Defence & National Security

Specialist data and evaluation for defence and national-security AI.

EO/IR imagery, remote sensing, autonomy, sensor fusion, speech and expert evaluation. Produced by specialist panels assembled for your project, reviewed until it meets your acceptance line, and delivered from Norway.

Scope a project See a hard frame reviewed

The capability field

Six data families. One delivery system.

Every project enters through the data it needs. Choose a family to see the tasks, the modalities and the specialists behind it, or follow one of five worked paths through the same system.

EO/IR, image and video

What the camera and the thermal sensor recorded, made trainable frame by frame.

Tasks · 8

  • Detection and bounding boxes
  • Tracking across frames
  • Semantic, instance and video segmentation
  • Small and low-contrast targets
  • Low light, glare and occlusion
  • Multi-camera review and re-identification
  • Model-assisted labels, human-verified
  • Agreement analysis and adjudication

Modalities

  • Thermal video
  • EO video
  • Still imagery

Specialist panel

  • EO/IR and thermal-imaging specialists
  • UAS operators
  • QA reviewers

Geospatial and remote sensing

Aerial and satellite imagery labelled by people who read terrain for a living.

Tasks · 7

  • Aerial imagery annotation
  • Satellite tile annotation
  • Human-only aerial segmentation
  • Infrastructure features
  • Map-anchored sensor traces
  • Schema alignment to your ontology
  • Remote-sensing review

Modalities

  • Aerial imagery
  • Satellite tiles
  • Geospatial traces

Specialist panel

  • GIS and geospatial professionals
  • Photogrammetry and remote-sensing specialists

Robotics and autonomous systems

Perception and behaviour data for systems that move on their own.

Tasks · 6

  • Perception datasets
  • Operator and task demonstrations
  • Trajectories and action labels
  • Outdoor navigation capture, stereo and RTK GPS
  • Synchronised sensor and video
  • Vision-language-action evaluation

Modalities

  • Stereo video
  • Egocentric video
  • Synchronised sensors

Specialist panel

  • Robotics engineers and operators
  • Autonomous-systems specialists

Sensor and multimodal

Every sensor on one clock, calibrated, aligned and checked before it is labelled.

Tasks · 6

  • LiDAR and point-cloud annotation
  • Radar, fused with camera and LiDAR
  • Depth and IMU
  • Timestamp alignment
  • Calibration and cross-sensor projection
  • Sensor QA and metadata

Modalities

  • LiDAR
  • Radar
  • Depth and IMU
  • Thermal

Specialist panel

  • Sensor specialists
  • Aerospace engineers and avionics specialists

Speech, audio and language

Speech and text in the languages, dialects and noise your system will meet.

Tasks · 7

  • Multilingual speech capture
  • Noisy and far-field speech
  • Low-resource languages, dialects and accents
  • Transcription, timestamps and diarisation
  • Linguistic QA
  • ASR and TTS evaluation
  • Translation and terminology

Modalities

  • Speech
  • Environmental audio
  • Text

Specialist panel

  • Technical linguists
  • Native-speaker reviewers

Expert evaluation

Specialist judgement on what a model produced, recorded against a rubric.

Tasks · 6

  • Rubric and benchmark design
  • Dual review and adjudication
  • Expert grading of model output
  • Red teaming
  • Robustness under domain shift
  • Media integrity and synthetic media

Modalities

  • Model output
  • Imagery
  • Speech
  • Text

Specialist panel

  • Engineers and domain experts
  • Model evaluators
  • QA reviewers and adjudicators

Five worked paths

  1. 01 · Family EO/IR, image and video
  2. 02 · Modality Thermal video, 30 fps, moving camera
  3. 03 · Task Boxes on small targets, tracked across frames
  4. 04 · Specialist panel EO/IR-literate annotators and QA reviewers
  5. 05 · Review Dual review on hard frames, HOTA and IDF1 per delivery
  6. 06 · Delivery EEA processing, YPAI-managed or in your environment

Specialist data operations

The panel is built for the project.

We source, qualify, calibrate and operate specialist panels to your specification. The people who label a thermal frame, a satellite tile or a dialect transcript are chosen for that work and tested on your data before production starts.

  1. 01

    Source

    Profiles drawn from our network and specialist channels to the project's specification.

  2. 02

    Qualify

    A test on your own sample. Only those who pass it work on the project.

  3. 03

    Calibrate

    Gold sets and agreement scoring with Cohen's and Fleiss' kappa, before and during production.

  4. 04

    Operate

    Production, dual review, adjudication and rework, with every decision logged.

Profiles we assemble · EO/IR, image and video

Air and sensing

  • UAS and drone operators
  • Aerospace engineers
  • Avionics specialists
  • EO/IR and thermal-imaging specialists
  • Sensor specialists

Ground and geospatial

  • GIS and geospatial professionals
  • Photogrammetry specialists
  • Remote-sensing specialists

Robotics and autonomy

  • Robotics engineers
  • Robotics operators
  • Autonomous-systems specialists

Maritime

  • Maritime specialists

Language

  • Technical linguists
  • Native-speaker reviewers
  • Language leads

Engineering and review

  • Engineers and domain experts
  • QA reviewers and adjudicators
  • Model evaluators
  • 210,000+ contributors
  • 50+ countries
  • 150+ languages

Network reach. Every panel is qualified against your specification before production.

