LHSDet: High-Resolution AI-Generated Image Detection via Visual Question Answering

2026-08-07

AI Summary

This work addresses the limitations of existing AI-generated image detection methods, which often suffer from the loss of low-level texture details due to downsampling and exhibit limited generalization capability. To overcome these challenges, the study formulates high-resolution generated image detection as a visual question answering task and introduces a novel three-branch multimodal architecture. This architecture integrates a custom-designed low-level texture branch, a high-level semantic branch based on SigLIP2, and textual descriptions generated by BLIP-2, enabling an end-to-end differentiable detection model. A hierarchical fusion mechanism is further devised to effectively combine features across low-level, high-level, and semantic modalities. The proposed approach achieves high accuracy and strong robustness on high-resolution images synthesized by diverse diffusion and autoregressive models, significantly enhancing generalization to unseen generative models.

Abstract

Driven by advances in diffusion models and autoregressive models, the fidelity and resolution of AI-generated images now rival those of real images. However, existing AI-generated image detection methods often downsample the images, inevitably overlooking critical low-level texture details in high-resolution AI-generated images, therefore limiting their detection performance. In addition, the ceaseless emergence of unknown generative models makes large-scale pre-training datasets inaccessible. To address these challenges, we propose a novel high-resolution AI-generated image detector, termed LHSDet. Specifically, we formulate the AI-generated image detection task as a Visual Question Answering problem, leveraging a fine-tuned vision-language framework to fully exploit the complementary information between visual and textual modalities. Recognizing that the default visual encoder of existing vision-language models is not tailored for AI-generated image detection, we redesign a visual encoder to better capture both the low-level and high-level artifacts inherent in AI-generated images. Furthermore, we incorporate a semantic-level textual branch to enable multi-modal feature fusion and detection. Consequently, LHSDet employs a triple-branch architecture to extract complementary multi-modal features: a low-level visual branch that aggregates non-overlapping patches for local texture cues, a high-level visual branch based on SigLIP2 for global perception feature extraction, and a semantic-level textual branch that generates captions using BLIP-2. Extensive experimental results demonstrate that LHSDet achieves high detection accuracy and robust performance across diverse generative models, including both diffusion and autoregressive models.

Problem

• AI-generated image detection
• high-resolution images
• low-level texture
• unknown generative models
• visual artifacts

Innovation

• Visual Question Answering
• High-Resolution Image Detection
• Vision-Language Model
• Multi-Modal Fusion
• AI-Generated Image Detection