Products
Solutions
Retail & eCommerce
Defense
Logistics
Autonomous Vehicles
Robotics
AR/VR
Content & Language
Smart Port Lab
Federal LLMs
Resources
Company
Customers
See all customers
3D Sensor Fusion
Cuboid
Segmentation
Use Cases
Computer Vision
Develop highly accurate perception models to locate and identify various objects, understand relationships between objects, predict behavior and more.
Detection & Tracking
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.
Prediction & Planning
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).
Lane & Boundary Detection
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.
How it works
Easy to Start, Optimize and Scale
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.
1client.createLidarAnnotationTask({
2 instruction: 'Label all cars, pedestrians, and cyclists in each frame.',
3 labels: ['car', 'pedestrian', 'cyclist'],
4 meters_per_unit: 2.3,
5 max_distance_meters: 30
6}, (err, task) => {
7 // do something with task
8});
ML-Powered Data Labeling
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.
Automated Quality Pipeline
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.
Sensor Agnostic
Confidently experiment with and deploy different sensors for 3D perception. We support all major sensors (LiDAR and RADAR) as well as different types of cameras (e.g. fisheye or panoramic)
Comprehensive Label Support
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.
Infinitely Long Tasks (Beta)
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.
Attributes Support
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.
Quality Assurance
Best-In-Class Quality
ML-accelerated, human-in-the-loop data annotation for industry-leading quality.
Super Human 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.
