Partners stitch up online fashion shopping

Partners stitch up online fashion shopping

- in Online shopping
Comments Off on Partners stitch up online fashion shopping

Image result for Partners stitch up online fashion shopping

Polytechnic University and the Alibaba Group have stitched together a deal to create the first-ever database that leads artificial intelligence to learn about fashion.

Starting out by teaching machines how to differentiate between high-collar knitwear and V necks, the database will improve searches on online shopping platforms to help customers find the right clothing.

The project is a collaboration between PolyU Institute of Textiles and Clothing and the vision and beauty team at Alibaba, which specializes in looks and applications.

Jia Menglei, a senior engineer in the Alibaba team, said: “There is a huge potential for AI applications in the fashion industry. In order for AI to understand fashion, which can be very subjective, we need to turn fashion knowledge and experience into language machines can understand.”

Currently, fashion image search technology on online shopping platforms is based on a whole fashion image. With that, people make use of one particular image to search for a match.

But if a potential buyer is interested in particular fashion attributes, such as specific collar or sleeve types, current technology cannot fulfill their needs.

This is where collaboration clicked into the picture.

PolyU and Alibaba built a database named FashionAI Dataset for systematic analysis and labeling of images.

The images are collated by characteristics, including sleeve length, collar type and skirt style. They are also categorized by “key points,” which means neck, chest, waist, hip, wrist and toe measures.

A combination of the two determines product category and style.

“Transforming fashion knowledge into determination of fashion-related attributes and item categorization is a very complicated and challenging task,” said Calvin Wong Wai-keung , associate head of the institute. “It is the most fundamental task in deep learning applications.”

Once the wrinkles are ironed out, customers will be able to upload a picture, single out a specific feature then check the AI racks for an item.