Case Study: Glacier Robotics labels 80,000 images in 10 days with AppRocket's AI data labeling pipeline

Glacier Robotics labels 80,000 images in 10 days with AppRocket

The customer, Glacier Robotics, faced the challenge of needing to label 80,000 recyclable product images within a strict two-week deadline to train its waste-sorting robots. Any compromise on speed or labeling quality would risk missing the deadline or degrading the robots' sorting accuracy in production. They engaged AppRocket to address this high-stakes data labeling challenge.

AppRocket built a streamlined data labeling workflow on AWS EC2 infrastructure, which included a dedicated ML verification model to flag suspect labels for human review and integrated QA protocols. This solution delivered all 80,000 verified images in just 10 days, four days ahead of schedule. The high-quality dataset enabled Glacier Robotics to successfully deploy its sorting robots across 10 pilot locations and two large-scale commercial sites.

Glacier Robotics