From March 11th to 13th, at the Embedded World 2025 exhibition, United Imaging officially unveiled its first self-developed deep binocular camera.
United Imaging's first self-developed deep binocular camera
Relying on a strong global supply chain and lean manufacturing system, United Imaging deeply integrates AIbase-driven product development capabilities into the entire design and production process to ensure efficient iteration and stable quality. With leading intelligent optimization algorithms and cost control advantages, it creates a highly cost-effective deep camera, providing accurate, efficient and reliable solutions for intelligent vision applications.
This camera is equipped with dual image sensors, using 1/4 1.3MP CMOS sensors, with an array size of 1280×1080 pixels; 2.7µm×2.7µm pixel size, with electronic synchronized global shutter, can achieve accurate image capture.
In terms of output, it supports 2x 1280×1080 @15fps and 2x 640x480 @15/30 (binning 2x2) resolution, and has rich output formats, covering point cloud maps, depth maps, RGB images, to meet diverse application needs. Its maximum signal-to-noise ratio is up to 38.3dB, the dynamic range is 64.6dB, the sensitivity is 1900mV/Lux-sec, coupled with a 63mm baseline, a 1.75mm focal length, a wide field of view of 115° (H) × 96° (V) × 157° (D), an aperture of 2.0 and a TV distortion of less than 20%, it can stably output high-quality images regardless of whether it is in dim environments or complex scenes.
In addition to the powerful image sensor, the camera also integrates a high-performance inertial measurement unit (IMU) and magnetometer. The accelerometer has a measurement range of ±8G, a resolution of 0.244mg, and a noise density of only 3.2mg; the gyroscope has a measurement range of ±1000dps, a resolution of 0.03dps, a noise density of 0.16dps, a sensitivity error of less than ±0.4%, and an output data rate of 400Hz. The magnetometer has a measurement range of ±2500µT in the Z axis and ±1300µT in the X/Y axis, a resolution of 0.3µT, an output data rate of 50Hz, and an accuracy of 1% within a test distance of 1cm - 300cm. These sophisticated sensors work together to provide the camera with a more comprehensive and accurate environmental perception capability.
Of course, the core highlight of this camera lies in its AI big model:
Strong adaptability to weak texture scenes: Through deep learning feature extraction, it can handle scenes where traditional algorithms fail, such as sky and white walls;
Improved accuracy and robustness: AI algorithms (such as representation-guided matching and loss function optimization) significantly reduce errors and generate denser point clouds;
Optimized dynamic scene processing: better robustness to problems such as motion blur and multi-path light interference.
United Imaging's self-developed deep binocular camera series will also deeply integrate the development trend of AI, accelerate iteration and update towards miniaturization, low cost, strong performance, and smarter, match the customized needs of customers, and develop adaptability for different scenes.
Stay tuned.
AI+Vision debuts at EW2025
Embedded World 2025 brings together about 1,000 exhibitors and over 32,000 professional visitors from around the world, focusing on the entire embedded industry chain and cutting-edge technologies of Industry 4.0. It is a core platform for technical exchanges, market expansion and insights into industry trends.
United Imaging is deeply engaged in "AIoT visual solutions". During the exhibition, the United Imaging booth was crowded with people, attracting many industry partners to stop and communicate. The camera modules, visual solutions, and imaging machines on display can quickly match solutions according to the specific needs of different industries and scenarios.
This exhibition focuses on the two core technology areas of camera modules and AI visual solutions. Static exhibits and dynamic demos enhance the experience for customers to understand and explore cutting-edge visual technologies. And there are widely watched star products:
Unimage Pi series is a highly integrated intelligent visual solution designed for AIoT scenarios. It is equipped with a dual-core RISC-V processor and provides multiple models and flexible memory options. It has rich interfaces and supports industrial control and multi-device expansion. It has built-in 9 types of high-precision visual models, covering security, transportation, health monitoring, human-computer interaction and other fields, taking into account high performance and easy deployment, and providing efficient and stable end-side computing power support for AI+everything visual applications.
DVS adopts the principle of bionic retina, has the ability to analyze the spatiotemporal events of optical flow fields, supports high-speed motion trajectory reconstruction, behavior analysis, and anomaly detection, and is widely used in intelligent interaction, autonomous driving, medical monitoring, industrial Internet of Things and other fields, showing excellent real-time performance, energy efficiency and environmental adaptability.
AI camera is a high-performance intelligent visual device with a built-in deep learning model. It can detect 17 key points of the human body in real time, support motion recognition and posture analysis, and is widely used in sports science, rehabilitation training, medical evaluation, human-computer interaction, security and entertainment and other fields. It is suitable for high-precision motion capture and real-time analysis in complex environments, providing a stable digital solution for human motion.
Multispectral cameras use high-precision spectral separation technology and are widely used in industrial inspection (PCB micro-defect identification, food sorting, color difference control) and agricultural monitoring (crop three-dimensional modeling, pest and disease warning). Combined with multispectral fusion algorithms, it provides a high-reliability visual solution for intelligent manufacturing and precision agriculture.
With the core concept of "Everything*Vision", United Imaging helps the Internet of Everything go from "seeing" to "understanding", enabling the intelligent upgrade of thousands of industries. We look forward to working with more industry peers in the vast AIoT market at home and abroad to explore more cutting-edge visual solutions and create a more "AI" future!
This paper is from Ulink Media, Shenzhen, China, the organizer of IOTE EXPO (IoT Expo in China)
