Classification of health insurance documents

1. Classification of health insurance documents with image URL or pdf URL input

API:

Params:

Demo Python:

import requests

api_key = "YOUR_API_KEY"
api_secret = "YOUR_API_SECRET"

image_url = 'https://example.com/image.png'

response = requests.get(
  "https://demo.computervision.com.vn/api/v2/ocr/document/get_label_claim?img=%s&format_type=url&get_thumb=false"
  % image_url,
  auth=(api_key, api_secret))

print(response.json())

2. Classification of health insurance documents with image file or pdf file input

API:

Params:

Body:

Demo Python:

import requests

api_key = "YOUR_API_KEY"
api_secret = "YOUR_API_SECRET"
image_path = '/path/to/your/image.jpg'

response = requests.post(
  "https://demo.computervision.com.vn/api/v2/ocr/document/get_label_claim?format_type=file&get_thumb=false",
  auth=(api_key, api_secret),
  files={'img': open(image_path, 'rb')})

print(response.json())

3. Classification of health insurance documents with JSON input

API:

Params:

Body:

{
  "img": "iVBORw0KGgoAAAANSU..." // string base64 of the image or pdf to extract
}

Demo Python:

import base64
import io
import requests
from PIL import Image
def get_byte_img(img):
    img_byte_arr = io.BytesIO()
    img.save(img_byte_arr, format='PNG')
    encoded_img = base64.encodebytes(img_byte_arr.getvalue()).decode('ascii')
    return encoded_img
api_key = "YOUR_API_KEY"
api_secret = "YOUR_API_SECRET"
img_name = "path_img"
encode_cmt = get_byte_img(Image.open(img_name))
response = requests.post(
    "https://demo.computervision.com.vn/api/v2/ocr/document/get_label_claim?format_type=base64&get_thumb=false",
    auth=(api_key, api_secret),
    json={'img' : encode_cmt})
print(response.json())

4. Response

The response will be a JSON with the following format:

{
  "data": array,
  "errorCode": string,
  "errorMessage": string
}

Each element in the data array (corresponding to each extracted page) will be a JSON with the following format:

{
  "id": [xxxx],
  "label": [xxxx],
  "image": [xxxx]
}

In there:

• id: The page order corresponds to the input file, starting from 0.
• label: Label of the respective page. Currently 17 labels are supported in the insurance profile, if it does not fall into one of the 17 defined labels, label will return None.
  • invoice
  • list_expense
  • claim_form
  • hospital_discharge_paper
  • id_doc
  • prescription
  • medical_report
  • discharge_report
  • bill
  • surgical_certificate
  • specify_vote
  • test_results
  • medical_examination
  • receipts
  • health_records
  • guarantee_confirmation
  • accident_report
• image

Error code table: