Introduction: The brain is Baidu Baidu AI core engine technology, including vision, speech, natural language processing, knowledge maps, depth of learning AI AI core technology and open platform.
Based on Xilinx Zynq UltraScale + EdgeBoard core MPSoC accelerated program is a key component of Baidu AI accelerated platform. Zynq which ARM processor chip integrated + GPU + FPGA (and Connections Video the Decode ) architecture, both a multi-core processing, and video stream decoding hardware processing capability, further comprising a programmable FPGA features. Built-in Linux 4.14.0 depth learning system and pre-installed environment, and Baidu brain model customization platform (AIStudio, EasyDL, EasyEdge) depth to get through, to achieve training model, deployment, reasoning and other one-stop service.
Baidu / Mill jointly launched FZ3 depth study calculated card is based on Xilinx XCZU3EG of Baidu brain EdgBaord acceleration platform. It can be embedded in a variety of product forms, the realization of landing a variety of AI edge applications of intelligent vision, smart security, advanced driver assistance systems (ADAS) and next-generation robots.
Mill / Baidu FZ3 depth study calculated the card with the diverse and constantly iterative model library provided by Baidu brain can easily recognize the scene and more human face, human, animal & objects, and text.
Features:
- Priced as low as 999RMB / 1299RMB, high cost ZYNQ / AI development board
- AI network model FPGA scalable computing architecture that can be flexibly adapted quickly Iteration
- Measured count up 1.2TOPS force, MobileNet up to 100FPS, powerful performance
- Baidu brain and seamless compatibility tools platform, one-stop lowering the threshold of AI applications
- Small, multiple interfaces, good scalability, easy to embed a variety of intelligent hardware
- Industrial configuration, a high standard of excellence choice of materials and technology, the quality of
AI powerful computing performance & low power consumption
Measured performance of up to 1.2TOPS, to quantify at cutting the situation MOBILENET up to 100FPS, more than 20 times the CPU performance, power consumption is only 5-10W.
Uncropped case model to quantify, calculate card performance:
Rich development resources and tools platform
Baidu brain seamless compatibility tools platform, one-stop AI reduce development threshold
Excellent hardware design
The board based on Xilinx Zynq UltraScale + MPSoC XCZU3EG, 4 Cortex-A53 + FPGA core architecture, onboard 2GB / 4GB DDR4 SDRAM (64bit, 2400MHz) + 8GB eMMC combination of storage. Small, functional, for embedding in a variety of product forms.
Interface Schematic:
1. Strong performance, small size, easy adaptation
2. Excellent quality & decade of life cycle
A wide range of application scenarios
Suitable for intelligent security, industrial inspection, medical diagnosis, unmanned aerial vehicles inspection, research, consumer, drone and other fields.
Currently FZ3A and FZ3B products have been pre-shock, for more information please visit the official website to find out about Kashmir Advisory: http://www.myir-tech.com/product/mys-czu3eg.htm