The PixLab Optical Character Recognition team is thrilled to announce that its document scanning API endpoint /DOCSCAN, is now able to scan U.S. Driver’s licenses and driving permits issued by jurisdictions from all the 50 U.S. states.

The /DOCSCAN API endpoint now allows any Website that is presented with a U.S Driver’s License, International Passport or ID Card to verify that the inputted information by the end user matches those present on the submitted or uploaded ID document image.

Usage & Code Samples

Given an input U.S driver’s license image issued by any of the 50 U.S. states, crop the license holder face, and extract fields of interest as follow:

Input U.S driver’s license image Car Vectors by Vecteezy

Extracted Fields Showcase

The extracted fields after successful call to the /DOCSCAN API endpoint are:

• License holder cropped face. This image will be stored on an AWS S3 bucket of your choice if you connect your target bucket from the PixLab Console.
• Issuing Country (USA obviously).
• Issuing State Name.
• Issuing State Two-Letter Code.
• License Number.
• License Holder’s Full Name.
• License Holder’s Address.
• License Holder’s Date of Birth (yyyy-mm-dd).
• License Issuing Date (yyyy-mm-dd).
• License Expiry Date (yyyy-mm-dd).
• License Holder’s Gender.

The code samples that is used to achieve such results are available via the following Gists:

• Python Gist. Also available on the PixLab Gitub Repository:
• PHP Gist. Also available on the PixLab Gitub Repository:

Algorithms Under the hood

• Face extraction is automatically performed using the /FACEDETECT API endpoint.
• /DOCSCAN already supports GET & POST HTTP methods so you can upload your document images directly from your application without relying on a foreign server. Refer to this Gist on how to do so.
• Upon the image processed on our server, it is automatically deleted. We do not keep trace or any log of your input images.
• Internally, we mainly rely on PP-OCR which is a practical ultra-lightweight OCR system that is mainly composed of three parts: Text Detection, Bounding Box Isolation, & Text Recognition. This combination produces highly accurate results in less than 5 seconds of processing.

Further Reading

• The official endpoint documentation is available to consult here, and the reader is more than encouraged to take a look at our production ready, code samples, available in various programming languages on the PixLab Code Samples Page.

The PixLab Document Scanner, development team is pleased to announce that is now fully support scanning Emirates (UAE) ID & Residence Cards via the /DOCSCAN API endpoint at real-time using your favorite programming language.

When invoked, the /DOCSCAN HTTP API endpoint shall Extract (crop) any detected face and transform the raw UAE ID/Residence Card content such as holder name, nationality, ID number, etc. into a JSON object ready to be consumed by your app.

Below, a typical output result of the /DOCSCAN API endpoint for an Emiratis (UAE) ID card input sample:

The code samples used to achieve such result are available to consult via the following gists:

• Python Code Samples for Scanning UAE ID Card: uae_emirates_id_card_scan.py
• PHP Code Samples for Scanning UAE ID Card: uae_emirates_id_card_scan.php
• PixLab Github Repository: github.com/symisc/pixlab

The same logic applies to scanning official travel documents like Visas, Passports, and ID Cards from many others countries in an unified manner, regardless of the underlying programming language used on your backend (Python, PHP, Ruby, JS, etc.) thanks to the DOCSCAN API endpoint as shown in previous blog posts:

• Passports & Travel Document Scan: Blog Announcement & Code Sample.
• Malaysia & Singapore ID Card Scan: Blog Announcement & Code Sample.
• Aadhar India ID Card Scan: Blog Announcement & Code Sample.

Algorithm Details

Internally, PixLab's document scanner engine is based on PP-OCR which is a practical ultra-lightweight OCR system, mainly composed of three parts: DB text detection, detection frame correction, and CRNN text recognition. DB stands for Real-time Scene Text Detection.

The system adopts 19 effective strategies from 8 aspects including backbone network selection and adjustment, prediction head design, data augmentation, learning rate transformation strategy, regularization parameter selection, pre-training model use, and automatic model tailoring and quantization to optimize and slim down the models of each module.

