Resizing_Forensics

• The experiments have been carried out using Tensorflow.keras with tensoflow version 2.4.1
• The training and testing have been done in google colaboratory
• Training of the model has been done on GPU

File Descriptions

• The folder proposed (preprocessing layer) contains all files and findings for our proposed method
• Proposed_Method.ipynb: Contains the creation and training of the deep learning model.
• Results.ipynb: Contains the results i.e accuracy vs resizing factor and accuracy vs Quality Factor 2.
• DataLoader.py : Contains the generator required for loading data during model training
• utilities.py: Contains the data preparation algorithm
• accuracyMatrix_1.py: Contains the script for generating the accuracy vs quality factor pairs matrix
• const.py: Contains script to generate training curves.
• model_1_3.hdf5: Saved weights for the trained model.
• logs_1_3.txt: Contains the training logs for the model.
• requirements.txt: version of certain libraries required. Do !pip install requirements.txt

Our Research Group

Supervisor : Prof. Manish Okade, Dept. of Electronics and Communication Engineering NIT Rourkela.

• Bibhash Pran Das, Dept of ECE, NIT Rourkela
• Mrutyunjay Biswal, Dept of ECE, NIT Rourkela
• Billa Nikhil Reddy, Dept of Electrical Engineering, NIT Rourkela
• Abharnta Panigrahi, Dept of Biomedical Engineering, NIT Rourkela