ID VERIFICATION OCR API

Our OCR software covers three types of ID documents, which are powered by our industry leading technology.​Passport OCR is capable of extracting information from Passports. ​

1. The following fields will be extracted from Passports (list fields)
2. ‍​Driver’s License OCR is capable of extracting information from Driver’s Licenses. The following fields will be extracted from Driver’s Licenses (list fields
3. Birth Certificate OCR is capable of extracting information from Birth Certificates. The following fields will be extracted from Birth Certificates (list fields)

The majority of customers benefit from Affinda’s ID document parser through our hosted solution. However, some users will the ID document Extractor to be deployed locally for various reasons. ​We have published an Affinda Self-Hosted Deployment Guide to aid customers who request this deployment method.

Affinda’s ID Document Extractor solution aims to accurately and quickly return key data from an ID document with as little human intervention as possible. The use of confidence levels and auto-validation is key to minimising this human intervention. ​​
‍
Confidence Levels​​

Artificial intelligence models are probabilistic. This means that the AI model will return what it believes is most likely to be correct, which will be the data it is most confident is the correct answer. Affinda returns the confidence levels to users, either via API or through the validation interface, so that users can more easily direct their attention to those fields that the model has less confidence in and therefore be more likely to be incorrect.​

The confidence levels shown take into consideration:​

• That the data point selected by the model is correct
• ​In the case of scanned images, the confidence that the model has that the capture of the text via OCR is correct​​

Affinda’s ID Document Extractor is capable of handling different date and number formats that exist across regions.​​

Dates will often be represented differently depending on the region. The most common date format difference is whether the month appears before or after days in the entire date. For example, dates in the United States are often presented as MM/DD/YY, whereas in the United Kingdom they are often presented as DD/MM/YY.

Additionally in passports, date formats may include the month denoted in English and the country’s primary or secondary language.​​

Affinda’s technology will parse all different date formats and return the extracted data in the ISO 8601 format YYYY-MM-DD.

The ID document OCR software is yet to automatically identify and categorise the type of ID document being submitted to the system. Currently users should submit the specified document to each extractor – i.e. a Passport to the Passport Extractor​​However, in the case that an incorrect document has been submitted to the extractor (e.g: a Driver’s License being submitted to the Passport extractor), the quality of the information extracted, and the model confidence levels will be very low. These metrics can be used to imply that an incorrect document has been submitted

The Machine Readable Zone (MRZ) is a codified element of identity documents which contains the basic personal details of the document holder including full name, document number, date of birth, sex and document expiration date.​​This information extracted can be used to validate the information extracted from the body of the passport (excluding the MRZ). This enhances the accuracy and the confidence in the= information extracted by the Passport extractor

Affinda has developed our data extraction solution to work either within our dedicated app or embedded within your platform. With the embedded option, workflows can remain largely unchanged to current processes with customisations available so that the tool fits seamlessly into your platform.​The validation interface can be customised to seamlessly fit within your platform. For more information on white-labeling and other customisations, see here.​​

How to embed the validation interface?​

1. ID documents posted to Affinda will extract and return via API most of the data from an ID document​
2. Within this API response, we return a value called meta.reviewUrl​
3. This is a signed and authenticated URL which can be embedded as an iFrame within your platform and allow members of your team to validate the data predicted and add any additional data points
4. ​Once an ID document is validated, the validated data can be requested and entered into your platform with full confidence in the accuracy

Why should I use the validation interface?​​

There are two key advantages of using the validation interface. ​​

1. Ensuring the integrity of data​
   Despite all of the benefits of using an AI-based engine for extracting data, the reality is that no model will ever be 100% accurate for every single ID document. The benefits of a 'human in the loop' model mean that predictions made by the model are validated and 100% accurate data can be guaranteed before further processing in your system.​
   The ease of use of the interface, and the already high accuracy of the model predictions, means that obtaining perfect data is now a matter of seconds, not minutes, per document.​
2. Creation of a feedback loop
   ‍​One key strength of AI models is that they can continue to learn over time. When using the validation interface, the data from these human corrections can be fed back into our models so that our engine can begin to learn different state and country formats and accuracy will continue to improve over time, further reducing the amount of human intervention required.