Self-Healing Product Catalogs: Drift-Aware Computer Vision for Retail & CPG on Databricks | Databricks

▤ Summary

Product catalogs constantly evolve and those changes directly affect model accuracy. Neurolabs runs image recognition across 20,000+ SKUs in 28 markets and generates millions of shelf crops each month, but only a few crops carry useful training signal. The selection and training loop now runs on Databricks: inference results land in Delta, each crop is scored for training value, and Unity Catalog governs which crops feed the next model while surfacing catalog-health metrics to CPG brands. The result is a self-healing product catalog and a computer vision model that improves with every shelf photo.

▦ Linked Companies

Neurolabs

$1M to $5M