---
title: "Video Data Collection at Scale | YPAI"
url: https://ypai.ai/video-data-collection/
description: "YPAI designs and operates video collection for AI teams whose dataset does not cover the model task, or whose licence does not cover the use they intend."
source: "src/copy/routes (route /video-data-collection/)"
---

# Video Data Collection at Scale

> YPAI designs and operates video collection for AI teams whose dataset does not cover the model task, or whose licence does not cover the use they intend.

Board markup remains in lib/video-data-collection/runtime-source/coverage-field.html because assembleHtmlBoard is required presentation logic. Every buyer-facing board string is owned by the board map below.

Collect the video your dataset does not cover

Your model may need footage that does not exist, misses the conditions it will meet in use, or cannot be used under its current licence. YPAI designs and operates a managed video collection project to close that gap.

Your engineers define what the footage must show and how it will be judged. YPAI runs the collection against that brief.

Share the footage gap. A member of the team will review what you send.

A managed video collection project at scale

YPAI manages sourcing and capture across the locations and conditions needed for the required coverage. Accepted footage is handed to engineering with the records needed to verify it before use.

Turn a coverage gap into recording instructions

A file count shows volume, not whether the footage covers the conditions the model will meet in use.

Locations / conditions / required share

A missing scenario becomes a coverage requirement. If a group or market needs a fixed share, the plan gives it a quota. The resulting instructions tell recorders what must appear in the frame and which conditions to reproduce.

Remote capture and moderated capture suit different tasks

The task determines whether coverage comes from distributed recording, active moderation, a fixed location, supplied hardware or synchronised streams. YPAI sets the final configuration with you before commitment.

Remote capture supports distributed coverage across markets. Participants record on approved devices and follow a defined capture flow.

A moderator gives instructions and intervenes during recording when the task requires it.

Some tasks need a fixed location or hardware supplied by YPAI. Others add camera or sensor streams that stay synchronised with the video.

Each asset keeps its own record

YPAI owns the collection platform and its human-review workflow. Each asset carries a source record; any derivative links back to that source. The review record names the reviewer and rubric, then stores the evidence and decision.

Technical checks follow the project plan. Human reviewers examine recordings selected under the sampling plan. The reviewer applies the rubric and records the evidence, decision and reason. A result that needs another judgement goes to adjudication or escalation.

If the acceptance or change-control path calls for rework or recapture, YPAI returns the recording with the reason.

Source, derivative and review records stay connected so the receiving team can inspect what changed, who reviewed it and which decision applies.

Engineering receives a delivery it can verify

The handover groups accepted media in a delivery manifest with the agreed metadata and QA or acceptance record. Engineering uses that package to verify the asset and version before it enters the model workflow.

Agree rights and processing before capture

You and YPAI identify the purpose and applicable lawful basis before processing personal data, then record any source rights, releases, licences or permitted-use terms separately and keep them with the asset.

You and YPAI set the processing location, retention period and relevant subprocessors before work starts. Standard contractual clauses are available for a transfer outside the EEA that you direct, while the media stays in the environment specific to you and agreed for the work.

YPAI uses separate workflows for data-subject requests and participant withdrawals. A withdrawal carries its own audit trail. Under the standard terms, erasure follows within 30 days after the contract ends.

YPAI checks the required coverage and capture setup against available capacity before setting the schedule. Before capture, you and YPAI agree the scope, acceptance terms, change control and price for the work.

A pilot tests the proposed method against the acceptance criteria agreed for the project. You review the result and its quality measurements, then decide how to respond to any recorded deviation.

After that review, you decide whether to commission production.

Tell us where the current footage falls short and what the new footage must show. A member of the team will review what you send.
