Classify a document automatically

Sends a document image without declaring a template. The OCR identifies the document type and side, checks that it is in the allowed_templates list and then runs the same validations and extraction as sending with a template.

Use it when the user may send more than one document type — for example, a CNH or an RG — and you do not want to ask them to choose before the capture.

Endpoint

ENDPOINT

/image

METHOD

POST

STATUS

200

This is the same endpoint used to send with a template. The request body is what changes the behavior: send allowed_templates instead of template. The two fields are mutually exclusive — sending both, or neither, returns 400.

Authentication

Send the API key in the Authorization header. See Authentication.

Request body

Field	Type	Required	Description	Size
document_b64	string	Yes	Document file in base64. JPEG, or a single-page PDF.	3 MB (JPEG), 30 MB (PDF)
allowed_templates	string[]	Yes	Document types you accept. Values in the Allowed templates table.	—

{
  "document_b64": "YOUR_BASE64_DOCUMENT",
  "allowed_templates": ["cnh", "rg"]
}

import requests

response = requests.post(
    "https://api.caas.qitech.app/ocr/image",
    headers={"Authorization": "YOUR_API_KEY"},
    json={
        "document_b64": "YOUR_BASE64_DOCUMENT",
        "allowed_templates": ["cnh", "rg"],
    },
)
print(response.status_code, response.json())

Allowed templates

Send the whole document (cnh) to accept either side, or a specific side (cnh_front) to accept only that side.

Value	Accepts
cnh	Physical CNH (driver's license), front (cnh_front) or back (cnh_back).
cnh_front, cnh_back	Only the given side of the physical CNH.
cnh_digital	PDF of the digital CNH.
rg	Physical RG or CIN (identity card), front (rg_front) or back (rg_back).
rg_front, rg_back	Only the given side of the physical RG or CIN.
cin_digital	PDF of the digital CIN.
rg_digital	PDF of the digital RG.
national_migration_registry	CRNM (national migration registry card), front or back.
national_migration_registry_front, national_migration_registry_back	Only the given side of the CRNM.
national_registry_of_foreigners	RNE (national registry of foreigners card), front or back.
national_registry_of_foreigners_front, national_registry_of_foreigners_back	Only the given side of the RNE.

Response

Field	Type	Description
ocr_key	uuid	Image key. Use it to reference the document in other QI Tech products.
template	string	Identified template, always including the side when the document has two: cnh_front, rg_back, cnh_digital...
is_inferred_template	boolean	true when the template was identified by automatic classification.
allowed_templates	string[]	The list sent in the request.
created_at	string	Processing date and time, in ISO 8601 (UTC).

{
  "ocr_key": "f1c0d2e1-f950-4360-896d-36588e443fc9",
  "template": "cnh_front",
  "is_inferred_template": true,
  "allowed_templates": ["cnh", "rg"],
  "created_at": "2026-10-16T13:45:12Z"
}

You can retrieve the image later by its ocr_key, as described in Sending a document.

Important

Do not use real personal or company data in the sandbox environment.

Errors

Rejected document

When the image is processed but rejected, the response is 400 and includes the ocr_key of the attempt. title gives the category and reason the cause:

{
  "title": "invalid",
  "description": "The uploaded image cannot be successfully processed.",
  "reason": "document_not_allowed",
  "ocr_key": "f1c0d2e1-f950-4360-896d-36588e443fc9"
}

Status	title	reason	When it happens
400	invalid	template_not_classified	The OCR did not recognize the document.
400	invalid	document_not_allowed	The document was recognized but is not in allowed_templates.
400	invalid	multiple_documents_found	There is more than one document in the image.
400	invalid	invalid_document	The image was not accepted as a valid document of the identified type.
400	invalid	missing_document_section	Part of the document is outside the image or covered.
400	invalid	missing_data	Required fields of the document could not be read.
400	invalid	qr_code_not_found	Digital PDF (CNH, RG or CIN) without a QR code.
400	invalid	not_digitally_signed	Digital PDF (CNH, RG or CIN) without the issuer's digital signature.
400	bad_quality	blurred_document	Blurred or low-quality image. JPEG only.
400	bad_quality	document_position_invalid	The document touches the edge of the image. JPEG only.
400	bad_quality	glare_on_document	Glare that hides or distorts information on the document.
400	bad_quality	color_irregularity	Irregular color distribution on the document.
400	bad_quality	face_not_found	No face detected in the document photo.
400	bad_quality	face_out_of_bounds	The face is partly outside the image.
400	bad_quality	face_brightness_problem	The face is too dark.
400	bad_quality	face_too_close	The face takes up too large an area of the image.
400	bad_quality	face_too_distant	The face takes up too small an area of the image.

Face validation (face_*) applies only to the sides that carry a photo: CNH front, RG or CIN front, CRNM front and RNE back.

Refused request

These errors happen before processing and do not generate an ocr_key:

Status	title	reason	When it happens
400	Invalid JSON	—	The request body is not valid JSON.
400	Request data failed validation	—	Body outside the contract: a required field is missing, an unknown field is sent, a value is outside the list, or template is sent together with allowed_templates.
400	Invalid base64	invalid_base64	document_b64 is not valid base64, or the file is neither JPEG nor PDF.
400	File is too big	max_file_size_exceeded	JPEG over 3 MB or PDF over 30 MB.
400	Document is too big	max_pages_exceeded	PDF with more than one page.
401	Authorization header not found	—	The Authorization header is missing.
401	Deactivated ApiKey	—	The API key has been deactivated.
403	Invalid ApiKey	—	The API key does not exist.
500	Internal Server Error	—	Internal failure. Try again later.

Generic HTTP statuses are listed in HTTP status.