We all have doubts when choosing a size when shopping online. We also do not know how the clothes will fit our figure. Of course, clothing manufacturers try to include as much information as possible in the product description, but usually, it is only the length of the waist, chest, and pants. And even with such a description, we need to know our real dimensions. How many people know all this data about themselves?

As a result of a survey of more than two hundred people, we found out that no more than 30% of respondents know their real size. This figure is indirectly confirmed by the statistics of online returns of clothing purchases. More than 40% of the respondents were unable to make the necessary measurements on their own. 20% of respondents deliberately distorted measurements for personal reasons. In addition, we found that most respondents do not want to know the actual measurements of the waist, hips, etc., but want to buy clothes matching their ideas about how they look in the mirror. Thus, this is indeed a problem, although not entirely obvious to the average user, but quite serious for retailers or clothing manufacturers.

Based on our own unique developments in the area of video analytics and 3D modelling in conditions of insufficient information, we were able to develop algorithms for determining any real size of a person from a photograph or video stream, whether linear or volumetric dimensions, with a high degree of accuracy (error within (0.5 — 1 cm)) and based on these measurements build 3D models of the user’s body.

As a result, the pre-digitized clothes are virtually put on the 3D model and the user can see how the clothes will fit exactly on his figure, indicating possible problematic sites.