Sens-VisualNews: A Benchmark Dataset for Sensational Image Detection

2026-05-11

AI Summary

This work proposes a novel task—sensational image detection—aimed at identifying visually sensational content in media imagery that is designed to capture attention and evoke strong emotional responses, thereby supporting misinformation detection. To facilitate research in this area, the authors introduce Sens-VisualNews, a benchmark dataset comprising 9,576 news images annotated with fine-grained labels across multiple sensational concepts and events. Leveraging multimodal large language models, the study systematically evaluates model sensitivity, performance, and robustness under both zero-shot and fine-tuned settings. Experimental results reveal significant limitations of current models in recognizing sensational visual content, highlighting a critical gap in existing research. This work not only establishes the first benchmark for sensational image detection but also provides foundational insights and directions for future investigations in this emerging domain.

Abstract

The detection of sensational content in media items can be a critical filtering mechanism for identifying check-worthy content and flagging potential disinformation, since such content triggers physiological arousal that often bypasses critical evaluation and accelerates viral sharing. In this paper we introduce the task of sensational image detection, which aims to determine whether an image contains shocking, provocative, or emotionally charged features to grab attention and trigger strong emotional responses. To support research on this task, we create a new benchmark dataset (called Sens-VisualNews) that contains 9,576 images from news items, annotated based on the (in-)existence of various sensational concepts and events in their visual content. Finally, using Sens-VisualNews, we study the prompt sensitivity, performance and robustness of a wide range of open SotA Multimodal LLMs, across both zero-shot and fine-tuned settings.

Problem

• sensational image detection
• disinformation
• emotional response
• media content filtering
• visual sensationalism

Innovation

• sensational image detection
• benchmark dataset
• multimodal LLMs
• visual disinformation
• emotional content analysis