.png){kind=icon}
返回
Deep Learning for Safeguarding Image Trustworthiness
.png){kind=icon}
PRE
AI
DOI：10.1080/08874417.2025.2547181.png)
.png){kind=icon}
原文求助.png){kind=icon}
分享.png){kind=icon}
收藏
摘要
En 中文
Nowadays, digital images are confronting a crisis of trustworthiness with the tremendous breakthroughs in digital imaging and steganography tools. However, many researchers have infiltrated deep learning (DL) frameworks in various research domains and applications. In this context, safeguarding trustworthiness in image data using deep learning frameworks has been proposed in this article. Deep learning frameworks such as CNN, VGG-19 and ResNet50 are used for fake image detection. Accuracy, precision, recall, F1-score and area under receiver operating characteristic (AUC) curve are used as performance metrics for comprehensive comparative analysis of the frameworks. From the comparison results, it is clear that ResNet50 outperforms when compared to other architectures under consideration with a total accuracy of 99% showing an improvement ranging from 5.3% to 76.78%. Further, ResNet50 exhibits superior performance when compared to its variants which are proposed by many authors.
Keyword:
Trustworthiness
deep learning
transfer learning
convolutional neural networks (CNN)
visual geometry group (VGG-19)
residual network (ResNet50)
期刊
IF:
4.2
论文数:
209
被引数:
3.1K
机构
引用论文
Lightweight and Resource-Constrained Learning Network for Face Recognition with Performance Optimization面向性能优化的轻量级资源受限人脸识别学习网络.png)
SENSORS
OAAI
SENSORSIF3.5
OAAI分享收藏
查看更多内容