2020 DeeperForensics 真实世界人脸伪造检测挑战赛：方法与结果
计算机视觉与模式识别
2021-02-19 v1 机器学习
摘要
本文报道了2020 DeeperForensics 真实世界人脸伪造检测挑战赛的方法与结果。该挑战赛采用 DeeperForensics-1.0 数据集，这是最广泛的公开可用真实世界人脸伪造检测数据集之一，包含60,000个视频，总计17.6百万帧。模型评估在一个高质量隐藏测试集上在线进行，该测试集具有多来源与多种失真。共有115名参与者注册了比赛，25支队伍提交了有效结果。我们将总结获胜方案，并对潜在研究方向进行讨论。
引用
@article{arxiv.2102.09471,
title = {DeeperForensics Challenge 2020 on Real-World Face Forgery Detection: Methods and Results},
author = {Liming Jiang and Zhengkui Guo and Wayne Wu and Zhaoyang Liu and Ziwei Liu and Chen Change Loy and Shuo Yang and Yuanjun Xiong and Wei Xia and Baoying Chen and Peiyu Zhuang and Sili Li and Shen Chen and Taiping Yao and Shouhong Ding and Jilin Li and Feiyue Huang and Liujuan Cao and Rongrong Ji and Changlei Lu and Ganchao Tan},
journal= {arXiv preprint arXiv:2102.09471},
year = {2021}
}
备注
Technical report. Challenge website: https://competitions.codalab.org/competitions/25228