Member-only story
New Vision-Language Solution for Extracting Tables and Figures from PDFs
As researchers and technical writers, we often find ourselves sifting through countless research papers, trying to extract the most relevant information to include in our own work.
One of the most time-consuming and challenging tasks in this process is identifying and extracting the tables and figures from these papers. Traditional methods of manually combing through PDFs or using basic text extraction tools often fall short, leaving us frustrated and wishing for a more efficient solution.
Fortunately, the recent advancements in computer vision and natural language processing have given rise to a new and powerful tool that can significantly improve the way we approach this problem.
The new solution
Introducing the TF-ID (Table/Figure Identifier) model, a fine-tuned vision language model that can do a much better job detecting and extracting tables and figures from research papers with remarkable precision, comparing with previous methods.

The TF-ID model is built upon the foundation of a pre-trained vision-language model called Florence-2, which was recently released by Microsoft. This powerful model has been fine-tuned by the researcher to specialize in the task of identifying tables and figures in academic papers. What sets TF-ID apart from other existing solutions is its ability to not only …
Curious to delve deeper into this?
Join Professor Mehdi and myself for a deep-dive discussion about this new approach:
👇
What You’ll Learn:
🔎 The very practical python library that you can use to achieve the goal.
🛠 Walk through of the Colab code that does the job!
🚀 Use cases, branding, and our new product launch!
Stay tuned as we continue exploring the development of knowledge-augmented AI systems to extract maximum value from…