1. EAIS Positioning
EAIS (Edge AI Server) is a series of software-defined hardware edge AI computing platform products jointly developed by OPEN AI LAB and industry partners, consisting of three main components:
1) High-performance embedded hardware platform (compatible with various computing power forms from 1-16T);
2) AIoT application development framework Tengine;
3) Business-level algorithmic application solutions;
In summary, it provides customers and partners with a high-efficiency, flexible AI edge platform.
EAIS Product Series (Partial)
2. Three Major Features of EAIS
As an edge AI server, EAIS has three major features:
First, EAIS is the center of edge AI computing power
Its flexible computing power configuration allows it to adapt to the computing needs of different scenarios, achieving cost-effective solutions. Through deep learning algorithms, it converts raw data into structured data, realizing multi-dimensional structuring of the physical world, providing comprehensive solutions and global intelligence for various scenarios. It supports the access of raw data from multiple devices through excellent transmission and processing capabilities, and its powerful computing power supports the efficient operation of multi-dimensional algorithms, while providing robust intelligent upgrade solutions for real-world scenarios.
Second, EAIS is the practical implementation of AI frameworks in embedded systems
EAIS is designed based on a highly reliable embedded architecture, serving as a sustainable deployment, continuous integration, and continuous development edge AI computing platform. The AI Framework (Tengine) provides universal support for various CPUs/GPUs/XPUs/DSPs across different SoCs, completing a unified abstraction of the hardware computing power layer, with comprehensive support for numerous operators across platforms, adapting to mainstream deep learning frameworks (Caffe/Mxnet/TensorFlow, etc.), and providing a complete toolchain for quantization/cropping/retraining, offering adaptable solutions for rapid algorithm migration and business deployment.
At the same time, OPEN AI LAB provides an algorithm automatic training platform (Auto-Training) for industry clients, facilitating the rapid customization of differentiated algorithmic application scenarios.
Third, EAIS is born for industry solutions
Based on a flexible underlying computing power platform and AI development framework, EAIS can support rapid algorithm and application-level expansion, quickly providing solutions for niche fields, and supporting rapid deployment and continuous upgrades.
Basic functions are as follows:
EAIS can greatly reduce the difficulty of system deployment and upgrades for partners, significantly lower usage costs and network pressure, providing more universal solutions for intelligent scenarios. The flexible and open model supports users to directly utilize the application solutions currently supported by EAIS, supports continuous downloads and updates for new application solutions supported by EAIS, and also supports users in developing their own algorithms and application models on the edge AI platform.
3. EAIS Application Scenarios and Industry Cases
Currently, EAIS has been widely promoted and used in industries such as safe cities, smart communities, smart transportation, safe campuses, and new retail, providing services and solutions for face detection and capture, abnormal behavior analysis, safety production management, VIP (stranger) precise identification, vehicle analysis, attribute analysis, and passenger flow statistics. By rapidly expanding with partners, it provides differentiated solution strategies for product solution providers and system integrators in various scenarios, where software defines hardware, and algorithms are applications.
01
Personnel Analysis:
Based on deep learning face technology, intelligent transformations have been made to high-definition IPCs built in key locations, achieving face detection, posture filtering, and quality selection at the front end, providing high-quality configurable face capture to the back-end big data platform for accurate and effective data information.Usage scenarios include face capture/tracking/deduplication, personnel identification (structured/comparison), multi-dimensional attributes (gender/age/glasses/hat/mask/attention angle/expression, etc.); human body detection analysis, etc.;
02
Safety Monitoring Analysis:
Based on deep learning object recognition technology, it achieves real-time analysis of safety helmets (clothing), flame detection analysis, and other abnormal behavior analyses;
03
Vehicle Analysis:
Currently supports vehicle recognition, including license plate recognition, vehicle type (color, brand, model, etc.) recognition; supports non-motor vehicle recognition, detection, and capture;
04
Object Recognition:
Based on deep learning technology, it analyzes video data from network cameras, detecting and recognizing multiple target objects, and displaying the corresponding recognition results in real-time. It has supported classification recognition and tracking of various common objects.
4. Significance of EAIS for the Industry
EAIS is designed based on a reliable embedded architecture, serving as a sustainable deployment, continuous integration, and continuous development edge AI computing platform, achieving compatibility and rapid support for lower hardware platforms and upper algorithm applications based on the AI application framework layer Tengine. In the future, it will further share and win-win with a wide range of hardware partners, algorithm partners, and scenario-based landing partners, jointly committed to providing faster, better, and more cost-effective comprehensive solutions.
Creating a new business model for edge services that supports rapid edge deployment of the entire series of deep learning algorithms, centered on scenario solutions, achieving a highly open platform, enabling a broader significance for various industries and universal AI.
For more information about EAIS, you can contact us via email:
