No vendor can sell you what nobody has photographed
Shelf imagery from Mekong Delta corner shops, Hanoi street signage, handwritten receipts from one specific province โ none of it exists until someone goes and captures it. We dispatch tasks to exactly where you need them, with constraints tight enough that what comes back is usable.
Tasks reach exactly the areas you need surveyed
You specify provinces, venue types, time windows and volume. The system pushes tasks to contributors in those areas โ not everyone receives every task.
Illustrative diagram of task dispatch, not contributor counts by province.
How a collection engagement runs
01
Agree the brief
Data type, volume, areas, quality constraints and delivery format. You get a quote with concrete dates, not an estimate.
02
Dispatch tasks
Tasks go to contributors in the right areas with specific instructions โ what to photograph, where to stand, what to avoid.
03
Check at intake
Every submission passes format checks, content matching against the brief, coordinate verification and deduplication. Failures are blocked at the door and never enter the archive.
04
Annotate and deliver
As images arrive, the annotation team processes them. You receive labeled data with a quality report โ not a pile of raw files and a problem.
Four constraints that make the data usable
The real risk in crowd collection is people photographing something just to get paid. We handle that with mechanisms that run automatically on every submission, not with trust.
Coordinate and radius constraints
Every submission carries GPS coordinates. When a brief requires surveying multiple locations, the system enforces a radius rule โ another shot from where you just stood is rejected, forcing contributors to move to a new site.
Content matched against the brief
Images are matched against the task instructions at intake โ not just the title, but the conditions you wrote. 'Receipt dated within 30 days', 'photographed directly, not off a screen' are conditions a machine can check.
Content-hash deduplication
Every file gets a hash from its contents. Re-submitting the same image โ renamed, or from a different account โ is recognised and refused.
Submission rate limits
Submitting faster than it would take to actually travel and shoot is a signature of automation. Those cases are blocked or routed to human review.
What you own
All data collected under your brief belongs to you
We do not resell it to anyone else โ that sits in the contract
An NDA is signed before we receive any file
Data that may contain personal information is anonymised before processing
Describe the dataset you need
Data type, areas, volume and timeline. If collection is not feasible within your budget, we will say so rather than take the job and deliver something unusable.