Case Study: Vida reduces false acceptance rate from 6% to 1% with Encord

Reducing Model False Positive Rate from 6% to 1% with Vida ID

Vida, a full-service verified digital identity platform operating in Southeast Asia, faced a challenge in training accurate facial recognition models. Most open-source datasets were not reflective of the region's demographics, leading to poor model performance and a high false acceptance rate of 6%. Vida needed a platform from a vendor like Encord to manage its large labeling team, annotate thousands of sensitive images, and maintain strict data privacy controls.

By implementing Encord's platform, Vida gained the tools to oversee its 20-person labeling team and ensure high-quality annotations. This enabled them to build new, representative datasets. The solution from Encord significantly improved dataset quality, which resulted in training new models that reduced the false acceptance rate from 6% to just 1%, greatly enhancing security for their customers.

Vida

Jeffrey Siaw

VP of Data Science