COVE: COntext and VEracity prediction for out-of-context images

2025-02-03

AI Summary

Multimodal misinformation arises from the disconnection between images and their textual descriptions—specifically, the absence of authentic contextual metadata (e.g., time, location, event) and the difficulty in verifying caption authenticity. Method: We propose the first decoupled two-stage framework: Stage I precisely reconstructs the image’s original context; Stage II leverages this context to verify caption veracity. Context reconstruction and factual verification are explicitly separated to ensure module interpretability and result reusability—i.e., one reconstructed context can support verification of multiple captions. We adopt a unified multimodal large language model architecture with joint fine-tuning for image-text alignment, context generation, and binary verification. Results: Our method outperforms state-of-the-art approaches by 5.1% (average) in context prediction accuracy; achieves significantly higher caption authenticity classification accuracy on real-world data; and receives strong human evaluation scores for context interpretability and practical reusability.

Technology Category

• Computer Vision: Multi-modal Vision
• Natural Language Processing: Fact-Checking / Misinformation Detection (NLP Focus)
• Machine Learning: Multimodal Learning

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Web Mining and Content Analysis: Mining multimedia, multimodal, multilingual, cross-lingual Web data
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs

Abstract

Images taken out of their context are the most prevalent form of multimodal misinformation. Debunking them requires (1) providing the true context of the image and (2) checking the veracity of the image's caption. However, existing automated fact-checking methods fail to tackle both objectives explicitly. In this work, we introduce COVE, a new method that predicts first the true COntext of the image and then uses it to predict the VEracity of the caption. COVE beats the SOTA context prediction model on all context items, often by more than five percentage points. It is competitive with the best veracity prediction models on synthetic data and outperforms them on real-world data, showing that it is beneficial to combine the two tasks sequentially. Finally, we conduct a human study that reveals that the predicted context is a reusable and interpretable artifact to verify new out-of-context captions for the same image. Our code and data are made available.

Problem

• Detecting out-of-context image misinformation
• Predicting true image context automatically
• Verifying caption veracity using predicted context

Innovation

• Predicts true context of out-of-context images
• Uses predicted context to verify caption veracity
• Sequentially combines context prediction and veracity checking