MSX-AIRIS
SDK for machine vision and industrial automation

The MSX-AIRIS helps developers to accelerate the industrialization of computer vision solutions. The skills allow the developer to customize the algorithms executed on the hardware module. With our developer-friendly tools based on a SDK combined with a Visual Studio Code plugin, machine vision engineers are able to develop on any platform : Linux, Windows and OS/X.

Integration, sensors and actuators interfaces, cyber-security and industrial communications are managed by the MSX-AIRIS. Thanks to our vision SDK, developers can focus on their business logic. The tool offers native deep learning model integration as well as high-level video processing libraries in Python. Productivity is greatly improved since the customer does not develop a complete system but a small application that meets his needs.

Faster time-to-market for your AI-based vision system

Production-ready

An edge platform that allows easy deployment of 2D and 3D computer vision processing, combined with deep learning algorithms, in a production-ready software system.

Developer-friendly

A development environment designed by developers, for developers. The Visual Studio Code plugin allows you to manage deployed skills and AI models, and to debug algorithms with a live visualization.

State of the art and proven standards

A solution based on a modern software stack, allowing developers to integrate existing OpenCV or Open3D algorithms, as well as deep learning models such as YOLO, ResNet or U-Net.

Technical specifications

Core

  • Lightweight for embedded applications and real-time processing
  • Dynamic behavior through execution of multiple ADDI-DATA Skills

Vision

  • Native support for deep learning models that enables GPU acceleration for classification, detection and segmentation
  • Native support for industrial color cameras (UVC, GigE, RTSP)
  • Native support for stereo vision cameras (Intel RealSense)
  • Automatic detection of surrounding cameras
  • Guided deep learning model training
  • Video recording

SDK

  • Designed to combine deep learning, classical computer vision and stereo vision
  • Easy application development with the MSX-AI-5000 Python library
  • Integration of standard computer vision Python packages (OpenCV,
    Open3D, scikit-image…)
  • Allows the same skill to be processed on multiple cameras
  • Skill management with triggered launch possible

Visual Code Studio Extension

  • Live visual probes to check algorithms at different processing steps for easy debugging on live sources or recorded videos
  • Custom models deployment, with GPU-accelerated engine for real-time inference* (*Check the supported model list)
  • Multi-target skill deployment
  • Cross-platform development environment

User interfaces

  • Android app & web interface
  • No-code system administration
  • Secure system authentication with privilege levels
  • Camera management, live playback and parameter settings

Communication interfaces

  • Secured & encrypted MQTT & OPC UA protocols for reporting
  • Communication clients available for custom control routines

Actuators

  • Compatible with ADDI-DATA industrial systems for measurement and regulation tasks
    • Analog I/O
    • 24 V Digital I/O
  • Industrial PLCs control
TOP