Engineering news
The company’s European branch announced the AI-enabled visual inspection process today, promising reduced costs, improved accuracy and quicker work in applications from manufacturing to healthcare.
Visual inspection is the main method of analysing images with the purpose of identifying regions of interest. It involves the painstaking analysis of large quantities of data in order to find a very small number of regions of interest. In manufacturing, this approach is used to identify defective products or defective areas in a product. In infrastructure maintenance, it finds cracks in bridges or holes in the road, while in healthcare it can be used to find patches of unhealthy tissue.
By automating visual inspection, analysis time can be greatly reduced and a substantial cost overhead removed, said Fujitsu. It also reportedly delivers more consistent results by reducing variation caused by different human interpretations.
The Fujitsu process uses a new type of defect detector that is trained on normal – non-defective – data, reducing dataset creation costs and enabling the detection of previously unseen patterns. That is combined with an AI-accelerated Graphical User Interface that enables ‘rapid cluster labelling’ based on automatically identified regions of interest, creating high-quality training datasets.
The combined result is “highly accurate, automatic, unsupervised defect detection (discovering 80-90% of regions of interest automatically) with 50-100 times faster dataset training and labelling,” said Fujitsu.
“This is the latest in a long line of Fujitsu’s novel AI technologies, applying our advanced data analytics and machine learning expertise to address complex non-intrusive testing applications,” said Dr Adel Rouz, CEO of Fujitsu Laboratories of Europe. “Once provided with a high volume of low-cost and easily available defect-free samples, our new defect detector technology is able rapidly to work out what to look for automatically.”
Potential applications include manufacturing, where cameras are placed at key points on a production line to continuously monitor product quality and identify potential defects. In the steel industry, for example, where a 2km long steel coil is produced every hour, 70,000 images are used to capture the surface of a single coil and over 1m during the course of a day. Some 80-90% of the defects could be automatically identified and assigned appropriate labels with the new process, Fujitsu said, covering around 200 types of defects.
Other potential applications include healthcare, where the technique could help diagnose and screen for abnormalities, such as chest abnormalities in X-ray scans. In the US alone, about 150m such health checks are performed each year. Manually labelling the enormous volume of resulting images is prohibitively costly and time-consuming but could be feasible with Fujitsu’s automated, unsupervised visual inspection technologies.
Content published by Professional Engineering does not necessarily represent the views of the Institution of Mechanical Engineers.