Success Story: AI Classification System for Ball Surface Inspection at ibg Prüfcomputer GmbH

Brief Summary:

• ibg, a leading company in non-destructive testing machinery manufacturing, is committed to advancing its inspection systems through the use of artificial intelligence in order to secure its long-term position as a market leader. After developing initial internal AI prototypes, ibg entered a partnership with Neuroforge to leverage their expertise in building a robust AI solution fully integrated into the machine infrastructure.

  
  

• Together with Neuroforge, ibg initiated a transformative project that began in June 2020. Initially a sparring collaboration for in-house AI research, it evolved into a fully functional AI-driven software application integrated into the company's existing systems.

  
  

• The result is a custom-built AI-supported software solution tailored to ibg’s systems that significantly enhances the company’s capabilities in material classification, boosts classification precision, and further solidifies ibg's position as a market leader.

  
  

Introduction

Adapting to technological transformation is essential for leading companies in industrial manufacturing and non-destructive testing (NDT). ibg Prüfcomputer GmbH (ibg), a key player in the manufacturing of non-destructive testing machinery, recognized early on the potential of artificial intelligence to optimize its product portfolio. As an industry innovator, ibg strives to enhance conventional NDT solutions through the integration of AI.

Neuroforge has supported ibg for several years as a sparring partner in its internal AI research. The foundation of our co-innovation lies in a shared and relentless pursuit of technological progress—with a clear mission to continuously evolve the solution and deliver new impulses to the industry.

We value partnerships where innovation is mutually reinforced—and that is precisely what characterizes our collaboration with ibg: While ibg contributed the visionary concept and a self-developed AI prototype, Neuroforge provided deep expertise in AI development and system integration. Through intensive technical collaboration, the existing models were transformed into production-ready components fit for real-world operation. This fusion of mutual expertise and eye-level collaboration laid the foundation for a new generation of AI-enabled testing systems.

Vision 

For over 40 years, ibg has been a pioneer in eddy current testing. Through continuous innovation, the company has become a global leader in eddy current systems for non-destructive testing. The idea of integrating artificial intelligence into this field first emerged in 2017.

ibg quickly realized the need to go beyond established methods and had already begun exploring AI integration into its existing machine learning-based platform. Together with Neuroforge, ibg developed a machine learning model to automate the classification of ceramic balls for quality assurance—a pivotal step to significantly improve inspection accuracy and efficiency while setting a new standard for NDT machinery.

Following promising results, ibg pursued the next step: integrating extended AI functionality to further improve accuracy, reduce errors, and sustainably lower operating costs for its customers. This required developing the existing models into fully functional, user-accessible components ready for seamless deployment in live systems. Based on prior experience, ibg created a concept for a graphical user interface to support this.

At this point, the partnership between ibg and Neuroforge deepened: What began as an R&D sparring relationship quickly evolved into a joint implementation of a fully integrated and production-ready AI solution.

With Neuroforge as development partner, ibg began building a new generation of AI-enabled testing systems aimed at surpassing the state of the art. The co-innovation was driven by ibg’s visionary approach and the need for specialized AI development and system integration know-how. We brought the required technical expertise to transform ibg’s models into robust components capable of high-precision, real-time material classification.

Solution

"Neuroforge provides genuine sparring and cutting-edge technological solutions, resulting in co-innovation that truly advances progress."

Wolfgang Korpus

CTO

ibg Prüfcomputer GmbH

The collaboration resulted in a custom-built software solution designed specifically for ibg and its customer requirements. It integrates advanced AI components into ibg’s new generation of ball inspection machines, significantly expanding their analytical capabilities and creating new inspection experiences for NDT applications.

At the foundation of the AI component is a tailored interface architecture that enables seamless integration into existing machine systems. Its central advantage lies in flexibility: The interface facilitates efficient and secure communication between various software modules within the machine network—ensuring full system compatibility without integration hurdles.

The core AI functionality was developed collaboratively to enhance ibg’s existing laser-based computer vision system. Originally, this technology could generate high-resolution C-scan images of ceramic ball surfaces via laser light reflection for defect inspection.

With AI integration, these images are now automatically analyzed: defects are identified, and the balls are categorized into “OK” or “not OK.” This process is powered by a deep learning model that emulates human pattern recognition, significantly enhancing the system’s precision in defect detection and classification.

To increase transparency, the AI system was equipped with an explainability feature that gives operators clear visual insights. When a ball is classified, the system highlights specific areas on the surface indicating detected flaws—including their location and signal intensity. This is especially valuable for ibg’s customers seeking to identify and trace manufacturing defects. Such detail allows for targeted improvements in production and enhances the overall quality assurance process.

During the development of the user interface, we closely followed ibg’s design specifications while contributing our own design expertise to optimize usability. The interface includes a 3D rendering of the ceramic ball and a visual explanation of detected defects. Below this, a historical log of inspection results is displayed, making it easy to identify patterns and recurring issues. This not only improves user experience but also boosts operational efficiency by making data more accessible and actionable.

A standout feature of the system is its adaptability: It supports the classification of a wide range of ball surfaces. Beyond ceramic balls, the current AI models also support steel balls. This flexibility is enabled by a built-in mechanism for importing and integrating new models—eliminating the need for costly reprogramming. If ibg wants to inspect a new object type in the future, a tailored AI model can be seamlessly integrated.

Working in close coordination with ibg’s internal control system team, the AI application was integrated into the overall system infrastructure. This includes automated data exchange between the AI component and the machine’s main computing system via PLC APIs, enabling automated processing and utilization of classification results.

To ensure that the AI systems could deliver real-time results aligned with performance demands, various optimization techniques were applied—including profiling, benchmarking, and flamegraphing. These measures ensured high throughput and efficient data flow across the entire AI pipeline.

The training process for the models was systematically structured in collaboration with ibg. By orchestrating the training pipeline effectively, new models were developed efficiently, relevant parameters were addressed, and explainability was continuously improved—contributing to the solution’s technical robustness and practical viability.

The collaboration not only met ibg’s technological goals—it also serves as a benchmark for future-facing innovation and further strengthens ibg’s leadership in the field of non-destructive testing.

Outcome

"The partnership with Neuroforge was a key factor in our success. Through technical exchange, we were able to jointly develop our internally developed AI models into operational components for real-world use, enabling precise material classification in real time. The integration of this AI component has not only significantly reduced the error rate, but also noticeably lowered operating costs for our customers. We have thus made a significant contribution to the technological advancement of our testing platform – and we are already working on additional features to expand our technological lead in the long term."

Jaroslaw Litau, Head of Application Development, ibg Prüfcomputer GmbH

The partnership with Neuroforge enabled ibg to introduce a new generation of AI-powered NDT systems that are leaving a lasting impression across the industry. This advanced technology has not only met expectations—it has exceeded them, delivering outstanding precision and operational efficiency. The successful market launch underscores the real-world value and competitive advantage that these AI-integrated machines offer.

At Neuroforge, we are proud of a collaboration that brought together ibg’s spirit of innovation and our deep expertise in AI development. The result is a solution that strategically expands ibg’s product offering while setting new standards in industrial inspection quality and efficiency.

This project is a powerful example of what co-innovation can achieve: ibg’s AI-enabled machines are now playing a vital role in optimizing quality assurance processes across modern manufacturing.

If you're looking to enhance your service or product with cutting-edge AI and Big Data expertise, Neuroforge is the right partner. We help you uncover the possibilities of AI—and empower you to turn them into real, tailor-made solutions.