Decoder. plain-English AI glossary

OCR

● Core

Converting images of text (scans, photos) into machine-readable text strings.

Think of it like

Transcribing a handwritten letter so you can search and edit it.

Example

Scan a receipt; OCR extracts "Total: $42.50, Date: 2025-01-16"; you can now search for this expense in your records.

How it actually works

Combines vision (detect text regions) and sequence modeling (recognize characters). Classic approaches (Tesseract) struggled with curved text and low quality; modern deep learning (CRAFT + attention-based sequence-to-sequence) is more robust. End-to-end models learn bounding box detection and character recognition jointly.

For product teams

Unlocks automation: invoice processing, document search, accessibility (read aloud).

For engineers

Text detection network (YOLO-style) finds text boxes; character recognition network (sequence-to-sequence) decodes characters within each box.

Related

Read anything AI without the jargon

Look up any term in plain English, or save terms as you read with the free Chrome extension.

Open DecoderAdd to Chrome