3D tracking
Combine LiDAR points and calibrated camera data with Find Similar and Auto-Tracking. Interpolate 3D cuboids between keyframes, then review and refine tracks in context.
Find SimilarAuto-Tracking3D interpolation
LIDAR & 3D ANNOTATION PLATFORM
LiDAR Annotation Platform Built for 3D Perception
Build reliable 3D training data with point-cloud labeling, camera-assisted tracking, shared ontologies, and governed quality control.
No credit card required
PRECISION ACROSS EVERY DIMENSION
Everything You Need for LiDAR Annotation
Label 3D objects and point-level segments, connect calibrated sensor views, and refine tracks in one workspace.
Precise 3D geometry and point labels
Fit cuboids in perspective and orthographic views. Paint point segments and label spatial boundaries with polygons, polylines, points, and spheres.
Point-level instance and semantic segmentation
Label point clouds by semantic class and separate individual objects. Refine point-level labels in 3D with calibrated camera context.
Nested ontologies and dynamic properties
Define classes with nested, conditional properties for 3D objects and scenes. Record changing states by frame and inspect their temporal ranges on the timeline.
Complete 3D annotation timelines
Review object tracks, visibility ranges, keyframes, and dynamic properties together. Navigate directly to the frame and object that need attention.
SENSOR CONTEXTSynchronized LiDAR + cameras
Calibrated multi-sensor views
Inspect LiDAR and radar point clouds alongside synchronized camera images. Review projected 3D shapes and linked camera boxes with the original sensor context.
Synchronized viewsCalibrated projections
Built for AI Data at Scale
Annotate complex LiDAR datasets faster with AI-assisted automation, scalable workflows, and lower operational costs.
15×
Faster LiDAR data annotation
Batch workflows, camera-assisted Find Similar and Auto-Tracking, and interpolation reduce repetitive manual LiDAR labeling.
85%
Automated LiDAR annotation
On average, 85% of LiDAR labels are pre-labeled automatically, then reviewed and refined by humans.
15×
Lower Training Data Costs
The cost per accepted label can be up to 15× lower as curation, annotation, and QA are automated.
NATIVE 3D LABELING TOOLS
LiDAR Annotation Types for Objects, Scenes, and Sequences
Choose the geometry and properties your perception task needs, then apply them consistently across every scene.
01
3D Cuboids
Label objects with oriented 3D boxes, precise dimensions, and rotation. Refine their fit in perspective, top, side, and rear views.
02
Point Segmentation
Paint point-level object and scene labels. Assign classes and properties while retaining the original point-cloud context.
03
3D Polygons
Outline spatial regions with 3D vertices and closed boundaries for consistent region annotation.
04
3D Spheres
Mark spherical regions with an editable centre and radius. Maintain class properties and temporal states.
05
3D Polylines
Trace curbs, barriers, and other open boundaries with connected 3D points.
06
3D Points
Mark precise spatial locations and landmarks with class labels and structured properties.
07
Linked Camera Boxes
Connect a 3D cuboid to its 2D box in a calibrated camera image and review both views together.
08
Object & Scene Properties
Apply reusable classifications and conditional attributes. Record frame-varying properties for objects and scenes.
09
Tracks & Keyframes
Maintain stable object identities, visibility ranges, and editable keyframes across a point-cloud sequence.
LIDAR ANNOTATION QUALITY ASSURANCE
Build quality into every 3D annotation.
Compare independent scene annotations, check labels against approved references, and resolve issues with the point cloud and frame context in view.
LiDAR Annotation Consensus
Compare independent annotations of the same LiDAR scene. Review disagreements in 3D geometry, classes, and object properties before approving labels.
Scene-level evidenceAgreement checksReview disagreements
Quality Gate (Honeypot)
Check scene annotations against approved reference labels hidden from annotators. Route results through pass, fail, or not-evaluated paths using configured criteria.
Validation & Issue Resolution
Find missing required properties and annotation issues. Assign corrections, inspect the affected object and frame, and resolve feedback before release.
LiDAR QA Analytics
Review benchmark results, consensus outcomes, open issues, and validation findings. Focus review on the scenes and objects that need attention.
