From Seeing to Understanding

Edge-based Camera Product Series

On-device AI · Supports Multiple Models · Custom Development

Contact Us +

Unimage Pi 

Series

资源 1 Intelligence Terminals

资源 1 Intelligent Driving and Transportation

资源 1 Industrial Identifcation

资源 1 Pet monitoring

资源 1 Smart Security

资源 1 Intelligent trafic management system

资源 1 Smart Home

资源 1 Smart Fitness Equipment and Devices

资源 1 Smart Payment

矢量智能对象.jpg

PCBA Display Diagram

矢量智能对象(1).jpg

Algorithmic capabilities

Key Parameters

Key Parameters

Visual Effects

2.png

Face Recognition

9(1769054234183).png

Pose Detection

3.png

Mask Detection

4.png

Gesture Recognition

5.png

Pedestrian Detection

6.png

Safety Helmet Detection

7.png

Cat/Dog Recognition

8.png

License Plate Recognition

1(1769054248559).png

Facial Expression Recognition

Key Features

Key Features

Al Camera Module

Series

Al Camera Module

The AI Vision Camera Module, powered by Kendryte’s K210 chip, is an integrated system-on-chip (SoC) that combines machine vision and listening capabilities. 

 

It features a dual-core 64 - bit RISC - V CPU with 1TOPS computing power and built - in hardware accelerators like KPU, FPU, and FFT. With core advantages of all - in - one audio - visual perception, proprietary IP cores, and strong programmability, it supports diverse multi - modal recognition and is designed for smart home, intelligent parks, smart farms, and more.

1 TOPS Computing Power

1 TOPS Computing Power

Micropython Programming

Micropython Programming

Self-learning Classification

Self-learning Classification

2.4-inch Capacitive Touch Screen

2.4-inch Capacitive Touch Screen

Open-Source Code

Open-Source Code

Face Detection/ Recognition

Face Detection/ Recognition

Feature Detection

Feature Detection

Mask Recognition

Mask Recognition

Digital Recognition

Digital Recognition

Al Visual Development

Al Visual Development

Barcode Recognition

Barcode Recognition

QR Code Recognition

QR Code Recognition

Road Sign Recognition

Road Sign Recognition

Road Sign Recognition

Road Sign Recognition

Compatible with Multiple MainControllers

Compatible with Multiple MainControllers

Handwritten Digit Recognition

Recognize digits from the image captured bycamen and display the recognized digit in theupper left comer.
Handwritten Digit Recognition

Facial Attribute Detection

Judge the gender, expression, etc. of agivee. This: model realizes rorle modelmalnct realizes the analysis of face
Facial Attribute Detection

Facial Mask Detection

Check whether a person wears a maskproperly through the camera, which canican be used for detection in public places.
Facial Mask Detection

Chart Display

Quickly build beautiful graphical interfacesthrough VGL, such as drawing variouesines, VGL, buttons, sliders and cher shapesoccreciriu glements to add iidr inteefaceeffects and more interactivity to applications.
Chart Display

68-point Facial Detection Model

Facial key point detction refes key areapositions of the facesitions of the face in agivem face image, imuding eybrows, yose,mouth, facial contour, etc.
68-point Facial Detection Model

Object Classification

Can recognize 20 different objects. When th:K210 vision camera recognizes an object,it can frame it in the image and displaythe corresponding name.
Object Classification

Main chip

Kendryte K210, Dual-core 64-bit RlSC-V Architecture

Computing Power

Up to 1TOPS

FLASH

On-board Nor Flash, ModelW25Q128JWS1Q, Capacity: 16MB 2MP Pixels

Camera

2MP Pixels

Screen

2.4-inch Capacitive Touch Screen, Resolution: 320*240

TF Card Slot

TF Card Slot*1, Max Support 512GB

USB

Type-C Interface, 3-in-1 Function: Serial Download,Communication, Power Supply 1.25T-4AW Wire-to-Board Pin, Two Sets of Serial Ports and 7Pin GPl0 Pins Led Out

Extended Interfaces

Type-C Interface, 3-in-1 Function: Serial Download,Communication, Power Supply 1.25T-4AW Wire-to-Board Pin, Two Sets of Serial Ports and 7Pin GPl0 Pins Led Out

Buttons

Reset Button*1; BOOT Button*1

LED

System Indicator*1; Power Indicator*1

Size

49.64mm*41.52mm (Overall Size)

Smart Companion PendantSeries

AI Smart Companion Pendant – Your Always-On Friend

Supports Multi-Language Switching No Language Barriers — More Versatile for Communication, Learning, and Translation
  • 资源 1Multilingual Dialogue Support

    Mandarin, Cantonese, Japanese, English, French, and more.

  • 资源 1Customizable Personality Settings

    Fiery, Calm and Dignified, Gentle, Shy and Reserved, Humorous, Frank and Innocent, and more.

