Machine Learning for Retail

Online fashion retailer Zalando has teamed up with Google, using machine learning to create a new virtual fashion experience letting shoppers create their own designs in 3D. The initiative called Project Muze uses a neural network trained with design preferences such as colours, textures and imagery. By answering a few questions such as mood and style the shopper will generate a design based on the answers. This together with data from the Google Fashion Trends Report and top trending objects from Zalando’s online platform.

The project taps in to the trend of using machine learning for retail to create a personalized experience using Google’s open-source platform TensorFlow combined with a set of aesthetic parameters. It launched at trend show Bread&butter in Berlin earlier this fall where a few handpicked designs were highlighted. The chosen pieces were later to be translated into real garments. Considering the visual experience the project clearly shows a big gap between digital and physical dimensions yet to be filled. However the concept of combining algorithms with 3D product interaction suggests a potential future where shoppers can virtually create their own products based on personal data.

Read more about 3D avatars and CGI in fashion