News

Smart solution - Trying on your favourite clothes online

22.05.2018

Trying on your favourite clothes online? Third-year HBO-ICT students have developed a machine learning tool that allows customers to easily create a 3D image of themselves using their own photos.

Thew Saxion students who developed the online tool for trying on clothes online

Ordering clothes online, trying them on in the comfort of their own home and returning items that do not fit; in recent years a growing number of people have started buying their clothes this way in online shops. Convenience and flexibility are the main advantages. However, there are also drawbacks to this new shopping trend. Trying on clothes online is not possible. People who order their favourite shirt in three sizes, keep the shirt that fits best and return the rest, burdening both the webshop and the environment. Parcel delivery vehicles drive back and forth and the webshops often have to cover the costs. There is certainly room for improvement.

That is exactly what the third-year HBO-ICT students Remco Booij, Bodhi Mulders and Adrian Brezoi thought. They were assigned by the research group Ambient Intelligence to develop a smart solution to the 'fitting' problem. Project manager Remco Booij (21) explains: "Within our minor Immersive Media, we are trying to find out how we can use the Machine Learning technique to develop a solution te this problem. We have a solution in mind, but are currently investigating whether it can actually be used in practice." The students are developing a tool to give the online shopper better insight into his/her own clothing sizes.

Immersive Media

At Immersive Media, the line between the physical and the digital, simulated world is becoming increasingly thinner. Consider a product like Virtual Reality glasses, for example, which can make you feel like you are in a different world. Remco and his colleagues form an important link in a national project that started in October 2016, under the supervision of the Hogeschool van Amsterdam. In this project, four universities, including Saxion, and 28 companies have combined forces. Several online clothing stores are also cooperating in the project. There is clearly a shared interest.

"Our research is progressing well. We are still on track and discovered that the solution we have in mind actually has a real chance of success. The lecturers of Creative Media and Game Technologies have given us excellent support in getting to grips with the concept of Machine Learning," says Remco.

Machine learning

Is there an easy way to explain this technique? "What Machine Learning actually comes down to is 'training' an algorithm," says the project manager. "You have to look at it as a series of programmed actions. With Machine Learning an algorithm improves itself. It becomes smarter all the time, as it were, based on the new data it incorporates. In our case, this concerns a tool which can make a fairly accurate estimate of the ideal size of a clothing item that somebody is interested in by using his/her body measurements

The students designed a tool that allows the customer who shops online to easily create a 3D-image of himself, based on a number of photos that he/she uploads. This 3D-body image is a great help in determining the right clothing size. Remco: "A customer can project his/her desired clothing item on it and immediately see whether a certain sweater, t-shirt or pants would fit. On average, we can now accurately determine a customer’s measurements to within 1.5 centimetres Of course we want to improve this even further, although we sometimes manage to within 1 millimetre from the actual size."

When will it be used in actual practice?

We have to exercise some patience, though, before being able to use this Saxion solution in practice, according to Remco. "We are exploring all sorts of possible applications within the consortium. This will tell us what works best. I imagine that, once it has been further developed, our idea will first be tested in small webshops. If they incur fewer losses thanks to our solution, we can take it a step further and introduce it to the larger companies. So it will take a while before the consumer can start using the application, but that does not matter. It is an exciting process. Since the moment we first presented our ideas at Saxion and our lecturer enthusiastically exclaimed that he immediately wanted to call a newspaper, I felt that we were really on the right track."

The students maintain their own blog at  fittingfashion.ml