The SV820 is a cost-effective AI SoC suitable for AI-IPC, smart doorbell, AI UVC and other smart terminal devices. It integrates the self-developed NPU with up to 0.5 TOPS AI computing power, which can efficiently support intelligent algorithms such as human detection, face detection/recognition/capture, license plate detection and recognition, crying/abnormal noise detection, and so on. The SV820 ensures professional security ISP image quality, has a built-in 512 Mb DDR2 and supports 4MP@30fps video encoding. The SV820 is hardware pin-to-pin compatible with SV822. The SV820 supports fast boot and image output and achieves low-power applications. These features help the SV820 empower the development of smart home, smart office, smart village and other AI scenarios.
View DetailsThe SV806 AI SoC integrates the self-developed NPU with up to 0.6 TOPS AI computing power. It is a turnkey on-device AI SoC suitable for face recognition terminals, which can realize non-contact recognition. Besides, it provides an overall solution that is easy to develop, deliver, and implement. The SV806 has responded quickly to the needs of COVID-19 pandemic prevention since 2020 and has been widely applied to intelligent access control devices to achieve non-contact temperature measurement, ID verification, and attendance recording. The SV806 enables Al technology to help people resume work and school in an epidemic.
View DetailsThe SV810 AI SoC integrates a self-developed NPU with up to 1.2 TOPS AI computing power, which can be applied to face recognition entrance and exit gates, community access control, office attendance and other scenarios. It can be efficiently adapted to face recognition and anti-spoofing detection algorithms, preventing spoofing attacks via photo or video and supporting a face library of up to 100,000 face IDs. Besides, the SV810 integrates the self-developed high-performance ISP, which can provide high-definition video image quality in low-light environments by using the wide dynamic range and low-light ISP design.
View Details