Edge Computing Box

Dual-core A55 | 2.5 Tops Computing Power

● 2.5 Tops@INT8 computing power, supports INT8/INT4/FP16 multi-precision mixed quantization.

● 4 channels or more of 1080p@30fps video encoding and decoding, IVE module independently provides image basic operator acceleration.

● Supports Caffe, ONNX/PyTorch deep learning frameworks, provides basic AI algorithm examples such as ResNet50 and YOLOv5.

● MindStudio graphical development environment, integrates model conversion and quantization, and network performance optimization tools to improve application development efficiency.

● Supports Docker containerized deployment and Ubuntu 20.04 system environment, facilitating cluster management and application expansion.

● Supports multiple network communication methods (dual Ethernet, WiFi/4G networking, etc.), providing convenience for remote management and control of edge computing nodes.

● Expandable with SATA hard drive or TF flash card to meet large-capacity data storage requirements for video image caching and application operation and maintenance logs.

● Provides serial communication control (RS232/RS485, USB), reserves visual monitoring and audio-visual alarm interfaces (HDMI, Audio).

Edge computing intelligent workstation, equipped with a dual-core Cortex-A55 architecture AI processor; built-in efficient neural network inference engine, providing 2.5 Tops@NPU computing power; supports multi-channel video image recognition hardware acceleration. High performance, low cost, stable and easy to develop, multi-point wiring, networked management and control; it can meet the application requirements for deploying lightweight intelligent video analysis at the edge; it is a new generation upgrade solution for economical intelligent workstation products, widely applied in smart city scenarios such as smart security, smart parks, and smart urban management.

Item

Type

Model Parameters

Description

Core

Configuration

Processor

CPU

ARM Cortex-A55@1GHz x2

NPU

AI Computing Power: 2.5 Tops@INT8, supports INT4/FP16

Memory

LPDDR4

2 GB (default)

Flash Storage

eMMC

32 GB (default)

Hardware

Features

Encoding and Decoding Performance

Video Encoding

4 channels 1080p@30fps H.265/H.264 protocol encoding

Video Decoding

4 channels 1080p@30fps H.265/H.264 protocol decoding

Image Encoding

Supports 3840x2160@60fps JPEG protocol encoding

Image Decoding

Supports 3840x2160@30fps JPEG protocol decoding

Ethernet Port

(Ethernet)

Supports access to 10/100/1000M network (default configuration: one 100M and one 1000M port)

ETH0 supports PoE power supply

RS232

Can connect to environmental sensors/actuator control modules, or other industrial communication control devices

(e.g., environmental parameter sensing devices like temperature/humidity/smoke/infrared detectors,

actuator control modules like cooling fans/stepper motors/alarm bells)

RS485

HDMI OUT

Supports 4K@30fps video output (can serve as a backup visual monitoring backend)

Line IN

Audio input port (standard 3.5mm audio jack)

Line OUT

Audio output port (standard 3.5mm audio jack)

USB 2.0

Can connect to USB flash drives, USB mice, USB keyboards, and other devices

DEBUG

System debug serial port (Type-C)

TF Card Slot

Can connect to TF flash card to expand storage space

GPIO

Can be used to output control signals to external expansion devices

(Relay control IO x1, Digital IO x2)

LED

PWR_LED + SYS_LED

RST-Key

Reset Button

Power Interface

Supports DC 12V/2A

Expandable

Functions

miniPCIe

Optional mSATA hard drive connection to increase storage space

Optional 4G/5G mobile communication module connection for remote wireless network management and control

(SATA hard drive, 4G-LTE module, choose one)

WiFi

Expandable WiFi short-range wireless networking, supports 2.4GHz band

Linux

File System

ext4

Buildroot, Ubuntu 20.04 (Kernel: Linux 5.10)

Software

Development

Media Processing

MPP

MPP platform, provides VI, VENC/VDEC(H264,H265), VO, and other APIs

AI Development

Graphical

Command Line

MindStudio IDE (provides development, debugging, model conversion,

model optimization, model analysis, and other functions for AI applications)

Supports Caffe, ONNX/PyTorch deep learning frameworks

Basic

Modules

Network Settings

Command Execution

Supports static/dynamic network parameter settings (DHCP)

Running Status

CPU, Memory, Disk

Device Information

Device serial number, software version number

Log Management

Running status, runtime errors, etc.

Time

NTP, manual time synchronization (RTC power-off protection)

Upgrade Management

Firmware Upgrade

Software tool upgrade, supports TF card upgrade

Application Areas

Widely applied in various fields such as smart factories, smart construction sites, smart communities, and smart parks.