Viwalk - content-based clothing recommendation engine

  • Completed on:In progress
  • Technologies: Python / Caffe / Lasagne / ConvNets
  • Client: Addwalk

Project Info:

Addwalk is a company that focuses on the implementation of AI (or machine learning) in fashion advertising. Addwalk’s method augments photos for fashion-related websites by displaying banners containing relevant fashion items. Craftinity created a system that is capable of detecting a variety of types of clothing within the photos. After detection, the system places the query image into different categories based on the percentage of likeness, and then runs an AI-based search engine. This search engine then finds the most similar fashion items in a product database. The interface of the search engine is smooth and simple; the only requirement to use the system is a photo. The search engine itself returns highly accurate results, allowing the user to find similar or nearly identical fashion items from a wide range of vendors. The effectiveness of the search engine allows Addwalk to enable their fashion industry clients to monetize their content and provides a high quality experience for the users.

We stumbled upon Craftinity and hired them for our machine learning backend for our fashion advertising network. They implemented state-of-the-art deep learning methods in image recognition and we both came up with a lot of ingenious ideas. Both parties liked working with each other so much that we begun a strategic partnership to improve and expand technology and to conquer next markets very soon.

Frank Cieslik Addwalk