ocr-document-classification
v1.3.7
Published
Document classification using tesseract.js and string-similarity-js.
Downloads
25
Maintainers
Readme
OCR Document Classification
Overview
The OCR Document Classification package provides a utility to classify documents based on their content. It uses OCR (Optical Character Recognition) to extract text from images and then determines the document type by matching extracted words with predefined target words using string similarity.
Installation
To install this package, use npm:
npm install ocr-document-classification
Usage
The main function exported by this package is classifyDocument
. Below is a detailed guide on how to use it.
Importing the Package
import { classifyDocument } from "ocr-document-classification";
import type { documentDictionary } from "ocr-document-classification";
Function: classifyDocument
Parameters
- file: The image file (File object) of the document to be classified.
- options (optional): An object containing the following optional properties:
- onProgress: A callback function to receive progress updates. It accepts a number between 0 and 100.
- customDocumentDictionary: An object containing custom document types and their associated target words.
- maxNumPages: A number specifying the maximum number of pages to process. Defaults to Infinity.
Returns
A Promise that resolves with an object containing:
classification
: The determined document type.text
: The extracted text from the document.
Classes
There exists a couple of default classes that can be useful to classify the most common documents. As you can see there exists multiple arrays for each key. This means that every word of only ONE of the arrays needs to be found in the document after OCR. You can also add your own class my creating a customDocumentDictionary.
const defaultDocumentDictionary: documentDictionary = {
MILITÆRBEVIS: [
["førstegangstjeneste", "bevis", "avtjent"],
["attest", "førstegangstjeneste"],
["fullført", "førstegangstjeneste"],
],
POLITIATTEST: [["politiattest", "politidistrikt"], ["police certificate"]],
KOMPETANSEBEVIS: [["omfatter", "opplæring", "utdanningsprogram"]],
LEGEERKLÆRING: [["legeerklæring", "fødselsnummer"]],
BOSTEDSATTEST: [
["registrerte", "opplysninger", "folkeregisteret"],
["bostedsattest", "bostedsadresse", "registrert"],
["registrert", "adressehistorikk", "folkeregisteret"],
],
};
Example
Here is an example of how to use the package can be used with a custom document dictionary in React:
import React, { useState, useEffect } from "react";
import { classifyDocument } from "ocr-document-classification";
function UploadClassification() {
const [documentFile, setDocumentFile] = useState<File | null>(null);
const [classification, setClassification] = useState("");
const [outputText, setOutputText] = useState("");
const [progress, setProgress] = useState(0);
const handleFileChange = (event: React.ChangeEvent<HTMLInputElement>) => {
const file = event.target.files && event.target.files[0];
setDocumentFile(file);
};
const customDocumentDictionary = {
Jobbsøknad: [["søknad", "stilling", "ledig"]],
};
useEffect(() => {
console.log("Progress: ", progress);
}, [progress]);
useEffect(() => {
if (documentFile) {
classifyDocument(documentFile, {
onProgress: setProgress,
customDocumentDictionary: customDocumentDictionary,
})
.then(({ classification, text }) => {
setClassification(classification);
setOutputText(text);
})
.catch((err) => {
console.error(err);
setOutputText("Error during OCR processing");
});
}
resetOCR();
}, [documentFile]);
function resetOCR() {
setClassification("");
setOutputText("");
setProgress(0);
}
return (
<>
<input
accept="image/jpeg, image/png"
type="file"
onChange={handleFileChange}
/>
<div>
<h3>Resultat av OCR</h3>
<p>{classification ? outputText : "Laster inn ..."}</p>
<h1>{classification}</h1>
</div>
</>
);
}
export default UploadClassification;
Dependencies
This package relies on the following dependencies:
string-similarity-js
: For calculating the similarity between strings.tesseract.js
: For performing OCR on the document image.pdfjs-dist
: For handling PDFs
LICENSE
This package is currently UNLICENSED.