Large European Automotive Manufacturer | Automotive

Developing an AI visual system for industrial Quality Assurance


Infotel’s Mission:

One of our long-standing clients approached us to discuss providing an AI solution to detect and prevent defects, optimise existing workflows and assist with design & innovation. Creating predictive maintenance to increase efficiency was one of the most important indicators of success for the project.

After outlining the most common applications of AI within the automotive sector, a well-structured and measurable plan needed to be introduced on a design level, as well as establishing algorithms to enhance the use of power units to build more efficient models.

Our client :

For this challenge, our client was one of the UK’s largest car manufacturers, producing over 500,000 automobiles per year as well as employing 7000 members of staff in the UK alone. Famous for their commitment to quality, innovation and reliability, their industry presence demonstrates decades of strong sales and distribution throughout Europe.


  • Achieving 98% accuracy in defect recognition after stabilisation of deep learning
  • Infotel’s work achieved high accuracy percentages on clearer subjects, (such as holes and missing badges) and in conditions that were easily optimised to reduce false positives
  • Our client has made headway on plans to implement our AI Visual System solution across the entire factory, covering various QA processes in different settings


  • Thanks to the previous work completed with our client as part of a long and trustworthy partnership with them, we were able to have an open and productive discussion about the main challenges of providing an innovative solution. By addressing issues like false positives and environmental variables, a goal was set to create a stable, scalable and efficient new system.
  • We identified the need to generate enough test samples for any deep learning training to provide enough meaningful data. This in turn presented its own challenge of working within existing processing parameters.
  • By introducing point-and-click automation, we planned to reduce any bottlenecking in images being processed.


We knew the answer to the client’s issue would lie in developing software that supports a reliable and adaptable system. By taking action to identify and address requirements at the beginning of the progress, we also identified opportunities to reduce costs for the client.

This lead to the team taking the following steps:

  • Ensuring the software created was adaptable, and suited its industrial environment (factory stations) for QA monitoring and provisioning of test images.
  • The AI visual system was built focussing on creation, training, and refinement of defect models.
  • The application automation process was tested on a live environment.
  • Measures were taken to automate the QA process, allowing detection and the flagging of defects up to 98% accuracy.
  • All work was designed so it could be easily reproduced, and deployed on a factory-wide scale.


With key staff members and project managers receiving direct support from our integration experts, results were impressive from concept to execution. Each stage of development was carefully considered, using the following programming and software to complete the task:

  • Microsoft Visual Software
  • Microsoft SQL Server (DB engine)
  • C# Coding
  • Google TensorFlow (open-source)
  • HTML, CSS, and Vanilla JS
  • Mosquitto (MQTT) transmission protocol to enable efficient messaging between different layers through a publish/subscribe model
  • Desktop application interface
  • Deep learning software



Infotel’s consultative services will always strive to work as an extension of your business, keeping communications succinct and precise.

  • 1 Project Manager
  • 2 AI Data Scientists
  • 1 Business Lead
  • 6 Business Pilots
  • 1 Front End/Back End Developer

Contact us for more information about our process.


Infotel UK’s thorough audit provided the client with an unbiased view of their architecture and AI system implementation. The recommendations not only helped adopt a new approach successfully, but also allowed the applications of various solution optimisations throughout the process.

Aligning processes with IT resources enabled more detailed auditing and reporting, enhancing visibility and accountability. The final roadmap phase introduced various plant-wide quality assurance processes, ensuring long-term high-performance and business continuity. The software has been developed to allow for adaptation to pretty much any industrial process, building in variables to:

  • Fine-tune to different environments and station sensors
  • Encompass a range of defect types
  • Apply alternative deep learning models
  • Enable the future potential to extend from object-detection to anomaly detection and optical character definition

Infotel UK’s expertise in Artificial Intelligence and deep machine learning enabled our client to seamlessly adapt to the solution provided: a new system with a strong emphasis on automating defect recognition in industrial QA.