Case study of an image inspection system using AI: Surface inspection process of mirrors.

Inspection accuracy improved from 60% to 99% by developing a system using AI deep learning technology, resulting in a reduction of visual inspection errors.

We would like to introduce a case where we developed a mirror surface inspection system using AI deep learning technology in the product inspection process of Murakami Kaimeido Co., Ltd., which boasts a high market share in rearview mirrors. This development has achieved an improvement in inspection accuracy (from 60% to 99%) and reduced the burden on inspectors who previously conducted visual inspections. 【Problems with the Conventional Detection System】 ■ Difficulty in quantifying image color and other attributes ■ Ambiguity in standards due to human determination of thresholds ■ Inspection accuracy of around 60% ■ Final confirmation by inspectors through visual checks 【Effects After Implementation】 ■ After full-line implementation, the number of inspection workers was reduced by 70% ■ With an eye on overseas expansion, data obtained from the automation of inspections can also be used to optimize upstream processes *For more details, please refer to the PDF document or feel free to contact us.

Other inspection equipment and devicesImage Processing Software