Maximizing Maximo Visual Inspection for Improved Non-Destructive Testing Accuracy

In this blog learn how to improve non-destructive testing (NDT) accuracy using Maximo visual inspection. Follow along with Karim, an experienced application consultant from MACS, as he walks you through the basics of NDT, an overview of surface testing, and a step-by-step guide to using the Maximo visual inspection application to quickly detect defects and anomalies. Boost your NDT game today with this easy-to-follow guide!

Visual inspection 4

How to Improve non-destructive testing accuracy using Maximo visual inspection

According to the NFX_60 standard, maintenance is a set of technical, administrative and management activities which aim to ensure the continuous and efficient operation of equipment. Despite implementing a rigorous maintenance strategy, defaults are inevitable, and they can appear in different forms and sizes which could not be visible. Induction motors could suffer from broken rotor bars or misalignment. This default could damage the entire system if serious actions are not undertaken.

Maintenance 4.0 has offered tools and techniques to detect small defaults that occur in large systems. However, it’s not always easy to connect equipment to multiple devices. There are boundaries to this practice like a budget limit or simply limits related to the equipment design. Therefore, if a piece of asset is not well equipped with instruments, it’s important at least to inspect it visually. In many cases, the asset must be disassembled in order to conduct a thorough inspection and therefore detect cracks and other forms of defaults. Nevertheless, the non-destructive testing could be an alternative to disassembly.


Karim Gmir

MACS EU – Application consultant

More content:

In this article, I will present an easy inspection solution which allows the maintenance operator to quickly detect defaults and other forms of anomalies. I will go through the non-destructive testing (or NDT) definition, surface testing overview and finally the application designed in Maximo visual inspection part of the IBM Maximo application suite.

What does Non-destructive testing mean?

Well, it speaks for itself. Unlike destructive testing, it simply means testing by collecting information about a piece of a equipment without destroying it or disassembling it. NDT offers a list of tools and techniques which allow maintenance operator to detect defaults not only on the surface but also inside a material.

In many industries, NDT turns out to be a key method in inspecting critical junction like weld which is the main topic of the application that I will describe later. As mentioned previously, weld is a critical part in large infrastructures, it can deteriorate with time and other factors such as:

  1. Poor weld quality
  2. Environmental damage: extreme temperature, moisture, air quality.
  3. Incorrect welding technique
  4. Fatigue

Six techniques for welding NDT exist in the industrial world:

  1. Ultrasonic testing NDT
  2. Magnetic particle testing
  3. Acoustic emission
  4. Radiography
  5. Eddy current
  6. Dye or fluorescent penetrant

In this article, I chose to build the application around dye penetrant testing by using the result from this test to train an artificial intelligence model which detects cracks on a piece of material. The following video presents a dye penetrant application.

The NDT results rely on visual inspection which is an operation performed by the maintenance operator. The material must be transported to the laboratory for testing and inspection. This could take time and effort for an easy and quick testing. Dye or fluorescent NDT can be conducted on a bigger scale like in large infrastructure of confined spaces.

In the next chapter, I will explain how it is possible to automate result interpretation of the dye penetrant-based NDT using Maximo visual inspection capabilities. Then, I will present an interesting end-to-end use case of non-destructive testing using drones and Maximo visual inspection.

What is Maximo visual inspection?

Maximo Visual Inspection is a software solution offered by IBM that uses artificial intelligence (AI) and machine learning (ML) technologies to automate the visual inspection process for assets such as equipment, machinery, and infrastructure.

It enables businesses to analyze images and videos captured through various devices, such as cameras and drones, to find defects, anomalies, and other issues that may affect asset performance or safety. The software can also help to predict and prevent future failures and supply data-driven insights for maintenance and repair decisions.

Maximo Visual Inspection (MVI) can be used across a range of industries, including manufacturing, transportation, utilities, and government. It aims to reduce the time and costs associated with manual inspections while improving accuracy and reliability.

MVI allows you to build and train models without writing any line of code. Only basic understanding of the business case is needed to build an application in Maximo visual inspection.

The following section describes the application I created for fluorescent penetrant-based NDT. I will share how I trained the model and how it is possible to use it to detect cracks or anomalies on weld joints or other surfaces.

How to detect cracks with Maximo visual inspection?

