KataScan OCR | Batch Text Extraction from Forms

Detailed instructions

1. Load files

Select an image (JPG/PNG) or PDF using the file picker.
You can select multiple files at once and use multi-page PDFs(up to 15 pages)。
When you load another file with the same layout, the fields from the previous file are retained.
Note: Images are internally upscaled to improve OCR accuracy. Loading may take longer when a file has many pages.

2. Set reading fields (up to 10)

On the canvas, point to the text you want to read anddrag with the mouseto draw a rectangular field.
Each field you create adds its settings to the legend above.
Choose the recognition language for the text, such as Japanese or alphanumeric only. Selecting alphanumeric only can reduce misrecognition.

3. Adjust fields and save a format

To redraw a field, select “Redraw” in the legend and drag again.
Columns in the results table follow the numbered order of the fields you created.
Enter unwanted characters, such as ¥, commas, or a currency unit, in “Characters to exclude” to remove them automatically.
If unwanted spaces appear, select “Remove all spaces”.
New“Format settings (JSON)”to download the field positions and excluded-character settings as a file on your device. Next time, choose the JSON file with “Load from file” to restore the fields instantly.

4. Run OCR, edit, and save results

Select “Run OCR” to recognize text on every page.
When processing finishes, the results table appears below.Click a yellow cell to edit its text directly.
NewWhen you select a cell, a preview of the correspondingoriginal image appears in a pop-up just above itso you can compare and proofread it easily.
After checking the results, save the data using“Download CSV” below the tableor use the control above to“Copy table” to paste directly into Excel or a spreadsheet with Ctrl+Vfor convenience.

5. Tips

Recommended environment: This tool is designed for desktop use. Use the latest version of Google Chrome, Microsoft Edge, or Safari. It may not work correctly on phones or tablets.
This tool is not well suited to handwritten text.
If numbers are not recognized correctly, redrawing the field may improve the result.
If OCR results below contain unwanted characters or symbols, copy them into the “Characters to exclude” field above and run OCR again to reduce manual cleanup.

Offline operation and privacy

All processing in this app, including image upscaling, image enhancement, and AI-based OCR,runs entirely on your computer in the browser without sending any data to an external server.

This helps protect forms containing confidential or personal information. In addition, once this page has loaded,all features continue to work offline even if you disconnect from the Internet.

Note: Because processing uses your PC instead of a server, it can use substantial memory and CPU. If it becomes slow, select “Reset” or reload the browser tab.

⚖️ Disclaimer:
This app uses AI image analysis for OCR and cannot guarantee 100% accuracy. Always review the recognized text yourself and correct it manually when needed.
The developers and provider are not liable for direct or indirect damages, data loss, or business issues arising from use of this app. Please use it at your own risk.

Tesseract.js (Apache License 2.0) / PDF.js (Apache License 2.0) / OpenCV.js (Apache License 2.0) / Tailwind CSS (MIT License)