Handwriting recognition in Azure Computer Vision API from locally stored image file

Question

I am trying to brush up my coding and cloud-computing skills by exploring Azure. I would like to automate some admin tasks that involves deciphering lot of handwritten documents and storing the text electronically.

The Python code below is a merge of two code sources.

1. A blog from Taygan Rifat https://www.taygan.co/blog/2018/4/28/image-processing-with-cognitive-services

2. Microsoft's own demo code over at https://learn.microsoft.com/en-us/azure/cognitive-services/computer-vision/quickstarts/python-hand-text

import json
import os
import sys
import requests
import time
import matplotlib.pyplot as plt
from matplotlib.patches import Polygon
from PIL import Image
from io import BytesIO

subscription_key = 'XX79fdc005d542XXXb5f29ce04ab1cXXX'
endpoint = 'https://handwritng.cognitiveservices.azure.com/'
analyze_url = endpoint + "vision/v3.0/analyze"
text_recognition_url = endpoint + "/vision/v3.0/read/analyze"

image_url = "https://3j2w6t1pktei3iwq0u47sym8-wpengine.netdna-ssl.com/wp-content/uploads/2014/08/Handwriting-sample-Katie.png"

headers = {'Ocp-Apim-Subscription-Key': subscription_key}
data = {'url': image_url}
response = requests.post(
    text_recognition_url, headers=headers, json=data)
response.raise_for_status()

# Extracting text requires two API calls: One call to submit the
# image for processing, the other to retrieve the text found in the image.

# Holds the URI used to retrieve the recognized text.
operation_url = response.headers["Operation-Location"]

# The recognized text isn't immediately available, so poll to wait for completion.
analysis = {}
poll = True
while (poll):
    response_final = requests.get(
        response.headers["Operation-Location"], headers=headers)
    analysis = response_final.json()

    print(json.dumps(analysis, indent=4))

    time.sleep(1)
    if ("analyzeResult" in analysis):
        poll = False
    if ("status" in analysis and analysis['status'] == 'failed'):
        poll = False

polygons = []
if ("analyzeResult" in analysis):
    # Extract the recognized text, with bounding boxes.
    polygons = [(line["boundingBox"], line["text"])
                for line in analysis["analyzeResult"]["readResults"][0]["lines"]]

# Display the image and overlay it with the extracted text.
image = Image.open(BytesIO(requests.get(image_url).content))
ax = plt.imshow(image)
for polygon in polygons:
    vertices = [(polygon[0][i], polygon[0][i + 1])
                for i in range(0, len(polygon[0]), 2)]
    text = polygon[1]
    print(text)
    patch = Polygon(vertices, closed=True, fill=False, linewidth=2, color='y')
    ax.axes.add_patch(patch)
    plt.text(vertices[0][0], vertices[0][1], text, fontsize=20, va="top")


plt.show()

What I would like to do is get some help in modifying the script so that it can work with image files stored locally (instead of using URLs).

Currently I am working around this by spinning up an IIS server on an Azure virtual machine and accessing the URL of the image I want to analyse via HTML. It's a little unwieldy (and somewhat unsecure for my purposes).

Thanks, WL

Answer 1

Here you go,

...
# You could also read the image file name from command line
# as the first argument passed to your script:

# try:
#    input_image = sys.argv[1]
# except:
#    sys.exit('No input. Pass input image file name as first argument.')

input_image = "your_input_image.jpg"
with open(input_image, 'rb') as f:
    data = f.read()
    headers = {
        'Ocp-Apim-Subscription-Key': subscription_key,
        'Content-type': 'application/octet-stream'
    }
    response = requests.post(
        text_recognition_url, headers=headers, data=data)
    response.raise_for_status()
...

and later on,

# Display the image and overlay it with the extracted text.
image = Image.open(input_image)
...

Most Azure Cognitive Services that accept an image URL also accept raw bytes as Content-type: application/octet-stream and binary image data as POST payload.

See Analyze image.

Supported input methods:

raw image binary or image URL.

Content-type:

Input requirements:

Supported image formats: JPEG, PNG, GIF, BMP.
Image file size must be less than 4MB.
Image dimensions must be at least 50 x 50.

By the way, if you ever need a quick web server for a future task, Python has your back:

# usage:
# python3 -m http.server [-h] [--cgi] [--bind ADDRESS]
#                        [--directory DIRECTORY] [port]

$ python3 -m http.server
Serving HTTP on :: port 8000 (http://[::]:8000/) ...