rocketcoder/florence-2-lg-ocr ❓🖼️✓ → 🖼️

▶️ 85 runs 📅 Jan 2026 ⚙️ Cog 0.16.9 🔗 GitHub 📄 Paper ⚖️ License
document-to-json ocr

Performance

19.3sTypical run time
~241sCold start (first call)
85Total runs

About

Vision Model that excels at batch OCR processing

Example Output

Output

Example output

Performance Metrics

19.31s Prediction Time
240.88s Total Time
All Input Parameters 
{
  "task": "OCR",
  "images": "https://replicate.delivery/pbxt/OLkpoxyuhASeHNRUGJnmAg6Si2iOqYWTA9ZAJOOUPwudrWw4/test_pages.zip",
  "include_image": false
}
Input Parameters
task Default: OCR
Task
images (required) Type: string
ZIP file containing images
include_image Type: booleanDefault: false
Include annotated output image
Output Schema

Output

Type: string • Format: uri

Example Execution Logs 
[Predict] Extracting zip file: /tmp/tmpl5r_4t57test_pages.zip...
[Predict] Found 4 images.
❌ Error on ._page1.png: cannot identify image file 'temp_processing/._page1.png'
❌ Error on ._page2.png: cannot identify image file 'temp_processing/._page2.png'
Version Details
Version ID
45b6934ad7c32346369da6019ee869d1e30de8bc9f3b2b35545c1f3a7added61
Version Created
January 3, 2026
Run on Replicate →