Pair Viewer

A web app for viewing images and corresponding OCR outputs side-by-side

Pair Viewer is an interactive web application that streamlines the review of digitized documents by presenting original images and OCR outputs side-by-side. I developed this tool while working with a colleague on experimental OCR techniques for transcribing handwriting and extracting text from low-quality images.

Visit the app at: https://pair-viewer.bowrey.info/

How Pair Viewer Works

Step 1: Upload Your Files

Start by uploading two zip files: one containing your image files and another containing your corresponding text files. The application supports standard image formats (.jpg, .jpeg, .png) and text formats (.md, .txt).

Upload interface showing two file upload sections for images and markdown files

Step 2: Review Pairing Summary

Files are intelligently paired by their filename stems. For example, document_001.jpg will be matched with document_001.md. The application displays a summary of the paring process. This shows:

  • Matched Pairs: The number of successfully paired image-text combinations
  • Unmatched Images: Any images without corresponding text files
  • Unmatched Markdown: Any text files without corresponding images
Pairing summary showing 40 matched pairs, 0 unmatched images, and 0 unmatched markdown files

Step 3: Review in the Viewer

Navigate sequentially through your paired files using the Previous and Next buttons.

Side-by-side viewer showing a handwritten historical document on the left and its OCR transcription on the right
Another document pair showing handwritten text on the left and corrected transcription on the right
Third document pair