An Unsupervised Learning Algorithm based deletion of Inter-frame Forgery Detection System

People utilize contemporary tools to edit captured video clips effortlessly. These video clips hold significance as evidence within a legal setting. Consequently, the identification of video tampering becomes crucial during legal proceedings. In order to improve the legal process, this paper presents an inter-frame forgery detection system especially designed for frame deletions. The system initially receives video clips and employs a statistical approach to determine whether they have been tampered with or are authentic. If the video system is determined to be forged, it will proceed with detecting frame deletions or other alterations. This detection process utilizes an unsupervised learning algorithm, specifically the K-mean learning algorithm. By accepting the cross-correlation coefficient value, the detection process categorizes the different types of forgery. The conducted tests on video clips sourced from two data sets: the University of Central Florida (UCF) 101 data sets and Surrey University Library for Forensic Analysis (SULFA) data sets to evaluate the system. The outcomes of the conducted investigation clearly showcase the effectiveness of the suggested system in precisely identifying instances of frame deletion forgery, achieving an impressive success rate of 87%.