Acquiring Accurate Body Measurements on a Smartphone from Supplied Colored Garments for Online Apparel Purchasing Platforms and E-Retailers

Abstract

Return rates for e-retail fashion companies are significantly higher than in-store sales. Twenty to fifty percent of online clothing sales are returned. Apparel retailers are haunted by returns based on sizing issues, with $62.4 billion in returns attributed to poor choices by the consumer in the USA. However, over the next ten years online sales are predicted to double, compounding the problem exponentially. Garment sizing and knowing your correct size for a particular garment or brand while online shopping is part of the problem. It is the combinations of body measurements that determine sizing and sizing labels in clothing not usually one measurement. Most consumers don’t know their body measurements when attempting to determine the size of a garment that they would like to purchase when shopping online and can have significant difficulty attempting to take their own measurements. This can lead to frustration and an incomplete sale or shopping cart abandonment. Many customers even resort buying a garment in two or more sizes and return the ones that do not fit, as they do not want to waste their time trying to determine which would be a perfect size. This adds to cost and waste affecting profitability. By the time these garments are returned to the vendor or manufacture they are out of season and usually not resalable at the original price because of the time lag and subsequent repackaging problems. This research focuses on creating a fast-personal garment apparatus, system, and method for measuring body dimensions extracted from two-dimensional (2D) images captured by a consumer. Measurements of the individual are taken from captured pictures or photographs from their smart phones while wearing one or more coded dimensioning garments that have markings at specific locations that can be aligned with characteristic body features and key measurement areas. Computer vision is used to track these markings and extract key body dimensions. TensorFlow, a machine learning software application, is incorporated for object detection can be used to recognize colors and patterns on the garment allowing the garment to act as a measurement device for the body. The extracted dimensions could further used to predict additional body information such as; size growth and fit information, for example with fitness apps and workout appeal, or simply predicting children’s wear and maternity wear needs as the body grows.