OpenForensics
收藏资源简介:
OpenForensics是由国家信息学研究所创建的大型数据集,专注于多面伪造检测和分割。该数据集包含115,325张不受限制的图像,涵盖多种背景和多个人物,年龄、性别、姿势、位置和面部遮挡各异。数据集通过丰富的面部级标注支持多种任务,如伪造类别、边界框、分割掩码、伪造边界和一般面部地标。OpenForensics不仅支持多面伪造检测和分割任务,还适用于一般的面部相关任务,旨在推动深度伪造预防和一般面部检测的研究。
OpenForensics is a large-scale dataset created by the National Institute of Informatics, focusing on multi-face forgery detection and segmentation. This dataset contains 115,325 unconstrained images, covering diverse backgrounds and multiple subjects with varying ages, genders, poses, positions, and facial occlusions. The dataset supports a wide range of tasks via rich facial-level annotations, including forgery categories, bounding boxes, segmentation masks, forgery boundaries, and general facial landmarks. OpenForensics not only supports multi-face forgery detection and segmentation tasks but also applies to general facial-related tasks, aiming to advance research in deepfake prevention and general facial detection.




