Search the Community

Edge Computing Access Screen is designed to provide an edge computing-based solution for access control systems. It can realize face recognition, data processing, fingerprint recognition and other functions, and improve the security and convenience of the access control system. In terms of application scenarios, intelligent monitoring and access control system is an important part of intelligent building security. The application of edge computing technology can increase the monitoring effect and response speed, avoid the security risks in the process of data transmission, and protect the privacy of users. Edge computing devices can process authentication faster and improve the response speed and security of access control systems. Hardware requirements for edge computing access control screen (1) Processor Select a high-performance and low-power embedded processor, the processor itself needs to come with arithmetic power to meet the needs of edge computing. (2) Memory Configure appropriate memory and flash for system and application data. (3) Interface Provide necessary interfaces between the carrier board and other devices, such as GPIO, UART, I2C, SPI, etc. (4) Communication module Support Wi-Fi or 4G to facilitate data transmission with the cloud platform. (5) Sensor Integrate multiple sensors, such as face recognition, fingerprint recognition, and RF card reader. Edge Computing Access Control Screen Design The FET3568-C SoM is recommended to be used as the hardware platform of edge computing access control screen. The system on module has a quad-core ARM Cortex-A55 processor with a main frequency of 2.0 GHz, and its own NPU has a computing power of 1TOPS, which can meet the needs of lightweight edge computing tasks. Memory: FET3568-C SoM supports LPDDR4 and eMMC storage, and can be configured with appropriate memory and flash memory to meet the needs of the access control system. Interface: Native GPIO, UART, I2C, SPI, Gigabit port, etc. can communicate with other lines. Communication module: FET3568-C supports wireless communication technologies such as Wi-Fi, Bluetooth, and GPS. The appropriate communication module can be selected according to the requirements. Sensors: FET3568-C supports multiple sensor interfaces, such as GPIO, I2C, SPI, etc., and can easily integrate multiple sensors, including face recognition, fingerprint recognition modules, etc. Power supply module: According to the requirements of the access control system, the appropriate power supply chip can be selected to provide stable and reliable power supply for the entire carrier board. Peripheral expansion: USB and SD card slots are convenient for secondary development and function expansion.

With the development of network technology, there is an increasing demand for storing video data over the network. As a result, surveillance systems that were previously centered around DVRs (Digital Video Recorders) have further evolved into NVR (Network Video Recorder) systems with network capabilities. NVR, which stands for Network Video Recorder, is the storage and forwarding component of a network video surveillance system. It is primarily responsible for functions such as accessing, storing, forwarding, decoding, and previewing network audio and video signals. In today's increasingly complex and widely applied surveillance scenarios, intelligence has become an inevitable trend in the development of video surveillance systems. Intelligent security monitoring is one such application of NVR intelligence. It utilizes advanced technologies and intelligent algorithms to enhance security, reduce monitoring workload, and provide real-time access, playing a crucial role in the field. They are suitable for various applications, ranging from home security to large-scale commercial and urban surveillance projects. With the constant advancement of technology, intelligent security surveillance systems will continue to evolve and improve, offering advanced features and greater convenience. Intelligent NVR can be utilized in security surveillance and, with algorithmic support, can achieve the following scenarios: Face Recognition Alarm: Comparative recognition of the target face can be performed through video or image streams. Fire Incident Visualization and Early Warning: Intelligent detection and identification of smoke and fire points, issuing early warnings before the fire spreads. Intrusion Detection and Alarm: The system can monitor an area and provide alerts for unauthorized intrusion by using object recognition and motion detection. Parking Space Vehicle Recognition: By using object recognition or license plate recognition, the system can detect and identify unauthorized or improperly parked vehicles (underground garages). Requirements Cases: A video surveillance system solution provider plans to launch a high-performance NVR (Network Video Recorder) device to meet the diverse monitoring needs of smart security applications. In the envisioned product positioning of this customer, this NVR can provide intelligent monitoring functions in various security surveillance scenarios, such as face recognition alarms, visualized fire hazard alerts, intrusion alarms, parking space vehicle recognition, etc. Unlike traditional NVRs, the implementation of intelligent monitoring functions requires AI computing power on the NVR's main control unit. This is because it needs to be able to run intelligent monitoring algorithms locally while performing traditional functions such as access, storage, forwarding, and decoding. Based on the above reasons, Forlinx Embedded recommends the FET3568-C SoM to customers as an ideal iteration tool for security back-end NVR devices. It adopts the RK3568 processor produced by Rockchip, featuring a quad-core Cortex-A55 architecture with a clock speed of up to 2GHz. Additionally, it integrates a 1 TOPS NPU, which enables lightweight local AI computing. It natively supports dual Gigabit Ethernet, which can be used for connecting network cameras, network communication, etc. It supports SATA 3.0 and PCIe 3.0, which can be used to connect mechanical hard drives or solid-state drives, enabling local storage of large video data capacity. It supports five different display interfaces: HDMI 2.0, eDP, LVDS, RGB Parallel, and MIPI-DSI. It also supports simultaneous and separate display on three screens, maximizing local preview and operation convenience. With the user-friendly model conversion tool RKNN-Toolkit, the FET3568-C allows for one-click conversion of popular architecture models such as Caffe, TensorFlow, TF-Lite, ONNX, PyTorch, Keras, and Darknet. This tool assists engineers in swiftly developing functions like face , object, action, license plate recognition, etc. Additionally, RK3568 supports various video input/output interfaces and offers hardware decoding of high-definition formats such as H.264, H.265, and VP9 at 4K resolution. It can decode multiple video sources simultaneously, making it highly compatible with NVR devices. Originally published at www.forlinx.net.