Case Study: Zefr doubles model output with Voxel51's FiftyOne

A Voxel51 Case Study

Preview of the Zefr Case Study

Zefr doubles model output per quarter with Voxel51

Zefr, a leading technology provider for contextual advertising and brand safety, faced significant challenges scaling its computer vision operations. The company needed to process hundreds of millions of social media impressions daily to develop classifiers that detect unsafe content, but its legacy tools resulted in slow data discovery, low annotation throughput, and a lack of dataset visualization. To overcome these hurdles, Zefr turned to Voxel51 and implemented its FiftyOne platform.

Voxel51's solution provided Zefr with a unified visual data pipeline that integrated vector search, bulk annotation, and comprehensive dataset evaluation. By using FiftyOne as a central hub connected to their Qdrant database, Zefr scaled to manage over half a billion data samples. This enabled them to double the number of classification models shipped per quarter, dramatically increase annotation velocity, and maintain a 99%+ brand safety rating for their clients on major social platforms.


View this case study…

Voxel51

52 Case Studies