文章总览 - 114
Towards Generalizable Deepfake Detection via Clustered and Adversarial Forgery Learning
Towards Generalizable Deepfake Detection via Clustered and Adversarial Forgery Learning

61
Robust Image Forgery Detection over Online Social Network Shared Images
Robust Image Forgery Detection over Online Social Network Shared Images

发表于CVPR2022。

62
科研工具合集
科研工具合集

63
Robust Camera Model Identification Over Online Social Network Shared Images via Multi-Scenario Learning
Robust Camera Model Identification Over Online Social Network Shared Images via Multi-Scenario Learning

发表于TIFS2023,随着互联网的蓬勃发展,在线社交网络(OSNs, online social networks)已成为图像共享和传输的主导渠道,但OSN传输下所有现有算法的性能都会严重下降,尤其是WeChat、QQ、Telegram和Dingding。为了减轻OSN的负面影响,在本工作中,我们提出了一种新的相机轨迹提取方法,该方法有望对各种OSN平台的传输具有鲁棒性。

64
experimet
experimet

65
Towards Modern Image Manipulation Localization A Large-Scale Dataset and Novel Methods
Towards Modern Image Manipulation Localization A Large-Scale Dataset and Novel Methods

发表于CVPR2024型,CAAA可以像素级自动和精确地注释大量的人工伪造的图像,进一步提出了一种新的度量QES,以方便不可靠注释的自动过滤。

66
Multi-view Feature Extraction via Tunable Prompts is Enough for Image Manipulation Localization
Multi-view Feature Extraction via Tunable Prompts is Enough for Image Manipulation Localization

发表于ACMMM2024,针对IML任务中公共训练数据集的稀缺,通过采用可调提示来利用预训练模型的丰富先验知识,即Prompt-IML框架,即插即用的特征对齐和融合模块。

*现有问题*: IML任务中公共训练数据集的稀缺直接阻碍了模型的性能。 *解决方案*: 提出了一个Prompt-IML框架,该框架通过采用可调提示来利用预训练模型的丰富先验知识。
具体情况 > 通过集成可调提示,从单个预先训练过的主干中提取和调整多视图特征,从而保持性能和鲁棒性 ![image-20240824155623293](../postimages/Multi-view-Feature-Extraction-via-Tunable-Prompts-is-Enough-for-Image-Manipulation-Localization/image-20240824155623293.png) > 特征对齐和融合的FAF模块 ![image-20240824220942200](../postimages/Multi-view-Feature-Extraction-via-Tunable-Prompts-is-Enough-for-Image-Manipulation-Localization/image-20240824220942200.png)
67
DH-GAN:Image manipulation localization via a dual homology-aware generative adversarial network
DH-GAN:Image manipulation localization via a dual homology-aware generative adversarial network

发表于Pattern Recognition 2024,双同源感知生成对抗网络(DH-GAN),选择性金字塔(SAP)校准多尺度特征。

68
IAD
IAD

69
PromptAD
PromptAD

70