Developer & Digital Tools · 🌏 Global

Image Table to Excel OCR

LIVE

Image table to Excel/CSV via client-side Korean tesseract.js OCR.

About this tool

Image Table to Excel OCR (imgocr) extracts tables from images β€” PDF screenshots, scanned receipts / contracts, generic screenshots β€” using Tesseract.js WASM (Korean + English) and exports them as XLSX, CSV, or JSON. All OCR runs in the browser; no images are uploaded (verifiable via DevTools Network). The detected table is editable cell-by-cell. Accuracy is 90–97% on clean scans / screenshots. Supports drag-and-drop, multi-image batch processing, and Korean / English simultaneous recognition. Useful in accounting, HR, research, and journalism whenever you face paper or image tables.

Use cases

Scenario 1

PDF table β†’ Excel

When a protected PDF blocks copy-paste, finance captures the page and converts it straight to Excel for analysis.

Scenario 2

Scanned receipts / contracts

Finance scans a stack of business-trip receipts, OCRs date / amount / vendor, and bulk-loads them into the accounting system.

Scenario 3

Academic / research data

A researcher extracts a statistical table screenshot from a paper PDF and converts it into CSV ready for R or Python analysis.

Scenario 4

Strict security environment

In environments where cloud OCR APIs are forbidden, digitize tables entirely on the client without any external transmission.

Features

  • Tesseract.js WASM Korean + English OCR β€” zero uploads
  • Auto row / column detection with editable result table
  • XLSX / CSV / JSON export
  • Drag-and-drop, multi-image batch processing
  • Pre-processing options (binarization, line removal)
  • Bilingual KR / EN UI

Frequently asked

Q. How accurate is the OCR?
A. Clean scans (β‰₯ 300 DPI) and Hangul screenshots in Myeongjo/Gothic typefaces hit 90–97% cell accuracy. Low-DPI (< 100), skewed photos, or handwriting drop accuracy β€” review the result editor before exporting.
Q. Are images uploaded to a server?
A. No. The OCR engine and Korean / English training data are cached in the browser and run 100% client-side. Verify via the DevTools Network tab.
Q. Can it handle mixed Korean + English?
A. Yes. Both language models load together so the same image yields both Korean and English cells. Japanese / Chinese etc. can be added as options.

Sources / references

Related tools

How we run it / disclaimer

This tool is advisory and does not constitute legal, tax, medical, or financial advice. All calculations and document generation run in your browser; inputs are never sent to a server. Ads follow Google AdSense policy and are kept separate from tool accuracy.