Splicing Image Forgery Detection by Deploying Deep Learning Model

• Published in: 2022 International Conference on Automation, Computing and Renewable Systems (ICACRS) pp. 1116 - 1120
• Main Authors: Krishnamoorthy, N, Amuthadevi, C., Geedtha, M. K., Reddy, Poli Lokeshwara, S, Anitha Rani K, Gopinathan, R.
• Format: Conference Proceeding
• Language: English
• Published: IEEE 13.12.2022
• Subjects: Accuracy; Computational modeling; Data models; Deep learning; Forgery; Image forgery; MobileNet; Renewable energy sources; Splicing; Training
• DOI: 10.1109/ICACRS55517.2022.10029055

The use of digital images has become more important in numerous industries, including journalism, medical diagnosis, law enforcement, and forensics. Images could be easily modified due to the availability of photo-altering tools and software, which can distort the original information. The availability of low-cost and simple image-editing software has also drastically decreased the time and money needed to engage in image tampering. This destroys the purity of the image, making it vulnerable to misuse by anyone. Copy-move, splicing, and enhancement falsification are prominent methods used to create fake images. Using a combination of two or more images, a new image is created called splicing, which may be shared across many online channels and used to spread information and impact target audiences. The discovery could have beneficial as well as negative results. Consequently, it is important to provide a method that can identify the splicing forgery in an image. The study used a Deep Learning (DL) model named MobileNet to distinguish between authentic and manipulated images. Both the training and testing of the DL model make use of the CASIA data. Resizing and enhancing methods are used to process the collected data. Accuracy, F1 score, and recall are employed to test the reliability of the created DL model.