Skip to main content
The headless methods take images your app already has and return the analysis. No camera, no screens.
These calls stand alone: they don’t create a workflow session or a reviewable session in the Admin Console. For onboarding, prefer a workflow.

Images: FileInput

Images must be JPEG or PNG, at most 15 MiB. The SDK uploads each one through a presigned URL before it calls the API.

scanDocument

Reads the fields on an ID.
  • type: .auto lets deepidv classify the document. It’s accepted only here, not by the capture views.
  • Only the front is read. A back image is uploaded and its key returned, but not read separately.
  • Returns DocumentScanResult. Its confidence is 0 to 1.

verifyIdentity

Reads the ID, detects the face on it and in the selfie, and compares them.

IdentityVerifyResult

Confidence scales differ: DocumentScanResult.confidence is 0 to 1, every confidence on IdentityVerifyResult is 0 to 100. The SDK passes them through unchanged.

Errors

Both methods throw DeepIDVError. The common kinds: .validation (image format or size, or a bad type), .insufficientFunds, .rateLimit, .network, .timeout. See Errors.

Getting the images