Fujitsu has trained an AI model to detect abnormalities in manufactured goods such as frayed threads and scratches.
As part of the manufacturing process, products often have to be closely inspected for abnormalities, defects or production errors. Now, new AI technology developed by Japanese company Fujitsu could significantly reduce the number of working hours needed for this process.
Fujitsu Laboratories, a research centre under the umbrella of the multinational ICT company, has developed an AI technology for image inspection that enables highly precise detection of a wide variety of external abnormalities on manufactured goods.
The technology leverages an AI model trained on images of products with simulated abnormalities, without having to prepare training data that uses actual images of defective products pulled from the inspection process of a production line.
The challenge with training AI to conduct quality control tasks is that models are often trained using weighted and summed indices for individual characteristics. This can make it difficult to create a model that fully grasps all the characteristics it needs to inspect.
Fujitsu has developed a method to train an AI model so that a normal image without a variety of anomalies such as shape, size and colour can be restored by artificially adding the simulated abnormalities to a normal image prepared for training.
According to the company, this technology achieved more than 98pc in an AUROC score, an evaluation metric used to check classification model performance.
The score was achieved in a class of products that have variations in their normal appearance, such as carpets with different fur patterns and colours on an individual basis, and printed circuit boards with different wiring shapes on different parts.
Fujitsu also verified the technology’s effectiveness in a real-world setting at the Nagano plant of Fujitsu Interconnect Technologies, a manufacturer of electronic equipment. According to the company, the AI reduced the number of hours required for inspecting printed circuit boards by 25pc.
The manufacturing sector has become a major focus for digitalisation in recent years. Last week, an €11m European project was announced that aims to use machine learning to enhance operations and achieve zero-defect manufacturing.
Fujitsu has also been applying its AI developments in other areas. Earlier this month, the company revealed details of a facial recognition system that can measure how focused someone is in online classes and meetings, using an AI model that detects small changes in muscle movements in a person’s facial expression