Skip to content

Latest commit

 

History

History
119 lines (75 loc) · 3.91 KB

README.md

File metadata and controls

119 lines (75 loc) · 3.91 KB
Raw

OCR Tool

A Python script that uses Azure Congitize Services to OCR text from images.

An Illustration

image with handwritten text

$ python main.py file --filepath images/example2.jpg
This is an example of some
text. I wonder if it will
get extracted perfectly

Usage

This tool allows two types of operation; you can either ask it to stream a local image to the Azure services, or you can pass the URL to a publicly accessible image.

If you run the script without any parameters you will see the following help text:

$ python main.py --help
Usage: main.py [OPTIONS] COMMAND [ARGS]...

Options:
  --help  Show this message and exit.

Commands:
  file  Sends a local file to OCR to extract text
  url   Sends a file, accessible via a public url, to OCR to extract text

Extracting Text from a Local Image File

The file command will allow you to extract text from a local image file.

The help text for this command:

$ python main.py file --help
Usage: main.py file [OPTIONS]

  Sends a local file to OCR to extract text

Options:
  --filepath PATH  Submits a local file to OCR  [required]
  --help       Show this message and exit.

For example:

$ python main.py file --filepath images/example1.jpg

Extracting Text from an Image File via Publicly Addressable URL

The url command will allow you to extract text from an image file at the end of a publicly addressable url.

The help text for this command:

$ python main.py url --help
Usage: main.py url [OPTIONS]

  Sends a file, accessible via a public url, to OCR to extract text

Options:
  --url TEXT  Submits a public url to OCR  [required]
  --help      Show this message and exit.

For example:

python main.py url --url https://raw.githubusercontent.com/Azure-Samples/cognitive-services-sample-data-files/master/ComputerVision/Images/printed_text.jpg

Setup

Create an Azure Cognitive Services Account

To use this script you will need to create an Azure Cognitive Services account.

https://azure.microsoft.com/en-gb/free/services/cognitive-services/

Once set up, you will need to grab the following two pieces of information:

  1. Your Computer Vision Subscription Key
  2. You Computer Vision endpoint URL

Create Environment Variables

Two environment variables are required. You can either set these up as environment variables before running the tool, or you can create a .env file and the code will load this file when executed.

The file .env-example is a template file. Copy or rename this file to .env, and then replace the {{REPLACE_ME}} text with the appropriate values.

The environment variables values that are required are as follows"

Name Description
COMPUTER_VISION_SUBSCRIPTION_KEY A token that identifies your Computer Vision subscription key. You get this value from the Azure Portal when you set up an Azure Cognitive Services account
COMPUTER_VISION_ENDPOINT A URL to an endpoint. You get this value from the Azure Portal when you set up an Azure Cognitive Services account

References

Python SDK, Cognitive Services, Computer Vision: Read printed and handwritten text

https://docs.microsoft.com/en-us/azure/cognitive-services/computer-vision/quickstarts-sdk/python-sdk#read-printed-and-handwritten-text

batch_read_file

https://docs.microsoft.com/en-us/python/api/azure-cognitiveservices-vision-computervision/azure.cognitiveservices.vision.computervision.operations.computervisionclientoperationsmixin?view=azure-python#batch-read-file-url--custom-headers-none--raw-false----operation-config-

batch_read_file_in_stream

https://docs.microsoft.com/en-us/python/api/azure-cognitiveservices-vision-computervision/azure.cognitiveservices.vision.computervision.operations.computervisionclientoperationsmixin?view=azure-python#batch-read-file-in-stream-image--custom-headers-none--raw-false--callback-none----operation-config-