Benchmark resultsReview outcomes
LIDAR DATASET CURATION & ANNOTATION WORKFLOWS
Curate LiDAR data and automate annotation workflows.
Search, version, and inspect LiDAR datasets, connect AI models, and move annotations through review and approval.
Dataset Versions
Create scene datasets, freeze selected versions, and attach them to projects without re-uploading. Keep data selection and releases traceable.
Search & Filter
Find scenes by folders, tags, classes, properties, assignees, and issue status. Build focused annotation queues from the data you need.
Embedding View
Visualize LiDAR datasets with your own embeddings to inspect clusters, outliers, and labeling issues before training.
Integrated Workflows
Build workflows that connect LiDAR annotation, review, and quality assurance in one continuous loop. Reduce handoffs and keep datasets moving from labeling to approval.
Bring Your LiDAR Model
Prepare reviewed 3D labels and versioned datasets for your LiDAR model pipeline. Talk to our team about your model integration requirements.
Solutions
LiDAR Annotation Across Industries and Use Cases
Build reviewed 3D training data for autonomous driving, robotics, warehouse perception, and infrastructure mapping.
Autonomous Driving & ADAS
Label road users, point-cloud scenes, and temporal tracks with calibrated camera context.
Explore autonomous driving
Robotics & Navigation
Annotate people, obstacles, and spatial boundaries for robot perception datasets.
Explore robotics
Warehouse Perception
Label forklifts, pallets, racks, and indoor scenes with consistent 3D geometry and properties.
Explore warehouse perception
Infrastructure Mapping
Annotate poles, barriers, structures, and corridor boundaries in point-cloud scans.
Explore infrastructure mapping
Questions, answered
LiDAR Annotation Platform FAQs
Answers about point-cloud formats, 3D labeling tools, calibrated camera views, tracking, quality assurance, and versioned datasets.
Talk with the Unitlab teamWhich LiDAR and point-cloud formats does Unitlab support?
+
Upload PCD, PLY, and BIN point clouds as individual files, ZIP scenes, or folders. Scene manifests can include multiple LiDAR or radar sensors, calibrated cameras, and frame poses. Unitlab also converts supported nuScenes recordings into scene inputs.
Read LiDAR data preparationWhich 3D annotation types can I use?
+
Unitlab supports oriented 3D cuboids, point segmentation, polylines, polygons, points, and spheres. Linked camera boxes connect to their 3D cuboids. Shared ontologies define classes, conditional properties, and scene-level labels.
Explore 3D annotation typesHow do calibrated LiDAR and camera views work together?
+
A scene keeps point clouds and camera images aligned by frame. Supplied calibration and sensor poses let annotators inspect projected 3D shapes, linked camera boxes, and point overlays in the original camera context.
Read camera calibration guideHow do Auto-Tracking, Find Similar, and interpolation work?
+
Find Similar uses a selected object in a calibrated camera image, then fits candidate cuboids to LiDAR points. Auto-Tracking follows camera evidence across frames and creates editable 3D keyframes while preserving manual corrections. Interpolation fills supported geometry between trusted keyframes, and batch view helps review cuboid drift.
Explore LiDAR tracking and interpolation Explore LiDAR Find SimilarCan LiDAR object properties change over time?
+
Yes. Define nested, conditional properties for objects and scenes, then set frame-varying values such as motion state or occlusion. The scene timeline displays object tracks, keyframes, visibility ranges, and property intervals together.
Explore dynamic propertiesHow is LiDAR annotation quality checked?
Consensus compares independent annotations of a LiDAR scene and opens disagreements for review. Quality Gate checks submissions against hidden approved references. Required-property validation and annotation issues support correction before release. Reviewers can inspect the relevant geometry and frame context.
Read LiDAR quality assurance guideHow can I curate and export LiDAR training datasets?
Organize scenes with folders and tags, filter by labels and review status, and freeze dataset versions. Export UUEF releases containing 3D annotations, temporal tracks, scene metadata, calibration, and point-label arrays for downstream training-data pipelines.
Explore LiDAR annotation exportsBUILD RELIABLE 3D TRAINING DATA
Build Production-Ready LiDAR Datasets with Unitlab
Bring point-cloud annotation, calibrated sensor context, quality assurance, and dataset operations into one governed workflow.
