egpivo / ocr

OCR Web Service

Table of Contents

• Building the OCR Web Service
  • Prerequisites
  • Instructions
• Running the OCR Web Service
  • Instructions
• Using the OCR Web Service
  • Synchronous Endpoint (imgsync)
  • Asynchronous Endpoint (imgasync)
  • Check Job Status
• Reference
  • Commands in Makefile
  • Core Hierarchy
  • Demo

Building the OCR Web Service

Prerequisites

• Docker installed

Instructions

1. Unzip the OCR repository:

   unzip ocr.zip
   cd ocr

2. Build the Docker image:

   make docker-build

Running the OCR Web Service

Instructions

1. Run the Docker container:

   make docker-run

   • This will start the OCR web service on http://localhost:8000. You can change the port in the docker run command if needed.

2. Access the Swagger Documentation:

• Visit http://localhost:8000/docs in your web browser to access the interactive Swagger documentation. This provides details on the available endpoints, request formats, and response structures.

Using the OCR Web Service

• The OCR web service provides both synchronous and asynchronous endpoints for text extraction from images.

Synchronous Endpoint (imgsync)

• URL: http://localhost:8000/imgsync
• Method: POST
• Request Body:

  {
    "data": "base64_encoded_image"
  }

• Response:

  {
    "extracted_text": "extracted_text"
  }

Asynchronous Endpoint (imgasync)

• URL: http://localhost:8000/imgasync
• Method: POST
• Request Body:

  {
    "data": "base64_encoded_image"
  }

• Response:

  {
    "job_id": "unique_job_id"
  }

Check Asynchronous Job Status

• URL: http://localhost:8000/imgasync/job/{job_id}
• Method: GET
• Response:

  {
    "job_id": "job_id",
    "status": "completed",
    "extracted_text": "extracted_text"
  }

Reference

Commands in Makefile

1. make clean: Removes Python compiled files and cache locally.
2. make install: Installs project dependencies locally.
3. make run: Runs the OCR web service using Uvicorn locally.
4. make test: Runs the tests using pytest locally.
5. make docker-run: Builds a Docker image for the OCR web service and runs it.

Core Hierarchy

ocr
├── __init__.py
├── core
│   ├── __init__.py
│   ├── job_manager.py
│   └── schemas
│       ├── __init__.py
│       └── image.py
├── dependencies.py
├── logger
│   ├── __init__.py
│   └── logger.py
├── main.py
└── routers
    ├── __init__.py
    ├── image_sync.py
    └── image_async.py
    └── utils.py

Demo by Swagger UI

1. /imgsync

• Input an encoded image data, get the result as

2. /imgasync

• Input an encoded image data, get the result as

3. /imgasync/job/

• Based on the job_id from /imgasync, we have