Data is the key part of the entire process. It is crucial to have clear pictures with high resolution to train the model. I don’t have any recommendation in terms of resolution, size and neither contrast but just make sure that your pictures are clear enough so that cracks can be easily discerned by human eye. The best case would be to use the same camera you intend to deploy with the model in production. These are few pictures I used to train the model.

click to enlarge
click to enlarge

The next step would be labeling the objects (or pictures). Maximo visual inspection offers an easy tool to draw and insert forms in order to identify “Label”. In our case, I used only one label which is cracks. Again, the purpose of this application is to identify cracks on surfaces after applying a fluorescent penetrant. This is how I did in MVI.

click to enlarge
click to enlarge

Having clear, high-resolution pictures is a good step. However, you should not expect good prediction results from a model trained with few pictures. For this matter, MVI comes in with a powerful algorithm which generates more pictures based on those imported and labelled by the user. MVI applies a high variety of filters and masks which it uses to feed the model. Depending on how many filters are applied, the dataset could grow to 10000 pictures. Use this functionality by pressing the “Augment data” button.

Click to enlarge

The next step would be to deploy the model and start testing. You can test by uploading an image directly from the platform, from your mobile phone or from another application by using MVI’s API. Here is an interesting article about Maximo visual inspection with more technical details.

What’s so special about MVI?

From a technical point of view, nothing. Artificial intelligence based on image recognition and classification has been on the market for a long time now. This platform has been developed using the exact same open-source tools available for any other organization. But here are my top 3 advantages of using Maximo visual inspection.

Click to enlarge

1. User Friendly

It took me less than half an hour to build the model I presented for this article, and I spent most of the time waiting for loading. I did not write any line of code. So, anyone from your maintenance department can train and deploy a model in few minutes.

I have been using IBM’s products and services for 6 years now. In my previous article “Optimize the MTTR using the MCSA in Maximo.”, I deployed an artificial intelligence model using IBM SPSS in a service called IBM studio. Although results were great and accurate and the application seemed to work perfectly, it was not easy to build. I had to create multiple layers like “Objects,” “Gateways” and “Pipelines” to name a few. It could be interesting for tech-savvy people to get familiar with these products especially when the intention is to build various applications. With Maximo visual inspection, it is completely different. The decision to build such a tool is driven by marketing. IBM has decided to speak directly to its target: “Maintenance people” who are craving for such capabilities.

By providing maintenance people with friendly to use and sophisticated application, IBM broke the ice by cutting these multiple layers that could discourage people from using this extraordinary tool.

2. Process automation

The Maximo visual inspection is part of the IBM Maximo application suite with all its components which makes MVI already connected to Monitor, Manage, Health and Predict. It is possible to raise alerts from MVI in Monitor in order to generate work orders or service requests on specific equipment. You can customize your business process in Maximo Manage by designing your own workflow, so you make sure that the right people are aware and alert at the right time.

These tools are provided in one full integrated product allowing to process maintenance work and track assets across your business operations.

3. Cost savings

As stated previously, MVI is easy to use and not a single line of code is needed to build artificial intelligence model based on image recognition. Neither data scientists nor developers are required to build such an application for your maintenance department. You do not need any additional plan for a server to run an application which would be deployed for the same purpose. You do not need to build any extra communication layers to process results form the AI model to your CMMS or EAM platform nor a team to make sure that interfaces are working fine. You bottom line will suffer if you opt for implementing external applications and interfaces.

MVI, part of Maximo application suite, is based on the appoint licensing model which is based on the principal that you pay as you go. In this article, you can learn how the appoint licensing model works.

Maintenance use cases with MVI

1. Inspection in hostile spaces

Safety is your number one priority. Before conducting any maintenance work, it is important to define the risks and hazards involved in the safety procedure. With the emergence of advanced robotics and drones, routine inspection across the plant can now be conducted remotely from a control room.

Robots and drones could be deployed in hostile environments to inspect critical infrastructure. These Nextgen workers would be well equipped with cameras and high precision localization system for the maintenance operator to track their progress in real time. They would take high resolution pictures which would be processed with Maximo visual inspection to detect cracks, or any sort of anomaly defined in the model.

2. Check precautions

With Maximo visual inspection, you can check whether your maintenance workers are well equipped to undergo certain work conditions. If work must be conducted in high altitude, you can make sure that operators wear oxygen masks and harnesses.

You can prevent people from working by revoking access to the work location if your CMMS system is connected to the building and all these steps would be automated using MVI, Monitor and Maximo Manage.


In this article, you learnt how to automate non-destructive testing by implementing Maximo visual inspection. If you are interested to know more about the Maximo Visual Inspection or how to implement it in your organization, please fill in the form below and one of our MACS team will be in touch to help you with your query.

Karim Gmir

MACS EU – Application consultant

Check out Karim’s Medium for more Maximo techincal content:

Contact us