Setup OCR API Using FastAPI

Create a virtual environment for our API

It is a best practice to create a virtual environment whenever you create a new project. You can follow one of the two options below:

• Using the Anaconda Prompt:

  • Start the Anaconda Prompt.

  • Run the following command:

    conda create -n env_name python=3.7

You can specify any env_name you like and the desired Python version.

• Activate your virtual environment by running the following command:

    conda activate env_name

• Your environment is now activated and you can install whichever packages you need in this virtual environment.

• Using venv package:

  • Start the Command Prompt.

  • If you do not have venv installed, you need to first install it by running the command shown below:

    pip install venv

• Then, run the following command to create the virtual environment:

    python3 -m venv path\env_name

You can specify any env_name you like. You need to specify the path where you want to store the files for your environment. By default, it will take the current directory.

• Activate your virtual environment by running the following command:

    path\env_name\Scripts\activate

• Your environment is now activated and you can install whichever packages you need in this virtual environment.

Create the default GET route

The first step in creating our REST API is to define a default (home) route that performs a GET operation. Whenever anyone tries to go to the URL:

http://localhost:8000/

We will just give a JSON response as shown below:

{
   "message" : "Visit the endpoint: /api/v1/extract_text to perform OCR."
}

This means that the actual OCR operation is going to perform at the URL:

http://localhost:8000/api/v1/extract_text

This endpoint will accept a request body containing the multipart/form-data, i.e., the images.

Note that we are going to pass multiple images in a single call, and our API should give the response in the form shown below:

{
   "image_name": "extracted_text"
}

Let’s first create the default GET route.

from fastapi import FastAPI

app = FastAPI()

@app.get("/")
def home():
    return {"message": "Visit the endpoint: /api/v1/extract_text to perform OCR."}

We are returning a response if anyone tries to access the route / then the API is going to provide this message to the user.

Create the read_image() function

Now that our GET route is done, we should create a function that will accept the image path and the language, and then perform the OCR operation.

We will be uploading the images on the server and then taking that image location (on the server) to perform OCR.

Let’s create the function and name it read_image().

import pytesseract
import asyncio

async def read_image(img_path, lang='eng'):
    try:
        text = pytesseract.image_to_string(img_path, lang=lang)
        await asyncio.sleep(2)
        return text
    except:
        return "[ERROR] Unable to process file: {0}".format(img_path)

Explanation

• We defined an async function read_image() that accepts the image path (which is uploaded on the server) and the language of the text present in the images.

• We created a try-except block because if any error occurs, we wanted to handle those things in our API as well.

• We performed the OCR using the same statement that was used in the previous article in the series.

• We paused the execution for just 2 seconds. At this point, your complex IO operations can also take place. This is the point where the coroutine switching will take place.

• Finally, we returned our extracted text.

• We also return any error that might have occurred during the extraction of text.

We will continue the next steps in our next article in this series.

This series is just a snapshot of the Build a REST API Using Python and Deploy it to Microsoft Azure course which covers a lot more things like FastAPI, Microsoft Azure, Deploying FastAPI applications to Azure, Monitoring the applications using Azure, and more projects. Do check it out and let me know if you have any questions.