Case Study: Scythe Robotics achieves faster, accurate off-road labeling with Segments.ai

Data labeling for Scythe Robotics’ precise off-road perception models

Scythe Robotics, a company developing autonomous commercial mowers, faced the challenge of creating precise perception models. This required a vast amount of accurately labeled 2D and 3D sensor data across more than 50 classes to distinguish between drivable areas and various obstacles on large lawns. To build these deep learning models, they partnered with the data labeling platform Segments.ai.

Segments.ai provided a solution where Scythe's team segmented drivable areas and obstacles on the platform for three years. They utilized ML-assisted tools like Superpixel and SAM to significantly speed up the labeling process while maintaining accuracy. The platform's Insights tab and integration capabilities allowed Scythe to efficiently manage their labeling workflow and billing. The partnership resulted in a streamlined, integrated labeling pipeline that was a game changer for Scythe's engineering team.

Scythe Robotics

Jose Rendon Leyva

Data Operations