Introducing Neousys edge AI embedded computers supporting dual Google Coral Edge TPUs which are flexible modules that can add on edge AI processing capability to a rugged computer. Each Edge TPU is capable of performing up to 4 tera-operations per second (TOPS) while using an astounding 2 watts only!
With the wide adaptation of Artificial Intelligence (AI) being deployed to computers at the edge, together, they form a real-time interactive link that has decentralized the edge computing infrastructure. Nevertheless, edge AI device deployments are often restricted by the size, installation space, cost and power consumption. To counter these obstacles, Neousys Technology offers ultra-compact fanless embedded computers—Nuvo-2700DS and POC-400. In addition to wide temperature operations, rich I/Os and wide range input voltage, they also support Google Edge TPU to execute AI applications independently.
The edge AI technology uses trained models for inference processes on edge devices to assist in making local decisions that minimize decision delays (reduce latency) and avoid consuming excess data bandwidth (bandwidth optimization and saving data transmission costs). Neousys’ Nuvo-2700DS and POC-400 feature Google Coral Edge TPU’s powerful performance, compact size and extremely low energy consumption. When implemented into edge devices, it provides additional computing power and self-processing capabilities. The incorporation of Google Coral Edge TPUs boost edge computers with the intelligence to drive advanced automated applications, such as smart surveillance, smart transportation, intelligent manufacturing, smart retail and many others.
Product Highlights
<30W Edge AI Platform
The inference capability of Neousys' AI platform is powered by Google Edge TPUs which support TensorFlow Lite with better system performance and processing speeds while consuming less than 30W of power.
Wide-temp fanless operation
Neousys' fanless computers with low power consumption are designed and validated for operation in wide-temperature ranging from -25°C up to 70°C. They are ideal for 24/7 deployments in harsh indoor and outdoor environments.
Rich I/O connections
The compact edge AI platforms provide rich and reliable I/Os connectivity such as USB3.1, COM, PoE+, GbE and DisplayPort for edge AI applications.
Wide range DC, ignition control and wireless connectivity
Neousys' Edge TPU platforms have wide-range DC input from 8 to 35V with ignition power control, and they also support WiFi 6, 5G/ 4G network modules for mobile applications.Neousys integrated platforms with Google Edge TPU:
 |
 |
|
| Nuvo-2700DS | POC-400 | |
|---|---|---|
| Description | Rugged 4x 4K Interactive Digital Signage System | Ultra-compact Fanless Embedded Computer |
| Processor | AMD Ryzen™ Embedded V1605B | Intel® Elkhart Lake Atom® x6425E |
| Memory | Up to 64 GB DDR4-2400 SDRAM | Up to 32 GB DDR4-3200 SDRAM |
| PoE+ | N/A | Optional IEEE 802.3at PoE+ |
| Ethernet | 2x Gigabit Ethernet ports by 2x Intel® I210 controller | 3x 2.5GBASE-T Ethernet ports by Intel® I225 GbE controller |
| USB | ||
| Mini PCI Express | Support 2x half-size mini PCIe for Google Edge TPU | Support 1x M.2 E key for Google Edge TPU |
| Video Port | 4x 4K DisplayPort | 2x 4K DisplayPort |
 |
|
|
Vertical Market
Fanless Edge AI for Public Space or Retail
As Edge AI platforms receive recorded images from cameras, it utilizes the Edge TPU for intelligent video analysis. The processed result can then be uploaded to the remote control center via the network. The data can be applied to applications such as people counting or crime prevention.
AI-Enabled Digital Signage
Edge computing has propelled the diverse development of interactive media. With cameras and specialized facial-recognition technology, the age, gender, race, and physical features of audience is analyzed, and an advertisement that matches the presumable interest will play on the digital signage in real time.