Effective Analysis of Real World Stego Images through Deep Learning Techniques

With the widespread application of deep learning in Steganography and Steganalysis, increasing data paves its way. The human eye is incapable of differentiating cover portray from a stego image. Regardless of whether or not people are aware of the embedding method, steganalysis is used to find any hidden messages in digital information. Moreover, deep neural networks play a major role in Image Processing. It is a difficult effort to identify steganography images in real-world scenarios that differ in size, adaptability, lighting circumstances, and unrecognized steganography techniques, as well as the payload or embedding capacity. When it comes to recognizing stego images from the real world, this research investigates the operation of two deep learning techniques that make use of Convolutional Neural Networks (CNN). The frameworks considered are Yedroudj-net and GBRAS-net. These models are trained with datasets from the real world and are assessed based on the accuracy of the framework.