In PP-OCR, Differentiable Binarization (DB) is used as text detector which is based on a simple segmentation network. It integrates feature extraction and sequence modeling. It adopts the Connectionist Temporal Classification (CTC) loss to avoid the inconsistency between prediction and label.

The algorithm is further optimized in five aspect where the detection model adopts the CML (Collaborative Mutual Learning) knowledge distillation strategy and CopyPaste data expansion strategy. The recognition model adopts the LCNet lightweight backbone network, U-DML knowledge distillation strategy and enhanced CTC loss function improvement, which further improves the inference speed and prediction effect.

Here at PixLab, we recently deployed on production, a brand new gender/age classification model available to our customers via the FACEMOTION API endpoint.

• The new model implementation is based on the ResNet-50 convolutional neural network (CNN) that is 50 layers deep. The network can easily classify images into 1000 object categories, such as keyboard, mouse, pencil, and many animals.

• The reference, implementation paper is from: Jiankang Deng, Jia Guo, Niannan Xue, Stefanos Zafeiriou: Additive angular margin loss for deep face recognition. In: Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR) (2019 (https://arxiv.org/abs/1801.07698).

• The Python/PHP code samples listed below should be able to easily output the age estimation, gender, and emotion pattern by just looking at the facial shape of any present human face in a given picture or video frame using our new classifcation model.

Python Code

• The PHP usage sample can be viewed here or on the PixLab Github repository.

• FACEMOTION is the sole endpoint needed to perform such a task. It should output the rectangle coordinates for each detected human face that you can pass verbatim if desired to other processing endpoints like CROP or MOGRIFY plus the age estimation, gender and emotion pattern of the target face based on its facial shape.
• Finally, all of our production ready, code samples are available to consult at our samples page or the PixLab Gihtub repository.

The PixLab Computer Vision Team is pleased to introduce the Pixel Generate API endpoint (/pixelgenerate) which let you in a single call, generate on the fly, images filled with random pixels of desired width & height using a mix of standard Image Processing and soon Machine Learning algorithms.

This endpoint is similar to /newimage except that the image contents is filled with random pixels. This is very useful for generating background (negative) samples for feeding Machine Learning training algorithms for example.

By default, this endpoint return a JSON object holding a link to the generated image output. But, you can set it via the Blob parameters to return the image binary contents instead.

Below, a Python snippet which generate on the fly a new image of height & width of 300x300 filled with random pixels using a single call to /pixelgenerate:

The code sample used to achieve such result is available to consult via the following Github link:

• Python code for generating random pixels: pixel_generate.py.
• The endpoint documentation is available to consult at pixlab.io/cmd?id=pixelgenerate.
• Others code samples including Passports & ID Scanning ,Face Blurring, etc. and for a general introduction to the PixLab API, refer to our Github Repository and Examples Page.

A Passport is a document that almost everyone has at some point in their lives. It is issued by the country’s government to its citizens and mainly being used for traveling purposes. It also serves as proof of nationality, name, and more importantly an Universally Unique ID for its owner.

Many services have been long-time accepting passports as identification documents from their customers to complete their KYC (Know Your Customer) form as required by the legislation in force. This is especially true and enforced for the Finance, HR or Travel sectors. In most cases, a human operator will verify the authenticity of the submitted document and grant validation or reject it.

Things can get really complicated if you have hundreds of KYC forms to checks, but also if your clients differ in nationality. Quickly, you will find yourself drowning in physical copies of passports in different languages that you can not even understand. Let alone the potential legal problems you can face with passport copies laying around the office. This is why, an automated & safe solution for Passports processing is required!

Modern Passport Structure

From the 1980s on wards, most countries started issuing passports containing an MRZ. MRZ stands for the Machine Readable Zone and is usually located at the bottom of the Passport as shown below:

Passports that contain an MRZ are referred to as MRPs, machine-readable passports (Almost all modern issued Passports have one). The structure of the MRZ is standardized by the ICAO Document 9303 and the International Electro-technical Commission as ISO/IEC 7501-1.

