Technology Overview
ITRI has developed a textile inspection and style matching system that uses incremental machine learning to identify corresponding textile parameters and styles. With this system, the adjustment time for new textile designs can be shrunk from 30 minutes to one minute, a reduction of 96%. By establishing an AI incremental learning framework, the system can attain a defect inspection accuracy of over 95%. The solution has integrated the textile industry’s defect image database, the machinery industry’s fabric inspection machines, and the ingenuities of academia and ITRI.
Applications & Benefits
It enables Taiwan’s machinery manufacturers to compete with international makers and possible business model transformation from a one-time buyout to a subscription-based service, thus prolonging product life cycles and increasing product values.