用简单模型+迭代优化，实现多语言文本伪造的精准检测与解释报告生成。
SEED: Simple ViT and Evolving Harness for Explainable Text Forgery Detection
- 分三步：合成数据增强、单个ViT联合检测与定位、大模型迭代生成报告。
- 在挑战赛中排名第三，支持多语言文本图像伪造检测与像素级定位。
- 适合需要可解释性伪造检测的金融、法律等高可信场景使用。
AI辅助图像编辑威胁金融、法律和身份记录的信任。ACM MM 2026的GenText-Forensics挑战要求生成结构化取证报告，包含检测、像素级定位和自然语言解释，针对以文本为中心的伪造图像。我们提出SEED，一个模块化系统：第一，基于相似性的增强管道，通过多样化的合成伪造图像扩充训练数据；第二，采用基于DINOv3的单个ViT模型，结合LoRA适配，联合完成检测与像素级定位，同时以极少可训练参数保留预训练先验；第三，演化调和机制利用多模态大模型（MLLM），根据检测结果生成完整取证报告，并通过提议者-评估者循环迭代优化报告质量。SEED在比赛中获得第三名。代码与数据已公开于https://github.com/KahimWong/GenText-Forensics-3rd-Place。
原文摘要 · Abstract (English)
AI-assisted image editing threatens trust in financial, legal, and identity records. The GenText-Forensics Challenge at ACM MM 2026 addresses this by requiring structured forensic reports, in which integrating detection, pixel-level localization, and natural language explanation for multilingual text-centric forgery images. We present SEED, a modular system with three components. First, a similarity-guided pipeline augments training with diverse synthetic forgeries. Second, a single ViT, built on DINOv3 with LoRA adaptation, jointly performs detection and pixel-level localization while preserving pre-trained priors with minimal trainable parameters. Third, an evolving harness takes the detector's predictions and generates a complete forensic report via an MLLM, iteratively improved through a proposer-evaluator loop optimizing report quality. SEED ranked 3rd in the GenText-Forensics Challenge. Code and data are available at https://github.com/KahimWong/GenText-Forensics-3rd-Place.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。