无需训练即可检测生成视频，基于时空统计差异实现高精度识别。
Training-free Detection of Generated Videos via Spatial-Temporal Likelihoods
- 构建联合建模空间与时间特征的似然评分框架
- 在新数据集ComGenVid上超越现有基线模型
- 适合需要快速部署、适配新生成器的检测场景
随着文本与图像生成技术的突破，视频生成领域快速发展，涌现出高度逼真且可控的序列内容。然而，这也带来了严重的信息误导风险，因此可靠地检测合成视频变得愈发重要。现有的基于图像的检测方法仅逐帧分析，忽略时序动态；而有监督的视频检测器对未见生成器泛化能力差，难以应对新模型不断涌现的现状。为此，本文提出无训练（training-free）的零样本检测方法STALL：它不依赖合成数据，而是通过真实数据统计建模，实现无需训练、模型无关的检测。STALL在概率框架下联合建模空间与时间证据，提供基于似然的视频评分。我们在两个公开基准和新提出的ComGenVid数据集（包含前沿生成模型）上评估该方法，结果表明其持续优于现有的图像与视频基线。代码与数据已公开于https://omerbenhayun.github.io/stall-video。
原文摘要 · Abstract (English)
Following major advances in text and image generation, the video domain has surged, producing highly realistic and controllable sequences. Along with this progress, these models also raise serious concerns about misinformation, making reliable detection of synthetic videos increasingly crucial. Image-based detectors are fundamentally limited because they operate per frame and ignore temporal dynamics, while supervised video detectors generalize poorly to unseen generators, a critical drawback given the rapid emergence of new models. These challenges motivate zero-shot approaches, which avoid synthetic data and instead score content against real-data statistics, enabling training-free, model-agnostic detection. We introduce STALL, a simple, training-free, theoretically justified detector that provides likelihood-based scoring for videos, jointly modeling spatial and temporal evidence within a probabilistic framework. We evaluate STALL on two public benchmarks and introduce ComGenVid, a new benchmark with state-of-the-art generative models. STALL consistently outperforms prior image- and video-based baselines. Code and data are available at https://omerbenhayun.github.io/stall-video.
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。