AI Image Inspection System - Mirror Surface Inspection Process

By introducing the AI image inspection system, we achieved a test result accuracy of 97%! After implementing it across all lines, the number of inspection workers was reduced by 70%.

The inspection system DEEP INSPECTION improves the speed and accuracy of the entire system and reduces the burden of visual inspections. Rist has developed a mirror surface inspection system using Deep Learning technology for the product inspection process of Murakami Kaimeido Co., Ltd., which boasts a high market share in rearview mirrors. This system has achieved an improvement in inspection accuracy (from 60% to 97%) and reduced the burden on inspectors performing visual checks. Rist's approach utilizes a convolutional network for multi-class classification. They developed a unique "Deep Inspection Confidence" and achieved a testing result accuracy of 97%. 【Issues with Conventional Detection Systems】 ■ Difficult to quantify image color tones, etc. ■ Ambiguity in standards due to human determination of thresholds ■ Inspection accuracy around 60% ■ Ultimately, inspectors confirm visually 【Features】 ■ After full-line implementation, the number of inspection workers is reduced by 70% ■ With an eye on overseas expansion, the aim is to optimize upstream processes based on data obtained from the automation of inspections. *For more details, please refer to the PDF document or feel free to contact us.

Analysis and prediction system