FigEx2: Visual-Conditioned Panel Detection and Captioning for Scientific Compound Figures

📰 ArXiv cs.AI

FigEx2 is a visual-conditioned framework for detecting and captioning panels in scientific compound figures

advanced Published 31 Mar 2026

Action Steps

Detect panels in scientific compound figures using visual features
Generate panel-wise captions based on the detected panels
Train the model using a large dataset of labeled compound figures
Evaluate the performance of FigEx2 using metrics such as precision, recall, and F1-score

Who Needs to Know This

Researchers and developers in AI and computer vision can benefit from FigEx2, as it improves the accessibility and understanding of scientific figures, while data scientists and analysts can utilize the generated captions for further analysis

Key Insight

💡 FigEx2 can automatically generate captions for scientific compound figures, improving their accessibility and usability

Key Takeaways

FigEx2 is a visual-conditioned framework for detecting and captioning panels in scientific compound figures

Full Article

Title: FigEx2: Visual-Conditioned Panel Detection and Captioning for Scientific Compound Figures

Abstract:
arXiv:2601.08026v4 Announce Type: replace-cross Abstract: Scientific compound figures combine multiple labeled panels into a single image. However, in a PMC-scale crawl of 346,567 compound figures, 16.3% have no caption and 1.8% only have captions shorter than ten words, causing them to be discarded by existing caption-decomposition pipelines. We propose FigEx2, a visual-conditioned framework that localizes panels and generates panel-wise captions directly from the image, converting otherwise un

Read full paper → ← Back to Reads