Sama Annotate

3D Point Cloud Annotation

Sama provides unparalleled tools and expertise to deliver uncompromisingly high-quality ground truth annotations for your LiDAR, Radar, and other 3D data.

3D Pointcloud annotation

Trusted by the world’s most innovative companies

3D Point Cloud Tools

Our advanced 3D LiDAR & Radar annotation tools and workspace enhancements enable rapid labeling of your moving objects across multiple frames and speed your model to market.

Feed your perception models with an object’s 3D pose and attributes. Annotate objects like vehicles, vulnerable road users, traffic signs, traffic lights, etc.

  • One-click cuboid

  • Use interpolation between key frames for enhanced annotation efficiency

  • Use static and dynamic labels to grab object properties for your detection and prediction models

Feed your drivable area models with ground truth annotation of curbs, road pavement, and lane markers.

  • Custom intensity sliders to facilitate object identification on low intensity points

  • ML-ground detection for “Z” values attribution

Feed your drivable area models with ground truth annotation of traffic islands, over drivable objects, or any other object on the ground.

  • Custom intensity sliders to facilitate object identification on low intensity points

  • ML-ground detection for “Z” values attribution

Annotate objects in 3D point clouds and 2D RGB images with fused object ID to give you all the valuable information from multiple sensors.

  • Sensor fusion between multiple sensor types, synchronizing all frame sequences

  • Projection of 3D objects within the point cloud into 2D RGB images

3D Point Cloud Features

Supports pre-annotations, including manual or model-generated labels for enhancing speed or annotation, or validating model performance.

Supports multiple sensors with a fused sensor architecture, which syncs all the assets to enhance the data for annotation.

Projection of 3D objects within the point cloud into 2D RGB images.

World coordinate conversion. By providing the sensor’s pose, our platform will convert your point clouds from local coordinates to world coordinates.

Extending a classic morphological approach with semi-supervised learning for our automatic ground detection.

Color map visualization for intensity values and custom ranges for focusing the color map on lower-intensity objects.

95%+

Data accuracy — guaranteed — compared to 88.5% for leading competitors.

160M

Events tracked per month to improve our product and processes with A/B testing.

59K

Lives impacted to date thanks to our purpose-driven business model.

Why Choose Sama

Sama 3D point cloud annotation experts leverage our ML-powered platform to deliver labels with unmatched accuracy and speed.

Efficient Annotation Platform

ML Assisted Annotation powered by MICROMODEL™ technology helps labelers annotate 2-4x more efficiently, while predictably producing 94-98% IOU.

Full Spectrum Labeling

Custom classes, multiple attributes per instance, and hierarchical relationships are available to meet your needs.

Multi-Level Quality Control

AutoQA, Gold Tasks, and advanced scripting detect errors early in the pipeline. Our skilled QA team focuses on solving edge cases that can slow the path to production.

Analytics and Reporting

Get deep insights into your training data, including edge case alerts and analysis of the classes in your dataset to bring your AI models to production faster.

Related Resources

Building a Robust Automotive LiDAR Annotation Quality Rubric

10 Min Read

Autonomous Vehicles’ Impact on Cities with Lyft’s Sarah Barnes

34 Min Listen
traffic light

The Traffic Light Problem for Autonomous Vehicles

5 Min Read

High-Quality Training Data From
Start to Scale.