Abstract
The recent digital revolution has sparked a growing interest in applying convolutional neural networks (CNNs) and deep learning to the field of image forensics. The proposed methods aim to train algorithms for solving a range of predetermined tasks. However, training a model that has been randomly initialized requires extensive time for computation as well as an enormous pool of training data to draw from. Moreover, such a model needs to be developed and redeveloped from the ground up if there are any alterations to the feature-space distribution. In addressing these problems, the present paper proposes a novel approach to training image forgery detection models. The method applies prior knowledge that has been transferred to the new model from previous steganalysis models. Additionally, because CNN models generally perform badly when transferred to other databases, transfer learning accomplished through knowledge transfer allows the model to be easily trained for other databases. The various models are then evaluated using image forgery techniques such as shearing, rotating, and scaling images. The experimental results, which show an image manipulation detection has validation accuracy of over 94.89%, indicate that the proposed transfer learning approach successfully accelerates CNN model convergence but does not improve image quality.
Highlights
Human beings are generally hard-wired to apply or transfer knowledge that they have learned in one skill to other related skills
Because convolutional neural networks (CNNs) middle layers and baseline architecture are alike, we aim to demonstrate the practicality and effectiveness of re-training the baseline by applying the original network weights instead of new CNN weights, with the aim of enhancing forgery detection abilities with more accurate image classification
The other advantage of using data generator is make many changes in each image which known as data augmentation to learn all possible image transformation to the used neural network after fixing the image scale according to the input layer which in our case is 126 × 126 × 32
Summary
Human beings are generally hard-wired to apply or transfer knowledge that they have learned in one skill to other related skills In this way, acquired knowledge can help solve problems in less time through the cross-utilization of knowledge which occurs during these transfers [1]. Considering this built-in problem-solving framework, let us look at two critical problems currently restricting progress in the field of image forensics. These problems are: (1) developing a sufficiently large and diverse body of annotated images for use as training models; and (2) including in the models non-image data in order to permit and enable broader application of the model(s). Finding a way to include non-image data in transfer learning is crucial if the models are to be applied to areas such as the medical field or insurance industry
Sign-in/Register to view highlights & summary Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callMore From: European Journal of Electrical Engineering and Computer Science
Disclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
