Hybrid machine learning techniques for image forensics.

In these days image authentication has become an essential tool as image enhancing software are becoming prevalent and established in today's world. As such, images can unfortunately be tampered in a multitude of ways by even nonexpert people using the assistance of digital forgery tools. Hence, a universal method to identify such an issueis critical because information is key in the modern world. In this paper, we propose a new framework wherein tampering of images using Error Level Analysis and transfer learning is proposed. The time taken for processing and complexity of using such a method is lower in comparison with existing solutions for image forgery since the images are not split into patches. [ABSTRACT FROM AUTHOR]