Synadel
Synadel addresses critical data bottlenecks in defense and security sectors: the lack of real-world data needed to train AI algorithms. With Synadel’s ability to create synthetic data, superior algorithms can be trained and the time it takes to respond to new threats is reduced from months to days.
The problem
In today’s globally challenging security landscape, organizations’ ability to quickly respond to new threats is constantly contested. Failure to respond timely can ultimately lead to the loss of lives, and even in less severe cases, financial damage can be extremely high. AI algorithms, such as automatic threat recognition in airport security inspection, are essential to tackling these issues.
Whenever a new threat is identified, the automatic threat recognition algorithms need to be retrained. However, in the case of airport security devices, the current retraining process requires collecting vast amounts of training data by scanning real luggage and individuals. Such a method is
- labor-intensive
- slow, up to several months
- burdened with data privacy and security issues.
The gap between identifying a threat and updating algorithms creates notable security vulnerabilities.
Our solution
To address this problem, Synadel will bring to market superior algorithms enabled by the groundbreaking prototype developed and tested in collaboration with key aviation security stakeholders. The prototype for generating synthetic data replicating data produced by the X-ray Computed Tomography (XCT) and the millimeter wave (mmWave) -based Advanced Imaging Technology (AIT) security inspection devices has been developed under a direct commission project with over $1M funding by Irregular Warfare Technical Support Directorate (IWTSD) / U.S. Department of Defense (DoD).
Synthetic data sets created with our solution are automatically annotated and can feature any new threat object and material. With our solution, data needed to train automatic threat recognition algorithms can be created as soon as a threat is identified. There is no longer a need to scan real luggage or individuals.
At the core of the solution is a physics-based simulation of complex X-ray and mmWave radar signal interactions with virtual materials processed as a computer graphics problem combined with generative AI. With this approach, data identical to the X-ray Computer Tomography and mmWave person scanning devices can be created efficiently.
The development of this solution has been enabled by years of research at VTT and our team's unique background in computer graphics, simulation, machine learning and physics behind the security inspection devices.
Our vision
Synadel will improve security critical operations by developing the best algorithms and by helping the ecosystem stakeholders respond to new threats faster.
In the long run, the aim is to extend the synthetic data offering enabled by our solution to wise scale use in security and defence sector. With our flexible core technology, a vast number of use cases demanding physical AI beyond visible light can be addressed.
Our goal is to grow Synadel to a leading AI algorithm provider in many of the most challenging imaging use-cases in the next 5 years, starting with security and defence.