LIDAR + CAMERA + RADAR + IMU

Sensor fusion annotation,time-synced across modalities.

Multi-modal annotation with calibration-verified cross-modality consistency. Single 3D label projects to every camera, every radar return, every thermal frame.

  • Cross-modal consistency
  • PTP-synchronized
Scope a Data Project See sample annotations
FUSION LiDAR x Camera x Radar

Schematic. Production runs against your raw streams with calibration verified across all sensors.

PROCUREMENT READINESS

Compliance posture for multi-modal perception data.

Article 10 + ISO 21448 + UN R157 alignment for AD-grade fusion. GDPR Article 9 where biometric camera data is present.

Per-delivery artefact pack

Annotation guideline (versioned)
Multi-sensor calibration verification log
Cross-modal projection consistency report
Per-class metrics per modality
Records of processing (Article 30)
Signed DPA + sub-processor list

EU AI Act Article 10

Data governance for high-risk multi-modal perception. Annotation provenance per sensor, bias-examination notes per modality.

GDPR Articles 7, 9, 28

Per-contributor consent. Article 9 biometric category where camera data identifies persons. Processor agreement. 30-day erasure SLA.

ISO 21448 + UN R157

SOTIF data-governance aligned. Multi-sensor edge-case taxonomy. EEA-resident processing outside US CLOUD Act reach.

Request a Procurement Readiness Brief →

We map the evidence package to your data, risk class, and deployment environment.

FUSION PATTERNS

Five sensor-fusion patterns.

What you procure depends on what your perception stack consumes. The double-spend failure mode is paying for separate per-sensor annotation when fusion-aligned labels would have shipped.

LiDAR + Camera

most common stack

Spatial geometry from LiDAR, semantics from camera. Single 3D box projects to all camera views via verified extrinsics.

USED FOR AD/ADAS perception baseline, robotics scene understanding

LiDAR + Camera + Radar

full AD stack

3D geometry + semantics + Doppler velocity. Weather-robust. Required for production AD beyond L2.

USED FOR Production AD, weather-robust, velocity-aware

LiDAR + IMU + GPS

geo-localized

Sub-metre geo-localization via SLAM + GNSS fusion. Survey-grade output with georeferenced point clouds.

USED FOR Surveying, mapping, mobile robotics SLAM

Camera + Thermal

low-light + night

RGB + long-wave thermal for low-light surveillance, automotive night vision, search & rescue.

USED FOR Surveillance, search & rescue, night automotive

Multi-camera array

360deg coverage

Synchronized multi-camera with cross-camera ID continuity. 360deg coverage for surround perception.

USED FOR Surround perception, sports broadcast, surveillance

WHERE FUSION IS PROCURED

Five verticals where multi-modal annotation is the only viable path.

Per-sensor annotation cannot resolve cross-modal disagreement. Fusion-aligned labels do.

Automotive AD / ADAS

Full perception stack: LiDAR + camera + radar + IMU + GPS. nuScenes / Waymo taxonomy. UN R157 + ISO 21448.

nuScenesALKSSOTIF

Robotics + AMR

LiDAR + camera + IMU for SLAM. Indoor + outdoor, dynamic environments. Grasp + navigable + obstacle.

SLAMGraspAMR

Surveillance + security

Multi-camera + thermal + radar. Cross-camera re-ID. Article 35 DPIA mandatory. Day + night coverage.

ReIDThermalDPIA

Surveying + mapping

LiDAR + IMU + GPS + camera. Geo-referenced delivery. LAS / LAZ + georeferencing checks. BIM-ready.

GeoIMULAS/LAZ

Defense + dual-use

Multi-spectral, thermal, radar, LiDAR fusion. Export-control aware. EEA-resident processing. Customer security clearance handling.

Multi-spectralExportEEA
Discuss your sensor mix

Or describe the sensors and YPAI will scope projection policy, kappa target, and delivery format.

HOW WE LABEL

Every project clears the same six gates.

Calibration verified BEFORE annotation. Cross-modal consistency checked at every stage. PTP-sync verified at delivery.

01

Schema + multi-sensor calibration

Class taxonomy locked. All sensor calibration data (per-sensor intrinsic, all extrinsics, PTP-sync) verified before annotation.

Deliverable: Schema + calibration verification report

02

Annotation guideline + projection policy

Class definitions with per-sensor visibility rules. Cross-modal projection policy (which sensor is primary, which corrects).

Deliverable: Annotation guideline document

03

Calibration round

Pilot batch on shared subset. Cross-modal IoU between annotators. Projection-error verification. Refine guidelines on findings.

Deliverable: Cross-modal IoU report

04

IAA gate

Production starts only when calibration clears cross-modal IoU 0.6+ and projection error under 1px at standard range.

Deliverable: Gate-pass attestation

05

Production with multi-modal viewer

Single 3D label drawn in primary sensor, projected to all secondaries. Annotator validates cross-modality. Single-tenant proprietary viewer.

