Treffer: Detection of image forgery using deep learning.

The advent of forged image detection in the digital realm has spurred research into advanced techniques. The main focus of this study is a thorough examination of the effectiveness of deep learning models, including VGG19, EfficientNet- B2, and ELA CNN, in identifying manipulated images. VGG Net, is a fitting choice for forged image detection due to its proven effectiveness in image classification tasks. Its architecture, with multiple layers of small convolutional filters, enables capturing intricate features crucial for identifying tampered images. In subsequent research, I have explored two additional deep learning models—EfficientNet-B2 and ELA CNN with their superior scalability and parameter efficiency make it a promising models for handling complex image manipulation scenarios. The CASIA 2.0 Image Tampering Detection Dataset, which consists of two classes—original and tampered images—is the dataset used for testing taking 3000 images from each class. The suggested method comprises a methodical methodology that includes gathering data from Kaggle. Although we first evaluate three deep learning models, we concentrate our analysis on VGG19. A thorough model evaluation that included a Classification Report and Confusion Matrix revealed that the model had a 72 per cent classification accuracy. Furthermore, the study describes how to create a graphical user interface (GUI) with Flask to enable user interaction and practical implementation. This work highlights the unique advantages and shortcomings of using deep learning models for forged image identification in the context of VGG19. It also provides subtle insights into this application. [ABSTRACT FROM AUTHOR]

Copyright of AIP Conference Proceedings is the property of American Institute of Physics and its content may not be copied or emailed to multiple sites without the copyright holder's express written permission. Additionally, content may not be used with any artificial intelligence tools or machine learning technologies. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)