Case Study: Blue River Technology cuts signal-production time by 50% with Labelbox

Blue River Technology cuts signal-production time by 50% with Labelbox

Blue River Technology, a company building smart agriculture equipment, faced the high cost of producing the intricate image-segmentation signal needed to train its computer vision models. This signal was required to teach its See & Spray robots to distinguish weeds from crops. The challenge was the expensive and time-consuming process of manually creating this training data from scratch.

Using Labelbox's platform with its model-assisted feedback workflows, Blue River's model could generate initial masks for contributors to correct, focusing their expert efforts on problem areas. This solution from Labelbox cut the average processing time to create training data by over 50%, saving millions of dollars per year. The result allowed the team to focus on higher-value correction work and supported the technology's goal of significantly reducing herbicide use.

Blue River Technology