Case Study: Tavisca achieves automated classification of 25M hotel images and 86% TOP-3 precision with Imagga

Tavisca - Customer Case Study

Tavisca, a travel technology company that powers thousands of travel websites, faced an unclassified dataset of more than 25 million hotel images aggregated from XML APIs, FTP feeds and other sources, causing duplicates, unsorted amenity and room photos, and inconsistent content across sites. To tackle this, Tavisca partnered with Imagga and used Imagga’s Categorization API and Image Categorization Training technology to build a custom hotel image classifier.

Working with Tavisca, Imagga trained a custom model (able to distinguish categories like balcony, lobby, living area, fitness, etc.) and provided a simple, easy-to-integrate API. Imagga’s solution automated classification of 25M images, achieved an 86% precision rate for top-3 predictions, and measurably improved real-time hotel mapping and the quality of photos served to millions of users monthly.

Tavisca

Vikram Sreekumar

Software Engineer