Flask API

• End point URL:

/processAadhar

• Allowed file types: 'png', 'jpg', 'jpeg'
• Imports and runs OCR_Dictionary

Aadhar Data Extraction

Contains 4 modules

• OCR.py
• PreProcessor.py
• AadharExtractor.py
• OCR_Dictionary.py

OCR.py

• Uses EasyOCR for recognizing text
• Takes the path of the images as the parameter and converts all the text from the image to a list of lines
• Returns the list of lines
• All EasyOCR parameters are set in this file

PreProcessor.py

• Has it's own implementation of EasyOCR with a different set of parameters for better pre-processing
• Identifies and removes non-english character. The bounding box is replaced by white pixels

AadharExtractor.py

• Contains the set of regular expression rules for extraction of required data from the list of lines received from the OCR module

OCR_Dictionary.py

• Imports and runs all the above modules
• Takes the image path and the dump path as parameters
• The dump path is used by the PreProcessor module to temporarily store the pre-processed image

Dockerfile

• The EasyOCR module crashed when it tries to download from the EasyOCR model hub
• To deal with this problem I have manually downloaded and added all the required models from JadedAI model hub to the required paths(inside the container)
• Incase in the future the models are depricated or the download link goes down you can download the models from the this google drive link

Running in Local Environment

• Clone this repository and install Docker on your local machine
• Build a new Docker image by Running

docker build -t <name of the container:<version number>> <path of folder that contains docker file>

• we can give '.' as the if the dockerfile is in the same directory. we can give the version number as latest
• Run the container using

docker run -p <port number of local machine>:<port number of docker container>

• The port number of the docker container has been set to 6000
• Now follow all the steps for running locally mentioned in this repository