Survey on Vision based Fake News Detection and its Impact Analysis
Survey on Vision based Fake News Detection and its Impact Analysis
复制标题
DOI：
10.23919/apsipaasc55919.2022.9980089
复制
发表时间：
2022-11
期刊：
影响因子：
--
通讯作者：
M. Raval;Mohendra Roy;M. Kuribayashi
中科院分区：
文献类型：
--
作者：
M. Raval;Mohendra Roy;M. Kuribayashi
虚假新闻是包含多媒体内容并歪曲其所报道事件的帖子。大量假新闻检测（FND）技术依赖于文本分析，而很少关注基于视觉内容的检测。基于深度学习的生成模型通过创建超现实的虚假媒体内容增加了复杂性，并且大多数检测技术在处理此类合成媒体时都会失败。因此，本文对基于视觉的 FND 进行了调查，并提高了我们对视觉内容在检测中的作用的理解。同时，研究假新闻的传播特征并找到影响分析也很重要。现有的综述论文没有结合检测和影响分析两部分。因此，本文提出的调查论文重点关注面部操纵的 FND，并对其影响进行分析。
Fake news is a post containing multimedia content and misrepresents the event that it is covering. A large number of fake news detection(FND) techniques rely on text analysis and very little attention has been paid to visual content-based detection. Deep learning-based generative models have increased the complexity by creating ultra-realistic phony media content and most detection techniques fail when dealing with such synthesized media. Therefore, the paper surveys the vision-based FND and improves our understanding of the role of visual content in detection. At the same time, it is important to study propagation characteristics and find an impact analysis of fake news. The existing review papers do not combine two parts - detection and impact analysis. Therefore, the proposed survey paper focuses on FND with face manipulation and also performs its impact analysis.