Multimodal AI Search

Multimodal AI Search: Next-Gen Multimodal AI for Enhanced E-Commerce Discovery and Personalization is a groundbreaking hackathon project designed to revolutionise the online shopping experience. At its core, this project leverages advanced multimodal AI technology to understand user input and deliver highly accurate and personalised search results based on the e-commerce data In today's e-commerce landscape, owners often face the challenge of setting up data for their inventory to be able to provide a rich search experience for their users. Even if the inventory has a certain product but it's attributes are not listed properly it won't be listed by traditional search algorithms. Multimodal Search addresses this issue by auto-generating the metadata required for the products just based on the product image and then allowing the user to search based on the generated attributes thus eliminating the chance of a human error Behind the scenes, our AI algorithms enhance the product data and generate all the required attributes of the product and our search algorithm is not based on keyword matching but a hybrid mix of keyword and vector based search. This not only improves the accuracy of search results but also tailors product recommendations to individual users, enhancing the overall shopping experience. The obtained results are re-ranked to further refine the results and are evaluated finally using Trulens