Multimodal Graph Learning for Deepfake Detection

计算机科学 稳健性(进化) 编码 人工智能 图形 地标 探测器 模式识别(心理学) 机器学习 理论计算机科学 生物化学 电信 基因 化学
作者
Zhiyuan Yan,Peng Sun,Yubo Lang,Shuo Du,Shanzhuo Zhang,Wei Wang,Lei Liu
出处
期刊:Cornell University - arXiv 被引量:1
标识
DOI:10.48550/arxiv.2209.05419
摘要

Existing deepfake detectors face several challenges in achieving robustness and generalization. One of the primary reasons is their limited ability to extract relevant information from forgery videos, especially in the presence of various artifacts such as spatial, frequency, temporal, and landmark mismatches. Current detectors rely on pixel-level features that are easily affected by unknown disturbances or facial landmarks that do not provide sufficient information. Furthermore, most detectors cannot utilize information from multiple domains for detection, leading to limited effectiveness in identifying deepfake videos. To address these limitations, we propose a novel framework, namely Multimodal Graph Learning (MGL) that leverages information from multiple modalities using two GNNs and several multimodal fusion modules. At the frame level, we employ a bi-directional cross-modal transformer and an adaptive gating mechanism to combine the features from the spatial and frequency domains with the geometric-enhanced landmark features captured by a GNN. At the video level, we use a Graph Attention Network (GAT) to represent each frame in a video as a node in a graph and encode temporal information into the edges of the graph to extract temporal inconsistency between frames. Our proposed method aims to effectively identify and utilize distinguishing features for deepfake detection. We evaluate the effectiveness of our method through extensive experiments on widely-used benchmarks and demonstrate that our method outperforms the state-of-the-art detectors in terms of generalization ability and robustness against unknown disturbances.
最长约 10秒,即可获得该文献文件

科研通智能强力驱动
Strongly Powered by AbleSci AI
更新
大幅提高文件上传限制,最高150M (2024-4-1)

科研通是完全免费的文献互助平台,具备全网最快的应助速度,最高的求助完成率。 对每一个文献求助,科研通都将尽心尽力,给求助人一个满意的交代。
实时播报
虚竹完成签到 ,获得积分10
刚刚
刚刚
1秒前
2秒前
3秒前
聚乙二醇发布了新的文献求助10
3秒前
kellymmbaby完成签到,获得积分10
4秒前
mbf发布了新的文献求助150
5秒前
5秒前
5秒前
mbf发布了新的文献求助10
6秒前
mbf发布了新的文献求助10
6秒前
6秒前
mbf发布了新的文献求助10
7秒前
mbf发布了新的文献求助10
7秒前
hahahah完成签到,获得积分20
7秒前
mbf发布了新的文献求助10
7秒前
infinity发布了新的文献求助10
7秒前
mbf发布了新的文献求助10
7秒前
7秒前
7秒前
7秒前
7秒前
mbf发布了新的文献求助10
8秒前
mbf发布了新的文献求助10
8秒前
mbf发布了新的文献求助10
8秒前
mbf发布了新的文献求助10
8秒前
落寞怜雪完成签到,获得积分10
8秒前
8秒前
mbf发布了新的文献求助10
8秒前
mbf发布了新的文献求助10
9秒前
mbf发布了新的文献求助10
9秒前
半青完成签到 ,获得积分10
9秒前
mbf发布了新的文献求助10
9秒前
9秒前
mbf发布了新的文献求助10
9秒前
9秒前
9秒前
清秀幼丝发布了新的文献求助10
9秒前
mbf发布了新的文献求助10
10秒前
高分求助中
Solution Manual for Strategic Compensation A Human Resource Management Approach 1200
Natural History of Mantodea 螳螂的自然史 1000
Glucuronolactone Market Outlook Report: Industry Size, Competition, Trends and Growth Opportunities by Region, YoY Forecasts from 2024 to 2031 800
A Photographic Guide to Mantis of China 常见螳螂野外识别手册 800
Zeitschrift für Orient-Archäologie 500
Autoregulatory progressive resistance exercise: linear versus a velocity-based flexible model 500
The Politics of Electricity Regulation 300
热门求助领域 (近24小时)
化学 医学 生物 材料科学 工程类 有机化学 生物化学 物理 内科学 纳米技术 计算机科学 化学工程 复合材料 基因 遗传学 物理化学 催化作用 细胞生物学 免疫学 冶金
热门帖子
关注 科研通微信公众号,转发送积分 3340086
求助须知:如何正确求助?哪些是违规求助? 2968135
关于积分的说明 8632438
捐赠科研通 2647668
什么是DOI,文献DOI怎么找? 1449744
科研通“疑难数据库(出版商)”最低求助积分说明 671534
邀请新用户注册赠送积分活动 660503