Case Study: Shotzr speeds up identifying images needing location metadata with Appen

How Shotzr used the Appen platform to improve the recommendation and search experience for their customers

Shotzr, a provider of contextual imagery for digital marketing with access to nearly 100M images, faced a scaling challenge: their in-house team could not keep up with labeling and location metadata needs (they had annotated about 20,000 images in 90 days). To improve search and recommendation relevance and avoid showing marketers hundreds of irrelevant results, Shotzr needed a way to quickly identify which images required specific location labels.

Shotzr turned to the Appen platform for high-quality training data to power ML models that automatically detect which images need location metadata. After the first Appen job they flagged over 17,000 images as not requiring additional labeling, expect to exclude roughly 61 million assets from location processing, and anticipate training 4x more classifiers than originally planned—freeing time to focus on images that benefit from location data and to build new automated labeling models.

Shotzr

Mark Lemmons

Co-Founder