Web OCR

---

About

A web interface for OCR of multiple languages. It uses tesseract ocr in the backend for processing. Currently, pdf files and image files are supported. Also, only english and hindi are currently added but other languages can be supported trivially

Main Features

• Simple web based gui for OCR.
• Supports both pdf and image inputs
• High quality ocr using LSTM based version of Tesseract OCR
• Sends mail on job completion

Requirements

• Tesseract OCR
• OCRmyPDF
• img2pdf
• Sendgrid (for sending email)
• AWS S3 (for storing resulting pdfs)

Installation

• Install OCRmyPDF
• Clone this repository and do npm install to install required packages
• Export the following environment variables:

S3_ACCESS_KEY          # AWS S3 access key
S3_SECRET_KEY          # AWS S3 secret key
SENDGRID_USER       # Sendgrid api id
SENDGRID_PASS       # Sendgrid api key
PORT                # Port to run webserver on
S3_BUCKET_NAME      # E.g. webocr
S3_ENDPOINT         # E.g. https://s3.ap-south-1.amazonaws.com
S3_REGION           # E.g. ap-south-1

Usage

• npm start to run the webserver.

Docker usage

docker run -d -e "PORT=<port>" -e "ACCESS_KEY=<access_key>" \
              -e "SECRET_KEY=<secret_key>" -e "SENDGRID_USER=<user>" \
              -e "SENDGRID_PASS=<key>" -e "CONFIG=<user db>" \
              -v <volume dir>:/home/docker/app/uploads ozym4nd145/ocr_web

OR

docker run -d --env-file <env.list file> -p 3000:3000 -v <volume dir>:/home/docker/app/uploads ozym4nd145/ocr_web