Validating data from a Document (NEW!)
Facial-CDV allows you to extract the data from an image of a document (e.g., CNH), and then validate the registration data and face image. For additional technical information about the endpoint, please refer to the validate/facial-cdv documentation in the API reference.
This endpoint includes the functionality of CDV - Cognitive Document Validation in Datavalid. CDV, or Computerized Document Verification, goes beyond OCR (Optical Character Recognition) by utilizing computer vision and artificial intelligence. It is designed to identify, classify, and extract textual data and photos from CNH (Carteira Nacional de Habilitação).
Notes
Currently, Facial-CDV does not perform:
- Scan of QR Code
- Does not validate the signature
- Extraction of affiliation and observation fields
- Extraction of document security numbers
Step by step
1. Get photos of the document
They can be taken in two forms, A or B: A) A picture of the entire CNH
- CNH shall be removed from the plastic;
- With CNH open, get a photo containing the entire document.
B) Two photos of the document, one of the front and one of the back
- CNH shall be removed from the plastic;
- Take a photograph of the front of the CNH with the document folded,
- Along with an image of the reverse side showing the section that includes the observations field.
Picture Quality
Pictures should be clear, without spots, with the right contrast and brightness. The precision of the data derived from the Document hinges upon the superiority of the image delivered from the document. The resolution of the facial image is also critical for validation.
Notice on the extraction of data
Retrieving data from the Document is a probabilistic, uncertain activity, and it is the responsibility of the user of the service to determine the validity of the document sent based on the values returned in the response.
2. Specifications for the photo
2.1 Images must have a minimum width of 1000 pixels.
The images must be a minimum of 1000 pixels wide, while preserving the original height of the image in pixels (height around 700 pixels).
Attention: Please ensure that the ratio between height and width is maintained!
The height of the document should be proportional to its width, following the original proportions of the document. Please avoid sending images with distorted aspect ratios. For the above example, keeping the original aspect ratio of the document, the height of the folded CNH front image would be 706 px. The examples below are not supported.
Disproportionately increased height should be avoided:
Disproportionately reduced height should be avoided:
2.2 The image must not be vertical or tilted (not yet supported)
2.3 Photo background must be uniform in color.
For optimal results, the document should be placed on a uniform background without any letters or numbers.
As an example, it's best to avoid taking pictures of the document on a newspaper or book page.
API Reference
For endpoint technical details, consult the validate/facial-cdv documentation in the API Reference
2.4 Observe the model of driver's license accepted
The example below illustrates that only the outdated CNH layout is supported right now.
NEW LAYOUT IS NOT SUPPORTED
The new driver's license model launched in 2022 is NOT yet SUPPORTED by Datavalid. Datavalid does not currently support CNH Digital layouts.