What clients ask before signing

Four control layers run automatically on every submission, plus gold-standard samples mixed into live work to measure each contributor's accuracy over time — not just at onboarding. Anything below the committed accuracy is redone at no extra charge, and that sits in the contract rather than in a sales promise.

Anything below the agreed accuracy is redone free. If it still falls short after the rework, we do not charge for that portion. The safest path for both sides is a small pilot batch to settle the standard before committing volume.

Depends on the model you choose: your own team (self-service), a contributor network with measured accuracy (crowdsourced), or a dedicated TucTak team (fully managed). For sensitive data, choose self-service or fully managed to limit how many people see it.

No. The contract is explicit: data you send us stays yours, and data we collect under your brief is yours as well. We sign an NDA before receiving any file.

Vehicle plates and faces are masked automatically before the image enters any other processing step. If your data contains another category of personal information that needs masking, tell us in advance and we configure the corresponding step.

No — ready datasets are licensed to multiple parties and the price reflects that. If you need exclusivity, that is the commissioned collection service: data captured to your brief belongs to you and is not resold.

Annotation-only work is typically 3–10 business days depending on volume and difficulty. Work involving field collection needs capture time on top, usually 3–6 weeks. Concrete dates go in the quote, not an estimate.

There is no hard floor. We take pilot batches of a few thousand images — in practice most clients start small to judge quality themselves before committing to volume, and we encourage that.

Per image, varying by label type — classification costs a fraction of detailed segmentation — and by delivery model. Collection work is priced separately from annotation. Quotes always itemise unit prices rather than collapsing into a single number.

We accept images and video in common formats, uploaded through the dashboard or API. We return whatever your pipeline expects: COCO, YOLO, Pascal VOC, or JSON against your own schema.

Per call, drawn down from a prepaid credit balance. The dashboard itemises every call and its cost, so you control spend without waiting for a month-end invoice.