---
title: OCR Reader
emoji: 🐢
colorFrom: green
colorTo: pink
sdk: gradio
sdk_version: 6.29.1
python_version: '3.12'
app_file: app.py
pinned: false
license: mit
short_description: Give image of text and get back the text the model reads.
---

# OCR Reader

Upload an image of printed text and get back the text the model reads. No code needed: upload, wait a few seconds, read the result.

## Model
[microsoft/Florence-2-large](https://huggingface.co/florence-community/Florence-2-large), run with the `<OCR>` task.

## Why this model
It is open (MIT license), reads whole images rather than single cropped lines, and is fast enough to run on a free ZeroGPU Space.

## Tests
| Input | Result |
|---|---|
| Clean screenshot of printed text | (fill in) |
| Slightly angled or dim photo of a page or sign | (fill in) |
| Handwriting (expected to fail) | (fill in) |

## Limitations
Handwriting, very stylized fonts, and tiny or blurry text are likely to be misread.

Check out the configuration reference at https://huggingface.co/docs/hub/spaces-config-reference