The MRZ is an area on the document that can easily be read by a machine using an OCR Reader Application or API. It’s not important for you to understand how it works, but if you look at it carefully, you will see that it contains most of the relevant information on the document, combined with additional characters and a checksum that can be extracted programmatically and automatically via API as we will see in the next section.

Once parsed, the following information are automatically extracted from the target MRZ and made immediately available to your app, thanks to the /docscan API endpoint:

• issuingCountry: The issuing country or organization, encoded in three characters.
• fullName: Passport holder full name. The name is entirely upper case.
• documentNumber: This is the passport number, as assigned by the issuing country. Each country is free to assign numbers using any system it likes.
• checkDigit: Check digits are calculated based on the previous field. Thus, the first check digit is based on the passport number, the next is based on the date of birth, the next on the expiration date, and the next on the personal number. The check digit is calculated using this algorithm.
• nationality: The issuing country or organization, encoded in three characters.
• dateOfBirth: The date of the passport holder's birth in YYMMDD form. Year is truncated to the least significant two digits. Single digit months or days are perpended with 0.
• sex: Sex of the passport holder, M for males, F for females, and < for non-specified.
• dateOfExpiry: The date the passport expires in YYMMDD form. Year is truncated to the least significant two digits. Single digit months or days are perpended with 0.
• personalNumber: This field is optional and can be used for any purpose that the issuing country desires.
• finalcheckDigit: This is a check digit for positions 1 to 10, 14 to 20, and 22 to 43 on the second line of the MRZ. Thus, the nationality and sex are not included in the check. The check digit is calculated using this algorithm.

Automatic Passport Processing

Fortunately for the developer wishing to automate Passports scanning, PixLab can automatically scan & extract passport MRZ but also help to detect possible fraudulent documents. This is made possible thanks to the /docscan API endpoint which let you in a single call scan government issued documents such as Passports, Visas or ID Cards from various countries.

Besides extracting MRZ, the /docscan API endpoint shall automatically crop any detected face and transform binary Machine Readable Zone into stream of text content (i.e. full name, issuing country, document number, date of expiry, etc.) ready to be consumed by your app in the JSON format.

Below, a typical output result of the /docscan endpoint for a passport input image:

What follow is the gist used to achieve such result:

Other document scanning code samples are available to consult via the following Github links:

• Python code for scanning Passports: passport_scan.py.
• PHP code for scanning Passports: passport_scan.php.

Face extraction is automatically performed using the /facedetect API endpoint. For a general purpose Optical Character Recognition engine, you should rely on the /OCR API endpoint instead. If you are dealing with PDF documents, you can convert them at first to raw images via the /pdftoimg endpoint.

Conclusion

The era we are in is more digitized than ever. Tasks that are repetitive are slowly being replaced by computers and robots. In many cases, they can perform these tasks faster, with a smaller amount of mistakes and in a more cost-effective manner. At PixLab we focus on building software to replace manual repetitive labor in administrative business processes. The processing and checking of passports can be very time-consuming. Using /docscan to automate your passport processing will enable you to save cost, on-board customers faster and reduce errors in administrative processes.

Our colleague Vincent just published an interesting blog post on how to automatically detect and blur faces at real-time using the PixLab API with a nice introduction on how modern face detection algorithms works under the hood and the privacy concerns related to such use of technology!

• Article Link: dev.to/unqlite_db/detect-blur-faces-programmatically-2cbp.
• DZone Article: dzone.com/articles/detect-and-blur-faces-programmatically.

• facedetect API Endpoint Documentation: pixlab.io/cmd?id=facedetect.
• morgify API Endpoint Documentation: pixlab.io/cmd?id=mogrify.
• Python Code Gist.
• PHP Code Gist.
• Github Repository: github.com/symisc/pixlab.

The PixLab OCR team is pleased to introduce the /docscan API endpoint which let you in a single call scan government issued documents such as Passports, Visas or ID Cards from various countries.

