pushp360 / holmes2

home for digital image analysis tools/ personal streamlit experiment

Holmes2

The open source home for digital image foresnics and analysis tools

Table of contents

• Holmes2
  • Table of contents
  • General info
  • Available features
  • Technologies
  • Setup
  • File structure
  • TODO
  • issues

General info

Holmes2 is designed to be a home for digital image forensics tools that is simple to extend with the aim of creating a platform for researchers and forensic experts to experiment and showcase various digital image forensics algorithms.

This project is my undergrad final project and i initially intended it to be an automated tool for detecting forgery then i realised i didnt have the time for that task, so i decided to pivot to make holmes more of a toolkit. Although the project has not relationship with a similar but more complete project called sherloq, I think thanks to streamlit it should be a lot easier to contribute new algorithms.

Note: Holmes is not meant to be an automated tool for deciding if an image if foreged or not. Holmes is still lacking UI polish, and several issues.

Available features

• Exif viewer
• Quantization tables matching(not yet complete)
• Error level analysis
• Noiseprint by grip-unina

Technologies

Project is created with:

• python: 3.8
• streamlit

Setup

To run this project

install necessary pip libraries

 pip install -r requirements.txt

To run holmes in browser

streamlit run holmes.py

File structure

• Holmes.py: is the main script, which contains streamlit page setup and page navigation

• Every other module should be named after what they accomplish.

TODO

• Using quantization tables from jpeg images we can classify the origins of an image.
• Imporve file structure
• UI changes

issues

• Currently runing noiseprint in an unpythonic fashion.