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

Hindi & English OCR with Keyword Search

This project implements a web-based prototype for Optical Character Recognition (OCR) on images containing text in both Hindi and English. It also includes a basic keyword search functionality based on the extracted text.

Features

Upload and process images containing Hindi and English text
Extract text from images using OCR
Perform keyword search on the extracted text
Web-based interface for easy interaction

Technology Stack

Python
Hugging Face Transformers (Qwen2-VL-2B-Instruct model)
PyTorch
Gradio (for web interface)

Setup and Installation

Clone the repository:

git clone [your-repo-url]
cd [your-repo-name]

Install the required dependencies:

pip install transformers torch gradio Pillow

Download the Qwen2-VL-2B-Instruct model: The model will be automatically downloaded when you run the application for the first time.

Usage

Run the application:
```
python app.py
```
Open the provided URL in your web browser.
Upload an image containing Hindi and/or English text.
(Optional) Enter a keyword to search within the extracted text.
View the OCR results and any keyword matches.

Limitations

The current implementation uses CPU for processing, which may be slower for large images.

Future Improvements

Implement GPU support for faster processing
Add support for multiple image uploads
Enhance the user interface for better user experience

Link

https://huggingface.co/spaces/pranshh/ocr-assignment

Acknowledgements

This project uses the Qwen2-VL-2B-Instruct model from Hugging Face Transformers.