PS-Battles数据集
收藏资源简介:
PS-Battles数据集是由巴塞尔大学数学与计算机科学系创建的一个图像集合,专门用于视觉领域中的媒体衍生和篡改检测。该数据集包含102,028张图像,分为11,142个子集,每个子集包含原始图像及其多个篡改后的衍生图像。数据集的内容来源于reddit上的photoshopbattles社区,该社区是一个活跃的数字艺术家社区,定期举办数字图像篡改比赛。数据集的创建过程涉及从reddit内容中提取所有相关的帖子和评论,并筛选出高质量的篡改图像。PS-Battles数据集的应用领域主要集中在图像篡改检测和衍生检测,旨在解决数字媒体中的篡改问题,提供一个可靠的基准数据集,以促进自动化篡改检测算法的研究和发展。
PS-Battles is an image collection developed by the Department of Mathematics and Computer Science, University of Basel, specifically tailored for media-derived and tampering detection in the visual domain. This dataset comprises 102,028 images divided into 11,142 subsets, with each subset containing an original image and multiple tampered derivative versions. The content of the PS-Battles dataset is sourced from the PhotoshopBattles community on Reddit, an active digital artist community that regularly hosts digital image tampering contests. The creation process of the PS-Battles dataset involves extracting all relevant posts and comments from Reddit content, followed by filtering out high-quality tampered images. The primary application scenarios of the PS-Battles dataset focus on image tampering and derivative detection, aiming to address the forgery problem in digital media and provide a reliable benchmark dataset to facilitate the research and development of automated tampering detection algorithms.

- 1The PS-Battles Dataset - an Image Collection for Image Manipulation Detection巴塞尔大学数学与计算机科学系 · 2018年



