Service 01
Data Collection & Generation
We plan, cast and shoot original video, audio and image datasets to your exact spec, produced end to end in house.
AIDataTrix runs training data collection end to end: we plan the shoot, cast and direct the people in it, and record the video, audio and images ourselves.
What we do
The work itself.
- Plan the shoot against your specification
- Cast and direct the people in it
- Shoot video, audio and image data in house
- Deliver the material with signed consent on file
To quote
What we need from you.
These are the questions we ask before we can put a number on it.
- Which modality: video, audio, image, or a mix?
- What scene, task or behaviour has to be on camera or on mic?
- How many participants, and any casting requirements: age range, languages spoken, attire?
- How much material: hours of footage, number of clips, or number of images?
- Resolution, frame rate or sample rate, if your pipeline fixes them?
- On a set, or a particular kind of location?
- Annotated as well, or raw and edited only?
- When do you need the first delivery?
Deliverables
What you receive.
- Raw and edited video
- Raw and edited audio
- Image sets
- Secure cloud transfer
Assurance
How the work is checked.
Signed consent from everyone we record. Standard practice on every shoot.
Questions
Asked and answered.
Both. For staged and scenario based datasets we plan, cast and shoot in house. For annotation work you send your own files and we label them to your spec.
Yes. Signed consent is obtained from everyone we record, as standard practice.
Depending on the engagement: raw and edited video and audio, annotated data as JSON, COCO or CSV, labelled image sets, Excel and CSV QA trackers, or licensed dataset files. Everything moves over secure cloud transfer.