Video Surveillance

Many of the same reasons why Xilinx SoCs and FPGAs are extensively used in the cloud are amplified when similar functions are implemented in surveillance cameras, network video recorders (NVR), digital video recorders (DVR), and video servers. The trend toward smart surveillance systems extending from the cloud to real-world environments, known as the edge, places a higher priority on efficiency in the form of performance/watt and latency from image capture to decision due to volumes, price points, and thermal constraints of edge devices. Neural network-based machine learning inference is rapidly becoming the dominant approach surveillance equipment manufacturers rely upon for differentiation, along with custom image signal processing (ISP). Xilinx offers efficient and scalable customization through:

• In-house or ecosystem IP support for the latest versions of MIPI, SLVS-EC, LVDS, and other sensor interfaces
• Ability to build or adapt ISPs for specific use cases and replace them on-the-fly (e.g. HDR, 4K, Infrared, Blacklight, etc.) as conditions change
• No-charge, optimized OpenCV functions, called xfOpenCV, to implement existing user vision algorithms on ARM processors or FPGA fabric or across both domains
• Heterogeneous architecture and SDSoC toolflow maximizes performance/watt by offloading or accelerating bottlenecks in processor-only implementations
• Real-time analytics and machine learning supported by in-house or ecosystem solutions such as PYNQ, DEEPHi, and others
• Connectivity to storage and network, including support for Time-Sensitive Networking (TSN)
• IEC 62443 cybersecurity with built-in hardware root-of-trust
• Integration of control, IEC 61508 SIL 3 functional safety and other functions for vision-guided robotics and non-standalone camera applications

Choose from an extensive assortment of kits from Xilinx and ecosystem partners that best match your needs by selecting the boards and kits button above and filtering by board function.