Tablets and smart phones are ubiquitous these days. Most have cameras, making it easy to take pictures for personal use, such as creating photo albums and posting to social media. But we can also use them to perform asset inspections: capture a snapshot of assets and analyze the content for discrepancies or abnormalities.

Image recognition tools and services, such as IBM® Watson® Visual Recognition, allow customized image analysis. Plus, artificial intelligence (AI) and machine learning (ML) platforms like IBM Watson are making it easier to create customized cognitive classifiers to do customized visual recognition. Using these technologies to inspect assets in the field allows the results to be readily available, which is valuable when combined with the asset record.

MaximoCoreMLVision sample app

We have created the MaximoCoreMLVision sample mobile application, which uses Watson AI and ML capabilities to perform image recognition on smart devices. It then integrates the image analysis results with IBM Maximo Asset Management by creating a work order. The sample app uses the following tools and SDKs:

  1. XCode, the iOS development environment, and an iPad
  2. Watson Visual Recognition service on IBM Cloud
  3. Watson Visual Recognition tool to create custom visual classifiers
  4. Apple Core ML to integrate trained models onto iOS smart devices
  5. Maximo Asset Management V7.6.0.9
  6. Maximo REST SDK framework for iOS to integrate with Maximo Asset Management

The MaximoCoreMLVision sample app supports offline mode or no-network connectivity. In offline mode, it performs visual recognition, but not Maximo integration, so to run this in full, you’ll need:

  • Maximo Asset Management V7.6.0.9 or V7.6.1 that is reachable from the device running the sample. Docker containers for Maximo Asset Management V7.6.0.9 developer edition for development purpose is available for IBMers and IBM Business Partners at the enterprise docker registry. Check out the setup page on the IBM Maximo Developer Center for instructions on how to get access to these containers.
  • An active IBM Bluemix® account. IBM Bluemix is a platform as a service (PaaS) that provides Watson Visual Recognition services. If you don’t have an IBMid, refer to IBM Bluemix to create one or access Bluemix if you already have one.
  • A valid Watson Visual Recognition API key. Watson Visual Recognition is available on IBM Bluemix and you can refer to Watson Visual Recognition pages to sign up and create Watson credentials.
  • A complete visual recognition classifier with at least one class. You can use Watson Studio or the visual recognition tool to create a working classifier.

The MaximoCoreMLVision sample is available on the Maximo developer GitHub. The readme page in the project contains steps to set up your XCode project, download and configure Apple Core ML and Maximo REST SDK frameworks into your project. The sample can be run on your Mac using XCode simulator or deployed on your iPad or iPhone using XCode deploy/run configurations. Refer to XCode documentation for more information.

MaximoCoreMLVision use case: two assembly lines at a ball factory

The image below shows representation of a ball factory that has two assembly lines to produce colored balls, one for orange and one for white. There are occasions where the balls from one assembly line crosses into the other assembly line, causing assembly line contaminations.

Identifying contaminations

Solutions to identification contaminations:

  • Create a new visual classifier with two classes on Watson Visual Recognition.
  • Use the MaximoCoreMLVision app on an iPad to load this classifier model.
  • Run classification on the device.
  • Update the local Maximo instance with a work order when the contamination of assembly lines is identified.

The image below shows the app that has classified an image containing orange balls and displays the result of the Watson Visual Recognition classification.

Result of the Watson Visual Recognition classification

You can see that the MaximoCoreMLVision app can be used to notify the contamination of assembly lines by creating a work order in Maximo. This can be extended to many different use cases by creating your own visual classifiers with your own Watson API keys, and results can be integrated to Maximo Asset Management.

The images below show the user interfaces to configure your Watson Visual Recognition keys, visual classifiers, Maximo connection and admin credentials. This allows the app to be used with different use cases and integrate with Maximo Asset Management.

View of user interfaces

Next steps

 

Join The Discussion

Your email address will not be published. Required fields are marked *