Signature detection for KYC intake

Where it sits in intake

A KYC file often includes an application or a consent page that is supposed to be signed, plus identity evidence that this API does not interpret. Run detection on the page that should carry the mark. Leave document authenticity, liveness, and watchlist checks on the systems you already use for those jobs.

1. Store the uploaded application page where this API can read it, or base64-encode it.
2. POST https://signature-detector.com/api/detect with x-api-key.
3. If result.count is 0, set the case to missing signature and skip downstream approval.
4. If marks came back, save detectionId and the boxes on the case. A reviewer can crop the region later. The API does not return the crop itself.

import requests

def application_needs_review(image_url: str, api_key: str) -> dict:
    response = requests.post(
        "https://signature-detector.com/api/detect",
        headers={
            "Content-Type": "application/json",
            "x-api-key": api_key,
        },
        json={"imageUrl": image_url},
        timeout=60,
    )
    body = response.json()
    if response.status_code == 402:
        raise RuntimeError("Out of detection credits")
    response.raise_for_status()

    marks = body["result"]["detections"]
    return {
        "application_status": "review" if not marks else "has-signature-mark",
        "detection_id": body["detectionId"],
        "boxes": [mark["bbox"] for mark in marks],
        "credits_remaining": body["creditsRemaining"],
    }

What not to automate

Do not treat has-signature-mark as "KYC passed." The box does not say the signature matches the person on the identity document. It does not say the form is complete. The only class in the published response is signature, so you cannot filter a separate logo or stamp class out of the JSON. A reviewer still has to reject anything that is not actually a signature. Confidence is the knob you have, and you apply it after the response. See check if a document is signed.

Packets that are PDFs

Account-opening packs are often PDFs. Send pdfUrl or pdfBase64 and inspect each item in the top-level pages array, using the contract on PDF signature detection. A page you require to be signed should not advance when its count is 0.

Cost for a pilot

• $10 per 100,000 credits, which is $0.0001 each. An image that is not served from cache spends 1 credit. A PDF spends 1 credit per page.
• 10 credits are added when the account is created.
• $1 minimum purchase. That is 10,000 credits. There is no subscription.
• Credits do not expire. The balance stays on the account until a detection spends it. A failed detection puts the credit back. Repeating an image or an identical PDF returns cached: true and does not spend a credit. On a PDF that hit is creditsUsed: 0.
• Buy credits from the dashboard or estimate a volume on the pricing page.

Ten free credits cover ten uncached page images, or one uncached 10-page PDF. A thousand application pages is 1,000 credits, or $0.10, when none of them hit the cache.