Alibaba AI improves quality efficiency by 90%
Alibaba has reported that, since October 29, its AI algorithm has been applied in “many” textile and garment factories across China, enabling “automatic” quality inspection for raw materials, grey cloth, finished cloth and ready-to-wear garments.
The technology’s defect identification accuracy exceeds 90%, which is higher than the manual worker version (70%). Alibaba data also shows a “five times over” improvement in defect detection efficiency.
As most quality inspections are currently carried out manually in China, the country’s production efficiency remains relatively low (caused by worker fatigue, among other factors).
After being trained in the properties of different types of cloth via Alibaba’s DAMO Academy, the AI technology is able to “correctly identify defects up to one-tenth of the diameter of a strand of [human] hair”, the company stated.
The AI can detect wrinkles, light spots and stains from grey cloth to finished fabric and is therefore able to “assist” quality inspection personnel in making quality control decisions.
Currently, Alibaba AI can inspect the quality of “mainstream” fabrics, such as cotton, chemical fibres, denim and leather.
Image: Alibaba’s Xunxi factory via The Global Times.
