Multimedia Forensic: An Approach for Splicing Detection based on Deep Visual Features

Fabrication of an image by utilizing a wide range of image altering tasks can lead to fraud. Since a measurable inspector must consider each of these. numerous techniques have emerged for calculation to identifying picture tampering activities and algorithms. In this paper, we propose a convolution neural network based network to deal with tampering and segmentation of visual data. Network is designed to preserve local and global characteristics of data to detect tampering. Inception module based CNN network is proposed intended to identify a picture’s content and tampered part using mask R-CNN. Through a progression of examinations, we showed that our proposed methodology can consequently figure out how to distinguish numerous picture tampering methods without depending on pre-chosen highlights or any pre-preparing and segment out tampered region. Evaluation results demonstrate that our proposed methodology can distinguish tampering over CASIA v1.0 and v2.0 with an accuracy of 98.76% and 97.92%, respectively.