarcity of labeled data. Tanks are not everyday objects, and acquiring enough annotated images for training is extremely difficult due to confidentiality of images.
How We Leveraged Synthetic Images to Train a Fall Detection Model
In the development of a computer vision fall detection model, one of the biggest challenges is obtaining high-quality, well-annotated image datasets. Real-world fall datasets are scarce due to privacy concerns, ethical constraints, and the difficulty of capturing diverse fall scenarios in real life. We tackled this challenge by leveraging synthetic images to train a highly […]