Add magic to magic mirror: a personalized garment size recommending algorithm

Some research of Internet of Things in online garment recommendation concentrates on collecting information of customer and garment without enough quantitative analysis on how comfortable people feel, which would make little contributions to personalized decisions about the size selection. In order to recommend the best garment size and help the producers to improve the design and production according to customer preference and purchasing records, an algorithm with functions of personalized-analysis, detail-showing and user-decision has been proposed. It can be applied in a garment recommendation algorithm both online and offline, leading to an increase in garment purchase. Also, it can be added to the magic mirror presented by Osaka´s Digital Fashion Co., which could collect geometry information of key body parts and show the try-on effects. The algorithm uses the garment stress of key body parts as a measurement of comfort level. And with the combination of moving range chart, maximum likelihood estimation and feedback, it is able to learn from records and give personalized recommendation.