  • 资源 1Image Recognition & Description

    Supports image recognition via camera, providing vivid and detailed descriptions of visual content.

  • 资源 1Compatible AI Models

    Qwen3, DeepSeekV3, Doubao 1.5, Qwen2.5, and more.

Hardware Features

HD Camera (2MP) Take photos for recognitionand vividly describe pictures

Complete Audio System Microphone input(dual microphones) and speaker output

Support Wi-Fi Bluetooth Richer loT applications

2.0-inch IPS HD color LCD screen Can beused for various display applications

Capacitive touch screen Enable fingercontrol applications

Exquisite and compact Screwless designEasy to disassemble and assemble

Hardware Features
Hardware Features

AI Vision Development Gimbal

Series

Developed based on OpenCV, ranging from basic OpenCV image processing to deep learning development frameworks, from simple image processing to object recognition, motion tracking, face recognition, identity recognition, and advanced AI recognition functions such as YOLO/SSD object detection, deep neural networks, Jetson-inference framework, PyTorch-based object recognition, gesture recognition, expression recognition, and facial feature recognition. 

Utilizing the Mediapipe image processing library, it supports cross-platform common ML solutions. This framework is well-adapted to the Jetson Nano with Python versions, offering very smooth performance. It enables the construction of a 3D depth camera for a series of 3D detection functions including gesture recognition, 3D facial mesh annotation, human body detection, and skeletal detection.

AI Vision Development Gimbal

EdgeSight — Intelligent Scene Understanding SystemSeries

Product Overview

EdgeSight is an intelligent system deeply integrating cutting-edge multimodal vision-language models (Vision-Language Model, VLM) with high-performance edge computing platforms. It enables cameras to transition from merely "seeing" to truly "understanding," capable of describing scenes, actions, and anomalies captured by cameras through natural language, just like human perception.

Core Technology & Product Features

  • 资源 1Edge-Optimized Multimodal Large Models (SmolVLM)

    Leverages the lightweight yet high-performance SmolVLM (Small Vision Language Model), a generative AI designed to comprehend images and videos. After deep optimization for edge devices, it achieves true "visual understanding" and "natural language interaction."

  • 资源 1Edge Computing Advantages (NVIDIA Orin Nano)

    Low Latency (Ultra-low delay), High Privacy (On-device data processing), and Low Cost (No cloud dependency).

  • 资源 1Wide Application Scenarios

    Smart Security, Smart Retail, Action Cameras, Machine Vision & Industrial Quality Inspection, Home Robots, and more.

Smart Security

Smart Security

Smart Retail

Smart Retail

Action Cameras

Action Cameras

Machine Vision

Machine Vision

Home Robots

Home Robots

Smart Warehouse

Smart Warehouse

Multispectral Palm Recognition System

Series

"The camera module in the multi-spectral palm recognition system developed by United Imaging is an imaging module designed for capturing palm-print, palm vein, or QR code data. This module employs an RGB+IR binocular imaging technology, allowing terminals to obtain high-definition palm-print, palm vein, or QR code image data through its Type-C interface. The image data is then used for palm-print/palm-vein recognition applications or QR code identification.


The system integrates a proprietary “palm-print and palm-vein multi-modal fusion" recognition algorithm, which combines two biometric features: the epidermal patterns of the palm and the subcutaneous venous distribution. This approach achieves an exceptionally high security standard, with a recognition speed of just 0.1 seconds. The system remains unaffected by sweat, temperature, or common contaminants during operation, and cannot be bypassed through methods such as 3D printing or mold replication. Additionally, it is contactless, enabling recognition from a distance of 8 to 15 centimeters, ensuring both safety and hygiene.

 

This system is particularly well-suited for high-security and privacy-sensitive fields, including mobile payments, e-commerce CA authentication, e-government, e-police services, smart cities, smart homes, hospitals, campuses, and building security management.

Multispectral Palm Recognition System
Multispectral Palm Recognition System

Product Features

Wide-angle lens ensures capture of palmprint, palm vein, and QR code images at a close range of 5–12 cm.

Global shutter image sensor eliminates motion distortion.

Supports an ultra-wide illumination range of 10–30,000 Lux.

Hardware-synchronized RGB-IR image capture guarantees simultaneous acquisition of palmprint and palm vein data.

Built-in fast AE (Auto Exposure) algorithm enhances rapid palm recognition experience.

Integrated proximity sensor dynamically detects hand movement, enabling contactless palmprint and palm vein acquisition.

Financial Payment

Financial Payment

Smart Home

Smart Home

Smart Retail

Smart Retail

Smart Transportation

Smart Transportation

Smart Building

Smart Building

Certificate Authority

Certificate Authority

Unimage PI

AI Camera

Smart Companion

Vision Gimbal

EdgeSight System

Palm ID

To become a global leader in AloT vision system solution

Co-creating a smarter future across industries through inneovative AloT vision systems