README.md

metadata

title: Handwriting Detection
emoji: 🐢
colorFrom: gray
colorTo: gray
sdk: streamlit
sdk_version: 1.43.2
app_file: app.py
pinned: false
license: mit
short_description: Streamlit app for handwritten text recognition using TrOCR.

Handwritten Text Recognition with TrOCR

This project is a Streamlit-based web application that recognizes handwritten text from images using Microsoft's TrOCR model.

Features

• Upload an image file or provide an image URL
• Recognize handwritten text using the TrOCR model
• Display the uploaded image and extracted text

Installation

To run this application, follow these steps:

1. Clone this repository:

   git clone https://github.com/your-username/your-repo.git
   cd your-repo
   

2. Install dependencies:

   pip install -r requirements.txt
   

Usage

Run the Streamlit application with:

streamlit run app.py

Dependencies

This project requires the following Python libraries:

• streamlit
• transformers
• PIL (Pillow)
• requests
• torch

To install them, run:

pip install streamlit transformers pillow requests torch

Model Details

This application uses the microsoft/trocr-base-handwritten model from the Hugging Face Transformers library. It processes images and extracts handwritten text using an encoder-decoder architecture.

Example

You can test the app using the following example image:

https://fki.tic.heia-fr.ch/static/img/a01-122-02-00.jpg

Simply paste the URL into the input field or upload your own handwritten image.

License

This project is licensed under the MIT License.