FileLingo

FileLingo is highly focused on performing OCR from files (like PDFs, images), processing and language translation.

Key Features:

OCR (Optical Character Recognition) Integration: The application can extract text from various image formats (jpg, jpeg, png) and PDFs using Tesseract-OCR and PyMuPDF. Supports multi-page PDFs, ensuring all content is captured and processed.

Multi-Language Summarization: Utilizes the bart-large-cnn model for summarizing extracted text. The summarization is performed in chunks to handle large documents effectively, ensuring concise and relevant summaries.

Language Translation: Translates the summarized text between over 50 languages using Facebook’s mBART-large model. The translation process is capable of handling complex language pairs, making it versatile for various linguistic needs.

Advanced Text Cleaning and Correction: The application includes two stages of text cleaning:

User-Friendly Interface: Built on Streamlit, providing an intuitive and accessible web interface. Features like file upload and language selection are made easy to use, ensuring a smooth user experience.

Chunked Processing for Large Texts: The application intelligently splits large summaries into manageable chunks for translation, ensuring that even lengthy documents are processed without errors.

Support for Multiple File Formats: Handles both PDF and image files, making it a versatile tool for various document types. Capable of processing both text-heavy and image-heavy documents, adapting the OCR and summarization approach accordingly.

Cross-Platform Compatibility: The application can be run on different operating systems (Windows, macOS, Linux) with appropriate Python and library installations.

Scalable and Modular Design: The code is designed to be easily extended, allowing for the integration of additional languages, models, or functionalities as needed. These features make FileLingo a powerful tool for multilingual document processing, offering robust capabilities in OCR, text summarization, and translation within a streamlined application.