Case Study: Automotus achieves 20% higher mAP and 35% smaller datasets with Cord

Automotus boosts model performance by 20% with Cord

Automotus, a company developing AI-powered parking and traffic management solutions, faced the challenge of managing a large volume of visual data from hundreds of cameras. The high cost of labeling all this data was prohibitive, and the Automotus team needed a way to identify and label only the most relevant data to improve their models efficiently. They partnered with the vendor Encord to address this data pipeline challenge.

By implementing Encord's platform for automated data curation and AI-assisted labeling, Automotus gained the ability to visually inspect, query, and sort their datasets. This allowed them to eliminate low-quality data, which reduced the dataset size required for annotation by 35% and cut labeling costs by 33%. Most importantly, using Encord's tools led to a 20% improvement in their model's performance, enabling more accurate analytics for their clients.

Automotus