
The advanced annotation platform for 3D sensor data.
Cuboid
Segmentation
Develop highly accurate perception models to locate and identify various objects, understand relationships between objects, predict behavior and more. Scale 3D Sensor Fusion supports a range of use cases from autonomous vehicles to robotics to AR/VR.
Develop 3D detection and tracking models with cuboid or segmentation annotation. The Dependent Tasks API can also be used to label some parts of tasks with cuboids (e.g. vehicles) and others with segmentation (e.g. vegetation) to leverage the benefits of both annotation types.
Develop prediction and planning models using attributes. Cuboid attributes for behavior and intent include gaze detection and turn signals. We also offer linked instance IDs describing the same object for long tasks, and linking of related objects (e.g. a vehicle and trailer).
Estimate the geometric structure of lanes and boundaries by using 3D Sensor Fusion segmentation annotation with intensity as an optional input parameter for greater accuracy. Talk To Us if you need line & spline or polygon annotation in 3D.
Build models you can trust while maximizing operational efficiency and reducing the cost of ML projects.
Cuboid
Segmentation
"Label all cars, pedestrians, and cyclists in each frame."
client.createLidarAnnotationTask({
instruction: 'Label all cars, pedestrians, and cyclists in each frame.',
labels: ['car', 'pedestrian', 'cyclist'],
meters_per_unit: 2.3,
max_distance_meters: 30
}, (err, task) => {
// do something with task
});
Receive large volumes of high-quality training data. Machine learning-powered pre-labeling and active tooling such as object tracking and automatic error detection ensures high quality at large volumes.
Have confidence in the quality of data. Quality assurance systems monitor and prevent errors, and varying levels of human review and consensus are provided according to customer requirements.
Confidently experiment with and deploy different sensors for 3D perception. Scale 3D Sensor Fusion supports all major sensors (LiDAR and RADAR) as well as different types of cameras (e.g. fisheye or panoramic).
Combine different annotation types in a singular task with the Dependent Tasks API. Label some objects with cuboids (e.g. vehicles) and others (e.g. vegetation) with segmentation to enhance object detection.
Annotate longer 3D scenes. An advanced stitching algorithm means no limit on scene length and linked instance IDs enables tracking accuracy even for objects that leave the scene for periods of time.
Gather metadata on annotated objects to understand correlations between objects (e.g. vehicle and trailer) and develop prediction and planning models. Attributes for intent include gaze detection and turn signals.
Get started today with on-demand, or chat with us about an enterprise plan.
Enterprise-grade SLAs include task completion times and tasks can be rapidly scaled up and down to meet your requirements.
Each enterprise customer is paired with a dedicated engagement manager who will ensure smooth on-boarding and continued data delivery.
Enterprise engagements provide upfront and volume-based discounts, and is the most cost-effective solution for high-quality labels. Plus with Scale AI, there are no platform fees.
ML-accelerated, human-in-the-loop data annotation for industry-leading quality.
3D Sensor Fusion tasks submitted to the platform are first pre-labeled by our proprietary ML-model, then manually reviewed by highly trained workers depending on the ML model confidence scores. All tasks receive additional layers of both human and ML-driven checks.
The resulting accuracy is consistently higher than what a human or synthetic labeling approach can achieve independently.
Scale 3D Sensor Fusion is trusted by leading machine learning teams to develop more accurate models.
Scale is a critical part of our stack, providing both software & trained resources to label our LIDAR and camera data with extremely high quality. When the team comes to me with new projects, I’m always confident that Scale will meet our needs & timelines.
Brandon Moak
CTO, Embark
Scale has a team of very talented engineers who really understand the problems in training machine learning models, of data annotation, and of data quality.
Oliver Cameron
Cofounder, Voyage