While autonomous vehicle companies are continuously improving and refining their AI/ML models, these models must represent real-world scenarios accurately, also known as ground truth.

iMerit has been working with a leading Robotaxi company, with more than 1200 employees, to improve their ground truth data for improved safety and control of autonomous vehicles.

Problem

Due to the high volume of data and complex annotation requirements, the Robotaxi company needed domain expertise and better quality control to ensure high-quality training datasets for the ML models.

Solution

iMerit team worked and reviewed different types of annotations for the Robotaxi company, such as full-scene semantic segmentation, mapping scenarios, and prediction work types, and shared expertise on QC for production data.

Results

Our workflow exceeded an accuracy threshold of 95% at scale. The team also shared recommendations for taxonomy upgrades, leading to a 250% improvement in annotation efficiency.

BOTTOM LINE IMPACT

95%

Annotation Accuracy

250%

Improved Efficiency

Over-achieving

Service Level Agreements