Difficult targets

When the object is only a few pixels, annotation becomes a review problem.

A target of a dozen pixels in a moving frame is easy to miss and easy to box wrongly. The work is done at full resolution, checked across frames and ruled on by a reviewer before it counts.

Frame 00417

Raw frame The sensor's frame as recorded. The target is a few pixels against the sky.

Full-resolution zoom The annotator works at the native pixel grid, not the preview.

Frame-level box A tight box on the target, drawn to the project's box rules.

Temporal consistency The box is checked against the frames before and after it. A jump is flagged.

Reviewer decision A second person accepts the frame or returns it with a reason.

Accepted The frame enters the dataset with its annotator, reviewer and version.

Reviewer

Accepted by the reviewer Logged with annotator, reviewer and dataset version. Returned to the annotator Logged with annotator, reviewer and dataset version.

Production scale

Frame rate

36,000 frames a week

156,000 frames a month

Every one of them is drawn and checked at full resolution.

The frame, the target and the readings are drawn by this page. Class list, box rules and acceptance thresholds are set in the project specification.

From specification to accepted data

Eight stations. One of them is held by a person.

Every project runs the same line. Automated checks catch what a machine can catch. A specialist rules on the rest, and nothing is accepted past that point without a decision.

  1. 01

    Specification

    Classes, rules, formats, volumes and the acceptance criteria, written with you.

  2. 02

    Qualification

    Sources and specialists tested against the specification.

  3. 03

    Collection and annotation

    Captured or labelled by the qualified panel, model-assisted where it helps.

  4. 04

    Automated QC

    Format, completeness, sensor timing and consistency checks on every item.

  5. 05

    Specialist review

    A qualified reviewer rules on each item or sample the plan defines.

  6. 06

    Adjudication and rework

    Disagreements resolved, returned items fixed and checked again.

  7. 07

    Accepted dataset

    Versioned, with the evidence for why each item passed.

  8. 08

    Controlled delivery

    Delivered into the environment the project specifies, with a manifest.

Modalities

What the sensors record, and what we do with it.

Modality Work 09

  1. 01 Thermal and EO imagery and video Boxes, tracks, segmentation, small-target review
  2. 02 Aerial and satellite imagery Geospatial annotation, human-only aerial segmentation
  3. 03 LiDAR and point clouds 3D boxes, segmentation, tracking, calibration checks
  4. 04 Radar Fused with camera and LiDAR, cross-sensor projection
  5. 05 Depth and IMU Capture, timestamp alignment, sensor QA
  6. 06 Stereo video and RTK GPS Outdoor navigation capture for ground autonomy
  7. 07 Speech and audio 48 kHz / 24-bit capture, transcription, diarisation, sound events
  8. 08 Text and documents Extraction, classification, terminology, translation
  9. 09 Synchronised multimodal One clock, calibration data, multi-sensor alignment

Delivery and data handling

Your data runs where your rules say it runs.

Three delivery models on one method. Choose where the data is stored, where the work is done and where the result is handed over.

Data stored
YPAI infrastructure in the EEA
Work performed
Qualified panel on YPAI tooling
Review
YPAI review console
Handover
Versioned delivery with manifest
Jurisdiction
Norway · GDPR
Processing
EEA by default
Contributors
Identity-verified
Isolation
Data isolation, client-specific environment
Control
Change control, versioned manifests
Erasure
30-day end-of-contract
Contracts
Standard DPA terms · SCCs available

Data residency in the EEA Private and EEA deployment

Evaluation

Specialist judgement, measured.

Some outputs only an expert can grade. We design the rubric with you, put qualified reviewers on it and record every judgement so the result can be audited.

  1. 01 Rubric and benchmark design What good looks like, written down before anyone grades.
  2. 02 Dual review and adjudication Two independent judgements, and a third where they disagree.
  3. 03 Expert grading of model output Engineers and domain experts score what the model produced.
  4. 04 Red teaming People trying to make the system fail, with every attempt logged.
  5. 05 Robustness under domain shift Degradation measured across terrain, weather, lighting, device and acoustics.
  6. 06 ASR and TTS evaluation Recognition and synthesis measured on your languages, dialects and noise.
  7. 07 Media integrity Real and generated pairs across face, body and voice for detection models.

When the data has to run inside a system, the same team builds it. AI Implementation

Scope a project

Start with one modality, one task and one acceptance line.

Tell us what the system has to see or hear, and under which rules. We reply with a feasibility read and a pilot plan against your own acceptance criteria.

What a good brief covers

  1. 01 Modality
  2. 02 Use case
  3. 03 Volume
  4. 04 Countries and languages
  5. 05 Specialist profile
  6. 06 Technical requirements
  7. 07 Acceptance criteria
  8. 08 Environment and data handling
  9. 09 Timetable

A bounded pilot carries its own acceptance criteria. You judge it against your own data.

The brief Name the modality, the task and the rules it runs under.

What the project needs (optional)

We use the information you provide to assess, respond to and manage your enquiry in accordance with our Privacy Policy.