Besides its accurate text scanning capabilities, the /docscan API endpoint shall automatically extract any detected face and transform binary data such as Passport Machine Readable Zone (MRZ) into stream of text payload (i.e. full name, issuing country, document number, date of expiry, etc.) ready to be consumed by your app in the JSON format.

Below, a typical output result of the /docscan endpoint for a passport input image:

The code samples used to achieve such result are available to consult via the following Github links:

• Python code for scanning Passports: passport_scan.py.
• PHP code for scanning Passports: passport_scan.php.

Face extraction is automatically performed using the /facedetect API endpoint. For a general purpose Optical Character Recognition engine, you should rely on the /OCR endpoint instead. If you are dealing with PDF documents, you can convert them at first to raw images via the /pdftoimg endpoint.

Below, a typical Python code snippet for scanning passports:

import requests
import json

# Given a government issued passport document, extract the user face and parse all MRZ fields.
#
# PixLab recommend that you connect your AWS S3 bucket via your dashboard at https://pixlab.io/dashboard
# so that any cropped face or MRZ crop is stored automatically on your S3 bucket rather than the PixLab one.
# This feature should give you full control over your analyzed media files.
#
# https://pixlab.io/#/cmd?id=docscan for additional information.

req = requests.get('https://api.pixlab.io/docscan',params={
    'img':'https://i.stack.imgur.com/oJY2K.png', # Passport sample
    'type':'passport', # Type of document we are a going to scan
    'key':'Pixlab_key'
})
reply = req.json()
if reply['status'] != 200:
    print (reply['error'])
else:
    print ("User Cropped Face: " + reply['face_url'])
    print ("MRZ Cropped Image: " + reply['mrz_img_url'])
    print ("Raw MRZ Text: " + reply['mrz_raw_text'])
    print ("MRZ Fields: ")
    # Display all parsed MRZ fields
    print ("\tIssuing Country: " + reply['fields']['issuingCountry'])
    print ("\tFull Name: "       + reply['fields']['fullName'])
    print ("\tDocument Number: " + reply['fields']['documentNumber'])
    print ("\tCheck Digit: "   + reply['fields']['checkDigit'])
    print ("\tNationality: "   + reply['fields']['nationality'])
    print ("\tDate Of Birth: " + reply['fields']['dateOfBirth'])
    print ("\tSex: "           + reply['fields']['sex'])
    print ("\tDate Of Expiry: "    + reply['fields']['dateOfExpiry'])
    print ("\tPersonal Number: "   + reply['fields']['personalNumber'])
    print ("\tFinal Check Digit: " + reply['fields']['finalcheckDigit'])

Finally, the official endpoint documentation is available to consult at pixlab.io/cmd?id=docscan and a set of working samples in various programming language are available at the PixLab samples pages.

The PixLab OCR team is pleased to announce that is now fully support scanning India Aadhar ID Cards besides Malaysia (MyKad) and Singapore identity cards as well governments issued Passports from all over the world via the /docscan API endpoint.

When invoked, the /docscan API endpoint shall Extract (crop) any detected face and transform the raw Aadhar ID card content such as holder name, gender, date of birth, ID number, etc. into a JSON object ready to be consumed by your app.

Below, a typical output result of the /docscan API endpoint for a Aadhar ID card input sample:

The same API call applies for Passports as well different ID cards from supported countries (you just specify the country name or ISO code):

The code samples used to achieve such result are available to consult via the following Github links:

• Python code for scanning Aadhar ID cards (as well other ID cards from different countries): id_card_scan.py.
• PHP code for scanning Aadhar ID cards: id_card_scan.php.
• Python code for scanning Passports: passport_scan.py.
• PHP code for scanning Passports: passport_scan.php.
• For converting PDF documents to raw images, you can rely on the /pdftoimg API endpoint as shown in this Python or PHP gist.

Face extraction is automatically performed using the /facedetect API endpoint. If you are dealing with PDF documents, you can convert them at first to raw images via the /pdftoimg endpoint.

Finally, the official endpoint documentation is available to consult at pixlab.io/cmd?id=docscan and a set of working samples in various programming language are available at the PixLab samples pages.