21/06/2023
🚀 Exciting breakthrough from Laigo! Our very own Machine Learning Engineer, Merxhan Bajrami, and our CEO, Yvonne Gaissmaier, have been hard at work innovating in the document processing sphere. Today, we're proud to introduce DocSplitAI.
DocSplitAI isn't just any tool - it's a paradigm shift in how we handle document segmentation in large merged PDFs. Leveraging the Donut model's binary classification approach, DocSplitAI has shown impressive proficiency in accurately identifying individual documents within vast PDF conglomerations.
But how does it work? DocSplitAI is an OCR-free model. This means it learns to recognize patterns and features in your documents without relying on optical character recognition. Instead, it has been fine-tuned on a dataset of 5,527 files, encompassing invoices, receipts, and letters in German, and manually labeled into new document and same document classes. With such a robust training regimen, DocSplitAI boasts an accuracy of 0.89, an f1 score of 0.92, a precision of 0.87, and a recall of 0.93. 💼
The implications of DocSplitAI are vast, particularly for industries like the legal and financial sectors, where large volumes of documents are a daily reality. No more labor-intensive, error-prone manual document separation - DocSplitAI streamlines your document processing and enables efficient data analysis. It's time to embrace the future of document management.
We're excited to share that this innovation has been unveiled at the 20th International Conference on Informatics and Information Technologies, CIIT 2023.
And the best part? This is only the beginning. Stay tuned for Part 2 of our research, where we will introduce a more powerful, rule-based system that takes DocSplitAI's capabilities to new heights. The future of document management is here and now. 🌐
And at this moment, you can experience the power of DocSplitAI firsthand. Visit our website https://lnkd.in/dfKxScX9 to explore and utilize this revolutionary tool. Discover how DocSplitAI can transform your document management processes. Embark on your journey towards a more efficient and streamlined workflow today!