Intelligent surveillance solution is one of the most widely adopted AI applications as it can be deployed for a broad range of use scenarios, delivering advanced video analysis with deep learning for almost all vertical sectors. Before they can leverage the benefits such as AI-enhanced detection accuracy and flexibility, however, businesses and organizations must overcome a significant numbers of challenges before and after a deep learning model is trained because AI-powered video surveillance solutions rely on visual data capture for analyzing/solving security, safety, and operations challenges; and such reliance demands simultaneous analysis of vast volumes of video footage.
A system integrator working with a software provider specialized in delivering on-premise and cloud-based video analytics solutions came to Lanner in search of a hardware appliance capable of hosting this software provider’s proprietary software architecture designed for running deep-learning intelligent video applications and the finished product is intended for enabling city-wise surveillance. This vision solution must accommodate Intel’s Open Visual Inference and Neural Network Optimization (OpenVINO™) toolkit for better integration and faster deployment when enabling AI capabilities at or near the edge.
Lanner’s LEC-2580 is a compact, fanless box PC with integrated pluggable Intel Movidius accelerator engine, making it a reliable hardware platform that can be installed in space-confined environments for for enabling AI and seamless integration with an Intel architecture-based gateway at the edge for applications such as 24/7 intelligent surveillance, traffic management, access control, retail and beyond.
The LEC-2580‘s processing prowess comes from Intel®’s Core™ i7-6600U, Core™ i5-6300U and Core™ i3-6100U processors. Its standout features include four RS-232/422/485 serial ports, four USB 3.0, two USB 2.0 ports and two HDMI ports, each supporting resolution up to 3840 x 2160 at 24Hz or 2560 x 1600 at 60Hz. For networking communications, the LEC-2580 offers six 10/100/1000 Mbps Ethernet ports while the LEC-2580P, a variant of the LEC-2580, comes with two 10/100/1000 Mbps Ethernet ports and four PoE ports for even greater flexibility.
Prior to deploying Lanner LEC-2580 and Intel OpenVINO™ toolkit, the end solution provided by this software provider carried out deep learning inferencing entirely in the cloud while visual decoding operations were handled at the Edge appliance. Offloading inferencing from the cloud by having processing take place at the edge reduces the required bandwidth between users’ Edge appliances and cloud servers, therefore greatly boosting efficiencies.
The eventual deployment saw not only an increase in cost-effectiveness and scalability but also significantly improved visual analysis accuracy at the edge.
Furthermore, Lanner’s close-knit and longstanding alliance with Intel is an invaluable asset and is instrumental in the delivery of this seamless solution. Lanner provides the robust hardware appliance optimized by its in-house engineering team; the software provider, the proprietary software for intelligent video applications; and Intel, the Intel® Movidius™ Myriad™ X vision processing unit (VPU), as well as the OpenVINO™ toolkit for smart vision application development.
Last but not least, the end solution is compatible with existing cameras, NVRs, and sensors, the said city government therefore benefited from backward compatibility and significant cost reduction.
The end solution since then has been deployed in its intended city for conducting simultaneous analysis of vast volumes of video footage in real-time and to this day being counted on for delivering high levels of accuracy and low false alarm rates even in complex situations.