Facebook AI- Facebook is integrating a new AI software called the Universal Product Recognition Model that automatically identifies the consumer goods placed for sale to the Facebook marketplace. This is the first step towards a future where every product can be identified and potentially shopped for, as part of Facebook’s mission to create a ‘social-first’ shopping experience.
Fashion will be at the epicentre of the Universal Product Recognition Model as Facebook is trying to develop a virtual fashion stylist that can offer users personalised fashion advice to build thier wardrobes, along with daily clothing suggestions based on the weather and thier schedules.
However, the idea of an AI fashion assistant has been around for a while and has in fact, been tried and tested, with mixed results. One such example is the Style Check feature offered by Amazon’s device, Echo Look. Echo Look is an Alexa-enabled camera that has the ability to take pictures with just voice commands. The Style Check tool then uses machine learning and data provided by numerous fashion experts to provide advice and recommendations that are in line with current trends and user preferences. However, the product is little heard of in the present day.
Facebook claims that the features that set itself apart in this market are scope and accuracy. The company’s product recognition tool, called GrokNet, can identify up to a thousand different traits and characteristics in the products listed, that range from product brand, colours and size. GrokNet has already been implemented within the Facebook Marketplace, where it identifies the products listed for sale and generate short descriptions of the product.
GrokNet has been developed from the vast array of data already present within the Facebook Marketplace, with around 100 million images. Facebook says this data is vital in creating a machine vision system that can identify products in challenging lighting and from dodgy angles — a part of the online shopping experience that isn’t going away.