Machine Learning-Based Image Forgery Detection Using Light Gradient-Boosting Machine

In recent days, due to the increasing development of digital automation, images have emerged as a significant way to interact as well as transfer messages in our community, and there was a high rise in the number of details transferred in the formation of virtual pictures in the day-to-day life especially with the disclosure of online medias such as Instagram, Twitter, and Facebook. Moreover, uploading pictures on social media and modifying those images with related software apps is considered a common method to do in current days. Even though, if every person does not perform with ominous meanings, but still, there is a noticeable rise in misconduct regarding the malignant image influence as well as updating. This research proposes an image/video forgery identification method by utilizing the light gradient boosting machine (Light-GBM) method to detect the fabrication in the visual data with an increased rate of accuracy. The performance, as well as comparative analysis, is estimated based on the performance metrics such as accuracy at 94.91%, sensitivity at 94.77%, and specificity at 93.26%, respectively, which is superior to the previous techniques.