biodeep
收藏资源简介:
biodeep数据集是由布加勒斯特大学计算机科学系创建的,旨在评估深度伪造检测器在处理分布外内容时的泛化能力。该数据集包含多种类型的深度伪造媒体,包括图像、视频和音频,以及多模态内容。数据集的创建过程涉及收集和整理现有的深度伪造检测基准数据,并引入新的生成模型生成的内容,以测试检测器的鲁棒性。biodeep数据集主要应用于深度伪造检测领域,旨在解决现有检测器在面对新型生成模型时性能下降的问题。
The BioDeep dataset was developed by the Department of Computer Science, University of Bucharest, with the primary goal of evaluating the generalization capability of deepfake detectors when handling out-of-distribution content. This dataset comprises various types of deepfake media, including images, videos, audio, and multimodal content. The construction of the BioDeep dataset involves collecting and curating existing deepfake detection benchmark datasets, as well as incorporating content generated by novel generative models to test the robustness of detectors. The BioDeep dataset is primarily applied in the deepfake detection research field, aiming to address the performance degradation problem of existing detectors when confronted with new generative models.




