A High-granularity Sizing System For Apparel Customization Based On Large-scale Simulated Data

The apparel sizing system based on large-scale data is the key to realize online customization and intelligent production of clothing.The modern clothing industry needs more accurate sizing system to achieve the mass customization of garment.China hasn't built up a sophisticated large-scale database for modern human body.The existing anthropometry database,collected 30 years ago,fails to represent the current physical condition of Chinese people accurately.Therefore,The national standard sizing system based on that old database can no longer meet the needs of modern garment companies.To solve those problems,this paper proposes a new method of simulating large-scale data of human body measurements,and develop a new apparel sizing system based on the large-scale data.Firstly,Monte Carlo method was used to generate 200,000 simulated data of human body measurements based on seed data of 15,000 subjects.Secondly,People will be divided into four shape groups according to BMI.Then,algorithm was presented for analyzing the large-scale simulated data and generating a series of apparel sizes.Finally,the new sizing system was evaluated for fit requirements in theoretical and practical way.We applied this research result in the enterprise production,and it proved the effectiveness of the new sizing system.In this paper,large-scale simulated data of human body measurements was provided and a new high-granularity sizing system was developed.It is concluded that the new sizing system has a higher coverage to accommodate the user population well.The new sizing system will help to reduce measurement costs and meet the requirement of those enterprises which produce small-volume,multiple-size,customized apparel.