Steganalysis using Machine Learning

This Repository supports the experiments and machine learning implementation for "Steganalysis using Machine Learning" research.

What is Steganalysis?

Steganography is a combination of two words “steganos” and “graphia” meaning “Covered Writing”. It is amethod of secret communication where no one other than theparties directly involved (sender and receiver) are aware of itsexistence. Steganalysis is the process of detection of messageshidden by the means of Stenography. The hidden message canbe in the form of an audio file, an image, a video file or plaintext (may or may not be encrypted).

Environment and Setup

We have use Google Cloud Platform's Compute Engine Instances and Jupyter Notebook.

Contents

1). Dataset

• Explores the dataset

2). SRNet

• Trained on 400 - 256x256 unbalanced cover + stego images
  
• 10 hours

3). SRNet_#2

• Trained on 400 - 256x256 balanced cover + stego images
  
• 18 hours