OCR API with bounding boxes

Here are some good OCR APIs that provide bounding boxes for detected text:

Cloud OCR APIs

1. Google Cloud Vision OCR

   • Returns text annotations with bounding polygons/bounding boxes
   • Good for: scanned documents, photos, handwriting
   • Output: words, lines, paragraphs with coordinates

2. AWS Textract

   • Returns bounding boxes for words, lines, forms, tables
   • Good for: documents, invoices, forms
   • Output: structured JSON with box coordinates

3. Azure AI Vision / Document Intelligence

   • OCR with bounding polygons
   • Good for: printed text, documents, layouts
   • Output: words, lines, layout elements with coordinates

Specialized OCR APIs

4. Mathpix

   • Great for: equations, scientific documents
   • Returns text, LaTeX, and bounding boxes

5. ABBYY Cloud OCR SDK

   • Strong document OCR
   • Returns detailed layout and coordinates

6. OCR.Space

   • Simple API, returns bounding boxes
   • Good for: lightweight use cases

Open-source / self-hosted options

7. Tesseract OCR

   • Can output bounding boxes via hOCR, TSV, or XML
   • Good for: local/offline OCR
   • Not as accurate as top cloud services in many cases

8. PaddleOCR

   • Modern OCR with bounding boxes
   • Good accuracy, especially on varied text and some multilingual use cases

If you want, I can also give you:

• a comparison table
• sample API responses with bounding boxes
• or a recommended choice based on your use case (receipts, forms, screenshots, books, etc.).