Perception, Verdict, and Evolution: Hindsight-Driven Self-Refining Forensics Agent for AI-Generated Image Detection

📄 arXiv: 2606.26552v1 📥 PDF

作者: Yangjun Wu, Keyu Yan, Yu Liu, Jingren Zhou, Fei Huang, Rong Zhang, Zhou Zhao, Fei Wu

分类: cs.CV, cs.AI

发布日期: 2026-06-25

备注: 10 pages

💡 一句话要点

提出ForeAgent以解决深度伪造图像检测中的灵活性不足问题

🎯 匹配领域: 支柱九：具身大模型 (Embodied Foundation Models)

关键词: 深度伪造检测 多模态融合 自我精炼 生成模型 图像取证

📋 核心要点

  1. 现有深度伪造检测方法对细粒度伪迹的敏感性不足，且依赖静态合成监督，导致灵活性和成本高。
  2. 本文提出ForeAgent框架，采用感知-判决架构和基于回顾的自我精炼策略，提升检测性能。
  3. ForeAgent在Chameleon基准上达到了82.18%的准确率，较AIDE提升16.41%，并在AIGCDetect-Benchmark上表现优异。

📝 摘要（中文）

随着生成模型的快速发展，现有的深度伪造检测方法面临重大挑战，尤其是在高度真实的AI生成图像广泛传播的背景下。尽管多模态大型语言模型（MLLMs）在此任务中展现出强大潜力，但现有方法存在对细粒度取证伪迹敏感性不足和依赖静态合成监督的局限。为了解决这些问题，本文提出了ForeAgent，一个具有迭代自我演化能力的取证框架。ForeAgent采用感知-判决架构，聚合多视角线索，并引入基于回顾的自我精炼策略，持续优化推理过程。实验结果表明，ForeAgent在Chameleon基准上达到了82.18%的准确率，相较于AIDE提升了16.41%。

🔬 方法详解

问题定义：本文旨在解决现有深度伪造图像检测方法在灵活性和对细粒度伪迹敏感性不足的问题。现有方法依赖静态合成监督，导致检测效果受限。

核心思路：提出ForeAgent框架，通过感知-判决架构聚合多视角特征，并引入基于回顾的自我精炼策略，实现持续自我优化。

技术框架：ForeAgent的整体架构包括感知模块、判决模块和自我精炼模块。感知模块负责提取语义、空间和频域特征，判决模块利用多模态大型语言模型进行逻辑判决，自我精炼模块则通过回顾失败案例进行优化。

关键创新：最重要的创新在于引入了基于回顾的自我精炼策略，使得ForeAgent能够在推理过程中不断反思和改进，生成更高质量的推理轨迹。

关键设计：在设计中，采用了双专家质量门控模块对合成样本进行严格过滤，并通过自我策划的高质量样本进行持续微调，确保模型的演化过程高效且有效。

🖼️ 关键图片

fig_0
fig_1
fig_2

📊 实验亮点

ForeAgent在Chameleon基准上达到了82.18%的准确率，相较于AIDE提升了16.41%。在AIGCDetect-Benchmark上，ForeAgent在16种生成器中实现了93.3%的平均准确率，显示出其在深度伪造检测中的优越性能。

🎯 应用场景

该研究的潜在应用领域包括社交媒体内容审核、新闻真实性验证和法律取证等。随着AI生成图像技术的普及，ForeAgent能够为相关领域提供更高效的检测工具，降低虚假信息传播的风险，具有重要的社会价值和实际意义。

📄 摘要（原文）

The rapid advancement of generative models presents a significant challenge to existing deepfake detection methods, particularly given the widespread dissemination of highly realistic AI-generated images. Although Multimodal Large Language Models (MLLMs) show strong potential for this task, existing approaches suffer from two key limitations: insufficient sensitivity to fine-grained forensic artifacts and reliance on static synthetic supervision from frontier models, leading to limited flexibility and high-cost. To address these issues, we propose ForeAgent, an agentic forensics framework for AI-generated image detection with iterative self-evolution. First, ForeAgent adopts a Perception-Verdict architecture that aggregates multi-view cues spanning semantic, spatial, and frequency-domain features, and leverages an MLLM as a verdict module to fuse these signals for a logical-grounded verdict. Second, to enable continual self-improvement, we introduce a Hindsight-Driven Self-Refining strategy following a Sampling-Reflection-Evolution paradigm. The agent performs inference rollouts on training instances. Guided by ground-truth labels as hindsight, it reflects on failure cases and low-quality reasoning trajectories to regenerate higher-quality reasoning traces. These synthesized samples are then strictly filtered through a dual-expert quality gating module. ForeAgent continuously evolves via fine-tuning on self-curated high-quality samples. Extensive experiments demonstrate that ForeAgent achieves state-of-the-art performance on the Chameleon benchmark, reaching 82.18% accuracy (+16.41% over AIDE), and achieves 93.3% mean accuracy on AIGCDetect-Benchmark across 16 generators. In addition, external evaluation shows that ForeAgent produces more consistent and causally grounded reasoning compared to GPT-5 and GPT-5-mini.