The system takes pictures of products flowing down the line, inspecting specified characteristics in one to two seconds. It can detect imperfections that are difficult to find with typical inspection equipment, such as stains, various types of damage and irregular colour. If X-ray images are used, even foreign material inside food products can be spotted.
Tests were performed at 30 companies, including leading auto parts manufacturers, and found similar defect detection rates to those of humans at over 90%. Only defective products would require reinspection by workers, allowing for a two-thirds reduction in such personnel. Existing inspection equipment at the companies tested has a detection rate of about 80%, and each line has people perform the reinspections.
NEC is strong in technologies that can catch and analyse the characteristics of and slight differences between photos. The AI creates a predictive model after learning colour and damage features from hundreds of images. Currently, specialists have to input such information for each product in advance. Such work can be eliminated with NEC's system.
AI-equipped inspection systems are still rare worldwide. Usage fees, excluding initial investment costs, are about 300,000 yen (US$2,692) per month for each production line.
If you want to read this article in Japanese, please see the following link: