Aotu* BrainFrame® Smart Vision Platform for Edge Computing

Published: 09/07/2021

Overview

Run video processing and AI inference on edge computing devices with this smart vision platform from Aotu*. Integrate with Intel® Edge Insights for Industrial for various vision applications.

This solution is provided by Aotu and can be requested at https://aotu.ai/en/download/

  • Programming Language: Python*, C/C++*, SQL*
  • Available Software: 
    • BrainFrame® Server and Client Software
    • VisionCapsules® Library 

Target System Requirements 

  • Intel® Core™ i7 processor  
  • Linux* is supported in general; Ubuntu* 20.04 is recommended 
  • At least 6 GB RAM
  • At least 250 GB hard drive

How It Works 

BrainFrame® has been deployed in multiple verticals such as industrial IoT, enterprise and retail for video analytics and real-time smart vision IoT applications. BrainFrame® has the following features:

  • Provides optimized real-time video processing and AI inference on edge computing devices. (Watch a YouTube* video.) 
  • Easy to use and fast to deploy: No code AI to deploy video AI solutions. Polished user interface with built-in video and AI overlay, smart lines and zones, alarm and alert interface, VisionCapsules® algorithm load/unload/configuration, identity interface, database and analytics dashboard, etc. 
  • Fully customizable by developers and users: REST API is 100% provided for you to build your own graphics user interface. You can use a selection of prebuilt algorithms or your own algorithms.

architecture diagram

Modules

BrainFrame® Server or Client can be installed on the same edge computer or different computers with TCP/IP connection. 

BrainFrame® Server and Client Software 

BrainFrame® Server is deployed on Linux through Docker*. It includes all video processing, AI inference, APIs services, database, and dashboard services, etc.  

BrainFrame® Client is used to access BrainFrame® Server. Linux* and Windows* are supported. The BrainFrame® dashboard can be accessed directly through the browser. 

VisionCapsules® Library 

Computer vision algorithms and deep learning models are encapsulated in the OpenVisionCapsules® format, which is designed to support downloadable and interoperable AI. When you download VisionCapsules®, the system will load the algorithm and start inference in seconds. The algorithm pipeline and algorithm fusion will be built automatically. 

BrainFrame® has built-in Intel® Distribution of OpenVINO™ toolkit optimization support. Check out the VisionCapsules® marked with “OpenVINO™”, “iGPU”, or “HDDL” for best performance on edge computers with Intel processors: VisionCapsules® download


Get Started

This solution is provided by Aotu and can be requested at https://aotu.ai/en/download/

  1. Install the BrainFrame® CLI on Linux. 
  2. Install and start the BrainFrame® server on Linux. 
  3. Install the BrainFrame® Client on Linux or Windows. 
  4. Get a License

Run the Solution 

  • Add a video stream
  • Select and download VisionCapsules®.
  • Install a VisionCapsule®
  • Add an Alarm.
  • Add a Zone.
  • Watch actionable insights on the BrainFrame® Client, and check out your data immediately on a dashboard through your browser.
  • If you decide an integration with your own application is needed or a customized algorithm is needed, check out the Tutorials for using REST APIs and building your own VisionCapsules®. 

Summary and Next Steps 

You successfully downloaded BrainFrame® and should be able to run BrainFrame® from end-to-end, watch Video/AI overlays, set up smart lines and zones, trigger alarms, save analytic results in database, analyze structured data from video and create graphs and view on dashboards. 


Learn More 

To continue learning, see the following guides and software resources: 


Release Notes 

BrainFrame® Release Notes.  


Support Forum 

Visit the support forum for developer support. 

Product and Performance Information

1

Performance varies by use, configuration and other factors. Learn more at www.Intel.com/PerformanceIndex.