Project fitting fashion
The project “fitting fashion using machine learning” was initiated by the Ambient Intelligence research group of Saxion University. This research group focuses on finding solutions for smart environments. It focuses specifically on safety and sports from a technical perspective.
In the past couple of years, the online ordering of clothing has gained popularity. As online shopping does not have the convenience of being able to try out clothing before the purchase is made, it has caused many consumers to order the clothing they want in multiple sizes, as to increase the chance of receiving a fitting garment. The remaining sizes are sent back by the consumer, which generates extra labour costs for the store that handled the initial purchase. If the store could somehow know an estimate of the consumer’s size, the amount of clothing that would have to be shipped could be severely reduced, causing labour costs to decrease.
For this particular project, the Ambient Intelligence research group partnered up with a national consortium that focuses on finding solutions for the same problem. At the time of writing, the consortium consists of four universities and twenty-eight companies, which all share their gained knowledge. The overarching project was initiated by the Amsterdam University of Applied Sciences in October 2016 and is expected to end around the 30th of September, 2018.
The main goal of this project is to reduce the number of returns and thus the costs for the online clothing stores. To accomplish this, many researchers and stores are working on solutions for obtaining a customer’s accurate body dimensions to enable the customer to pick the clothing size that fits them best.
Even though some of these solutions are quite successful in terms of creating accurate 3D models, most of the solutions require a customer to perform various difficult tasks, such as having a full-body scan done, or taking a photograph using a special camera lens. Because most customers do not wish to put in this much effort, especially since the aforementioned solutions would see very small/no returns to the expended effort for the customer, these solutions are generally not viable to be put into production.
This project attempts to solve the issue of requiring too much effort from the customer. In the proposed solution, the customer would only be required to take a single photo of themselves and, with the help of machine learning, a 3D model of the customer would be created based on said photo which could aid them in ordering the properly sized clothing in the first place.