Case studies / Defense
Multi-stream edge AI video analytics for a defense target-acquisition product
At TerraSense Analytics: technical leadership of MIST, with NVIDIA DeepStream and GStreamer pipelines processing multiple video streams with AI at the edge, Cassandra and Kafka for data, and an integrated GUI.
- Industry
- Defense
- Role
- Architect and technical lead
- Technologies
- NVIDIA DeepStream, GStreamer, Edge AI, Cassandra, Kafka, GUI
Problem
What the
Approach
The role combined technical leadership of the MIST product team with architecture of the system. It covered the project timelines, the high-level system architecture, and guiding the integration of machine learning into the product.
- Multi-stream video pipelines. The video pipelines were built with GStreamer and NVIDIA DeepStream. Data types included Electo-optical, infrared, scanline radar
- Edge AI deployment. Nvidia Jetson based devices
- Data streaming and storage. Cassandra databases and Kafka streaming were built to carry and store the pipeline’s output.
- GUI integration. Integrated with multiple external vendor GUI’s.
Result
Reliable pipelines that operated at the required framerate.