Deliverable: Multi-modal annotations

06

QA + cross-modal audit + delivery

Per-class IoU per modality. Cross-modal projection error report. Time-sync drift audit. Article 30 records, signed DPA.

Deliverable: Final delivery + multi-modal metrics

Six gates. One trail of evidence. Calibration verified before AND after annotation.

Start the scoping process →

WHAT WE DELIVER

Hover any view to see the annotation.

Schematic previews. Production work runs against your raw streams, multi-camera, calibration-verified.

HOVER reveal annotation

LiDAR + Camera (AD)

Vehicle, pedestrian, cyclist, lane

HOVER reveal annotation

LiDAR + Camera + Radar

Vehicle with velocity, weather-robust

HOVER reveal annotation

LiDAR + IMU + GPS (SLAM)

Map, trajectory, landmark

HOVER reveal annotation

Camera + Thermal (night)

Person, vehicle, animal, hot-spot

HOVER reveal annotation

Multi-camera surround

Cross-camera handoff, full 360

HOVER reveal annotation

Multi-spectral aerial

Vegetation, water, structure, anomaly

PUBLIC BENCHMARK COVERAGE

Taxonomy-aligned across the multi-modal corpora.

Multi-sensor public benchmarks are rare and mostly research-licensed. Production work runs against your own streams under engagement DPA.

nuScenes

Research only

Multi-modal AD

Samples
40,000 scans, 6 cams + 5 radars + LiDAR
Schema
nuScenes JSON
License
Non-commercial research
Taxonomy reference

Waymo Open

Research only

Multi-sensor AD

Samples
1,150 scenes, 5 LiDARs + 5 cams
Schema
TFRecord, proto
License
Waymo dataset license
Evaluation reference

Argoverse 2

Mixed

Forecasting

Samples
1,000 scenarios, 7 cams + LiDAR
Schema
AV2 SDK
License
CC BY-NC-SA 4.0
Evaluation reference

A2D2

Commercial

Automotive multi-modal

Samples
41,277 frames, 6 cams + 5 LiDARs
Schema
PNG + JSON
License
CC BY-ND 4.0
Evaluation reference

KITTI-360

Research only

Multi-sensor urban

Samples
320km driving, 360deg sensors
Schema
KITTI format
License
CC BY-NC-SA
Taxonomy reference

Pandaset

Research only

AD multi-modal

Samples
103 scenes, LiDAR + 6 cams
Schema
Pandaset SDK
License
CC BY-NC-SA 4.0
Evaluation reference

Multi-sensor licensing is the binding constraint for commercial training. Verify per engagement.

WHAT YOU RECEIVE

Every delivery ships with the cross-modal artefact pack.

The records a regulated multi-modal perception buyer expects.

Multi-sensor calibration verification.

Intrinsic + extrinsic verification per sensor. PTP-sync timestamp drift report. Cross-modal projection error matrix. Calibration log preserved per delivery.

Cross-modal projection consistency.

3D box projected to every camera, every radar return, every thermal frame. Per-class projection error in pixels. Time-offset audit. Velocity delta where radar present.

Per-modality + cross-modality metrics.

Per-class IoU per modality. Cross-modal Hungarian matching accuracy. Identity preservation across sensor handoff. Per-distance and per-modality breakdown.

Time-sync drift audit.

PTP-sync drift over the recording. Frame-pair correspondence. Interpolation policy for sensors at different rates. Drift threshold flags.

Records of processing, DPA, and sub-processor list.

Article 30 records of processing, signed Article 28 DPA, lawful-basis documentation, 30-day erasure SLA, and full sub-processor list per modality.

START A PROJECT

Brief us. We reply within one business day.

Short brief now, deeper scoping in the reply.

Capability lanes (NER, RLHF, etc.), languages, volume, regulatory context.

QUESTIONS BUYERS ACTUALLY ASK

Frequently asked questions

Per-sensor intrinsic parameters, all pairwise extrinsics (LiDAR-camera, LiDAR-radar, camera-camera), PTP-synchronized timestamps with sub-millisecond skew. Calibration verified before annotation; drift checks across the recording.

Documented projection policy decides which sensor is primary for each class. Annotator validates cross-modality before each label is finalized. Cross-modal disagreement flagged in the delivery report.

Yes. Thermal IR (long-wave), thermal NIR, multi-spectral satellite / drone bands. Calibration-verified across modalities. Class taxonomy adapts (thermal-only classes like hot-spot supported).

EEA-resident. Norwegian company, EEA contributor network, EEA infrastructure. 30-day GDPR Article 17 erasure SLA. Outside US CLOUD Act reach. Defense / dual-use engagements supported with additional access controls.

nuScenes JSON, KITTI-360 format, AV2 SDK, custom Protocol Buffers. Per-modality output plus cross-modal manifest. Time-aligned multi-sensor packages.

GDPR-Native EU AI Act Article 10 EEA Operations Consent Evidence

Brief us on your multi-modal perception project.

One business day reply. NDA on request. DPA included.

Cross-modal consistency PTP-synchronized delivery
Or connect on LinkedIn →

Your information is never shared. We respond with the next scoping step.