Trust & fit

Who is this data for?

Short “suitable for / not suitable for” guidance for teams buying pose and motion exports, plus printable license summaries to reduce legal back-and-forth.

Ready-made datasets

Enterprise & bulk available

Buy Human Pose Dataset (COCO / JSON) for AI Training

No annotation pipeline setup required — data is ready to plug into your training workflow.

View free dataset samples on GitHubIncludes walking + running sample exports (JSONL + manifests + stats).

Browse curated export bundles. Click a title to view full specifications, schema, and examples.

Companies & robotics teams: enterprise, bulk & custom supply →

Or generate via Dataset Lab
DatasetPriceActionVideosFramesFormatIndividualsGumroadVolume & customEnterprise
139-Video Dexterous Hand Movements Pose Dataset
Specs on detail page.Motion intelligence
$950 (one-time)Dexterous Hand Movements139123732JSONLBuy Enterprise
High-Quality Dexterous Hand Movements Dataset
Specs on detail page.Motion intelligence
$440 (one-time)Dexterous Hand movements5041160JSONLBuy Enterprise
$450 (one-time)Full Locomotion Bundle (Walking + Jogging + Running)19040020JSONLBuy Enterprise
$370 (one-time)walking10022923JSONLBuy Enterprise
$330 (one-time)jogging6114550JSONLBuy Enterprise
$250 (one-time)running204500JSONLBuy Enterprise
$350 (one-time)running10625944JSONLBuy Enterprise
$250 (one-time)dancing2011886JSONLBuy Enterprise

Showing 18 of 9 datasets

Note: prices and checkout are handled on Gumroad. Datasets are sold as exports (JSONL + metadata), not as raw videos.

Companies and robotics teams usually prefer custom enterprise deals instead of fixed subscription plans. View enterprise & bulk options or email info@qvision.space for volume pricing and recurring delivery.

Datasets

Simple Pricing That Converts

Annual offer:prices shown below already include a 30% annual discount (vs. reference list price).

Frame quotas for motion / pose dataset exports. Pay securely with PayPal. We are ready to deliver ready-made datasets — custom scope and licensing quoted separately (see below).

Think of each plan as a frames quota pack. You can purchase multiple packs over time as your needs grow.

Want free samples to validate the format and quality before buying? Browse our public GitHub repository with ready-to-parse exports.

View free dataset samples on GitHubIncludes walking + running sample exports (JSONL + manifests + stats).

Batch size (videos per job): depends on your machine (RAM, CPU). Jobs with at least 24 videos in one run have completed successfully in our tests; use smaller batches on lighter PCs to avoid the engine stopping mid-job.

Videos on this site are examples only. They illustrate quality and format — they are not offered for sale as raw clips.

Custom ready-made data (curated bundles by size and labels) is quoted per scope — contact us for a formal quote. We do not list enterprise or bulk pricing on this page.

PlanFrames quotaPrice (annual −30%)Best forAction
Free Trial7 videos$0Trying the workflowTry Lab
Basic7,500
$99 list$69per year · 30% off list
Serious experiments
Standard25,000
$256 list$179per year · 30% off list
Most popular — production pilots
Pro100,000
$713 list$499per year · 30% off list
Teams and product builds
Scale500,000
$2,856 list$1,999per year · 30% off list
Large-scale training pipelines

How this pricing compares (illustrative)

Third-party keypoint / pose annotation APIs often publish tiered rates on the order of roughly $1–$10+ per 100 images depending on volume and features (public marketing pages; not an endorsement). Custom dataset vendors usually quote enterprise projects on a custom basis for curated, licensed bundles.

Our plans are annual frame quotas for the Quality Vision export pipeline (JSONL, manifests, optional smoothing & augments) — closer to a productized subscription than pay-per-frame API pricing.

Reference export: 24-video lab job (internal QA)

  • All 24 sources were decoded and processed (diagnostics: 24/24 rows).
  • 99 HQ-approved frames before augmentation — the authentic frame count for that batch (not every source clip contributed frames; filters reject soft blur, low visibility, etc.).
  • 297 rows in merged data.jsonl includes augmented variants (e.g. mirror + keypoint noise), so row count is larger than the 99 unique accepted frames.
  • Plan frame quotas (e.g. 25,000 / year on Standard) are annual subscription caps, not a valuation of one export. A small lab batch is a valid marketing sample; standalone resale value depends on scope and license.

Ready-made dataset bundles

Need labeled pose / motion JSONL delivered without running the lab yourself? We can scope a bundle (frames, actions, format). Contact us for pricing — final terms are always confirmed in writing.

Sample videos

Clips shown in demos or documentation are samples for evaluation — not a catalog of videos for purchase. Commercial use requires the right dataset tier or a custom contract.

Need bulk supply, custom extraction from your footage, or a long-term data partnership? Open the enterprise & robotics page — same motion quality, scoped to your roadmap (contact: info@qvision.space).

Payments are processed by PayPal. Taxes or local fees may apply per PayPal and your region.