waymo-research/waymo-open-dataset resource

Health v2 · maintenance only

Flags: no_license

Adoption not part of the score

3398 stars · 699 forks observed · 2026-08-28

What it is AI-extracted, prompt v1, taxonomy v1, 2026-08-29, confidence not recorded

The Waymo Open Dataset is a large-scale public collection of autonomous driving datasets (Perception, Motion, and End-to-End Driving) accompanied by an Apache-licensed Python/TensorFlow codebase with dataset format definitions, evaluation metrics, and helper functions. It supports research challenges in perception, motion forecasting, sim agents, scenario generation, and vision-based end-to-end driving.

Use cases

• train 3D object detection models on lidar point clouds
• benchmark motion forecasting models against a leaderboard
• evaluate camera segmentation and panoptic segmentation models
• build end-to-end driving models from camera data and routing commands
• generate simulated agent behavior for scenario generation research
• reconstruct 3D object shapes and NeRFs from multi-sensor data

When to choose

• you need large-scale, high-quality labeled autonomous driving sensor data for research
• you want standardized evaluation metrics for perception or motion prediction tasks
• you are participating in Waymo Open Dataset Challenges

When to avoid

• you need a permissively licensed dataset - the data itself has non-commercial restrictions
• you want to evaluate real-life vehicle performance or safety
• you need a lightweight dataset without large download and compute requirements

Facets

dataset · maturity active

machine-learning computer-vision benchmarking data-science autonomous-vehicles machine-learning computer-vision simulation python autonomous-driving lidar point-cloud motion-forecasting tensorflow evaluation-metrics non-commercial-license linux gpu

Member repositories

For agents

markdown · JSON · MCP: product_card(name="waymo-research/waymo-open-dataset")