SICK AppSpace SensorApps
Intelligent Inspection

Solves complex machine vision applications with deep learning

Your Benefits

  • Deep learning solves complex machine vision applications where rule-based tools are insufficient
  • Reduce waste and warranty claims and improve reliability and productivity
  • Cost-efficient ownership as deep learning runs on device
  • Fast and easy on-device application buildup as well as optimized accuracy and execution speed with training in SICK dStudio
  • User-friendly and quick to learn
  • Combined benefits from traditional rule-based tools with deep learning side-by-side
  • Easily expand and customize functionality

Intelligent Inspection

Overview

Solves complex machine vision applications with deep learning

The deep learning-powered Intelligent Inspection toolset of SICK Nova enables powerful anomaly detection and object classification that is not possible with rule-based machine vision. The combination of an example-based approach with on-device training and user-friendly interface paves the way for simplified solution development. The anomaly detection and classification tools ensure that inspected items fulfill required quality and sorting demands, which helps to improve yield, reduce waste and increase customer satisfaction. In addition, all traditional rule-based machine vision software tools from Quality Inspection toolset are included.

At a glance
  • Inspection by deep learning technology
  • Anomaly detection tool and classification tool
  • Runs on supported SICK vision sensors
  • On-device or dStudio-based labeling, training and evaluation
  • Web user interface
  • Traditional rule-based machine vision software tools included
  • SICK Nova Tools plug-in support

 

Artificial Intelligence Teaser Box Image
Artificial Intelligence Teaser Box Image

Industrial Artificial Intelligence

Unlocking the full potential of Intelligent Sensors with deep learning solutions

Learn more

Advantages

Machine vision beyond traditional limits

AI Anomaly detection tool

The anomaly detection tool finds anomalies in the image and is suitable for complex applications with unpredictable defects. The application areas include inspections of surfaces or welded, glued or soldered pieces as well as injection mold tool inspection.

Training is done solely on good images, and the tool provides the result OK or NOK. It then also presents the detected defect on an anomaly heatmap.

The anomaly detection tool allows for fast and easy on-device application buildup with the capability to train up to 100 images directly on-device (or with the SICK AppEngine as emulator).

AI Classification tool

The classification tool sorts visually similar objects and is suitable for complex classification and sorting tasks of deformable, variable, organic and reflective material. Convenient for assembly verification and defect classification.

This tool is trained on similar amount of images from all classes and provides the object class that has been recognized somewhere in the image as a result.

AI Classification supports on-device training, allowing quick and easy application buildup. Often only requiring a small number of images, AI classification can train directly on-device with up to 100 images.

AI Classification (dStudio) supports training using the SICK dStudio service for labeling, training and evaluation in order to generate an optimized neural network for increased accuracy and execution speed. With this solution, it’s possible to train with more than 100 images.

Reduce waste and warranty claims and improve reliability and productivity

Fast and easy on-device application buildup. For optimized accuracy and execution speed of classification tasks, the labelling, training and evaluation can be done with SICK dStudio.

Combined benefits from traditional rule-based tools with deep learning side-by-side

The Intelligent Inspection toolset runs on supported SICK vision sensors and contributes to fast and easy on-device application buildup as well as optimized accuracy and execution speed thanks to training in SICK dStudio.

Traditional rule-based machine vision tools from Quality Inspection toolset are included, making it possible to use the benefits of the existing tool in tandem with deep learning capabilities.

Expand functionality with Nova plug-in tools available in the SICK AppPool or quickly create a custom plug-in tool with the SICK Nova plug-in API.

Supported Products

Product families
Products

Applications

Technical overview

 
Show all
Hide all
  • Technical data overview

    Technical data overview

    Task

    Training of neural networks for classifying images

    Presence inspection

    Quality inspection

    Classification

    Anomaly detection

    Measuring, 2D

    Technology

    Classification of images based on artificial neural networks

    Deep Learning

    Image analysis

    2D snapshot

    Language

    English

    German

    French

    Italian

    Spanish

    Japanese

    Korean

    Chinese

    Supported products

    InspectorP series

    SIM1012

    SIM1004

    SIM2000

    SIM2500

    SIM4000

    SICK AppEngine

    InspectorP61x

    InspectorP62x

    InspectorP63x

    InspectorP64x

    InspectorP65x

    SIM2x00 + picoCam2 / midiCam2

    Minimum screen resolution1,366 px x 768 px
    Supported browsersGoogle Chrome (version 80 or higher)
Show all
Hide all
All technical data can be found accompanying the individual product

Downloads