Starting as a project in my final year of university, this project combines several machine learning techniques to create a virtual stylist with the intention of estimating a user’s ‘style’ and helping them explore new ones.
The stylist program uses runway images sourced from Vogue Runway across multiple years and designers in order to capture a variety of potential styles. Features are then extracted in order to group the images. Prominent colours are extracted using pixel clustering and outfit silhouettes are found as Hu Moments – the date of the show the outfit is featured in is also included.
The stylist then explores this outfit cluster space, adjusting it’s search by branching out or exploiting known user preferences.
This project is still in development but if you wish to view the report, please email me at stephanieelawal@gmail.com!
the tech perspective 