Jumia Lens: A Visual Product Search Engine for Personalized Shopping on Jumia

A visual search engine, designed by Paul Okewunmi, for Jumia that lets users search for products by uploading an image. It uses computer vision to find similar or identical products within the store's inventory, saving users time and providing a more personalized shopping experience.

Abstract (from Lanfrica Talks)
In the dynamic realm of e-commerce, where convenience meets consumer desire, I'm thrilled to share my project. It introduces a novel visual search engine for Jumia, one of the leading online marketplaces in Africa. Imagine a shopping experience that transcends the barriers of text-based queries and embraces the sheer power of images. Picture a platform where you're no longer confined to describing what you seek; instead, you can effortlessly upload images and instantly unlock a world of exact or astonishingly similar products for purchase.

The project began with a comprehensive data collection phase, involving the scraping of 3650 product images across eight diverse categories. These categories encompass Computer Accessories, Mouse Varieties (ranging from gaming to office use), Wrist Watches (including fitness bands and chain watches), Office/Ergonomic/Gaming Chairs, and Backpacks.

This dataset lays the groundwork for my model, serving not only the Visual Image Search Engine but also contributing to datasets for broader e-commerce applications.