A9.1 - AI based object recognition for telemetry applications
- Event
- ETTC 2024 - European Test and Telemetry Conference
2024-06-11 - 2024-06-13
Nuremberg - Chapter
- Imaging & Video
- Author(s)
- S. Hardt, M. Faber - Safran Data Systems GmbH, Bergisch Gladbach (Germany)
- Pages
- 125 - 129
- DOI
- 10.5162/ETTC2024/A9.1
- ISBN
- 978-3-910600-02-7
- Price
- free
Abstract
In today's Flight Test Instrumentation (FTI), the bandwidth of data to be analyzed is constantly increasing. This is mainly due to the need for more video channels in parallel and higher resolution video signals. This raises the problem of how to transfer all the relevant image information to the ground station during the flight. A common solution to this problem is to use a reduced bitrate or reduced framerate of the video channels for the telemetry downlink to a ground receiving station. This reduction in signal quality can cause problems during data analysis. This paper presents an approach that uses Artificial Intelligence (AI) based object recognition to detect the relevant image content of a video signal. Based on this AI detection, different actions can be performed, e.g. the detected content can be extracted and only this information is streamed in high quality to the ground station. An embedded video toolbox is provided for the instrumentation engineer to configure the AI system. The instrumentation engineer has the ability to define custom compositions and extraction of video streams. The defined outputs create video streams that contain only the relevant information of the image content in an extremely bandwidth saving manner.