Using U-NET with Grasshopper Optimisation to Spot Image Forgery on Social Media

Kalpuerty Derwe^*
*Correspondence: Kalpuerty Derwe, Department of Computer Science, Laurentian University, Sudbury, ON P3E 2C6, Canada, Email:

Keywords

Digital age • Splicing • Digital manipulation

Introduction

Deep learning techniques have shown remarkable success in various image analysis tasks, including image classification, segmentation, and detection. U-Net, Convolutional Neural Network (CNN) architecture was specifically designed for image segmentation tasks. Its unique U-shaped structure, with a contracting path and an expansive path, enables it to capture both local and global features of an image. U-Net's encoder captures image features at different scales, while the decoder reconstructs the segmented output. This architecture is well-suited for image forgery detection as it can learn intricate features that might indicate manipulation, such as inconsistencies in lighting, shadows, and textures [1,2]. Grasshopper Optimization Algorithm (GOA) is a nature-inspired optimization technique based on the swarming behaviour of grasshoppers. GOA is particularly effective in solving complex optimization problems due to its ability to escape local optima and explore the solution space extensively. In the context of image forgery detection, GOA can be used to fine-tune the parameters of the U-Net architecture for optimal performance. The algorithm iteratively adjusts the parameters to minimize the detection error, thereby enhancing the accuracy of the forgery detection model [3].

Literature Review

Image forgery encompasses a range of techniques, including copy-move forgery, splicing, retouching, and more. Copy-move forgery involves duplicating and pasting a portion of an image within the same image or across multiple images. Splicing combines different parts of different images to create a new, deceptive image. Retouching alters the appearance of an image by modifying certain aspects, making it challenging to distinguish from the original. These techniques demand innovative solutions for accurate detection. Deep learning has emerged as a powerful tool for various computer vision tasks, including image forgery detection. U-Net, convolutional neural network architecture is widely used for tasks like image segmentation due to its ability to capture fine-grained features. U-Net's architecture consists of a contracting path that captures context and a symmetric expanding path that enables precise localization. By training the network on a dataset of both authentic and forged images, it learns to differentiate between the two categories based on learned features [4,5].

Discussion

The rapid proliferation of social media has led to an increase in the manipulation and dissemination of digital images. Image forgery, including techniques like splicing, cloning, and retouching, has become a significant concern, impacting the credibility of online content. To combat this issue, researchers and practitioners have turned to advanced technologies such as deep learning and optimization algorithms. This article explores the integration of U-Net, a deep convolutional neural network, and Grasshopper Optimization, a metaheuristic algorithm, to enhance the detection of image forgery on social media platforms [6]. The age of social media has transformed the way information is shared, but it has also brought about challenges related to the authenticity of content. Image forgery, which involves manipulating or altering images to deceive viewers, poses a threat to the credibility of visual information online. Traditional methods of detecting image forgery often fall short due to the increasing sophistication of manipulation techniques. This article delves into a novel approach that combines the power of deep learning and metaheuristic optimization to enhance the accuracy of spotting forged images.

Conclusion

As image forgery continues to pose a threat to the authenticity of content on social media, innovative approaches are required to enhance detection accuracy. The integration of U-Net, a powerful deep learning architecture, with Grasshopper Optimization, a versatile metaheuristic algorithm, offers a promising solution. By harnessing the capabilities of both techniques, we can develop more robust and accurate forgery detection models. As technology advances, this approach can play a pivotal role in restoring trust and credibility to the digital landscape.

Acknowledgement

None.

Conflict of Interest

None.

References

1. Nabi, Syed Tufael, Munish Kumar, Paramjeet Singh and Naveen Aggarwal, et al. "A comprehensive survey of image and video forgery techniques: Variants, challenges and future directions." Multimed Syst 28 (2022): 939-992.

Google Scholar, Crossref, Indexed at

2. Qazi, Emad Ul Haq, Tanveer Zia and Abdulrazaq Almorjan. "Deep learning-based digital image forgery detection system." Appl Sci 12 (2022): 28-51.

Google Scholar, Crossref, Indexed at

3. Abd Warif, Nor Bakiah, Ainuddin Wahid Abdul Wahab, Mohd Yamani Idna Idris and Roziana Ramli, et al. "Copy-move forgery detection: Survey, challenges and future directions." J Netw Comput Appl 75 (2016): 259-278.

Google Scholar, Crossref, Indexed at

4. Zhao, Fei, Wenchang Shi, Bo Qin and Bin Liang. "Image forgery detection using segmentation and swarm intelligent algorithm." Wuhan Univ J Nat Sci 22 (2017): 141-148.

Google Scholar, Crossref, Indexed at

5. AlShariah, Njood Mohammed, A. Khader and J. Saudagar. "Detecting fake images on social media using machine learning." Int J Adv Comput Sci Appl 10 (2019): 170-176.

Google Scholar, Crossref, Indexed at

6. Liu, Yaqi, Qingxiao Guan and Xianfeng Zhao. "Copy-move forgery detection based on convolutional kernel network." Multimed Tools Appl 77 (2018): 18269-18293.

Google Scholar, Crossref, Indexed at