mnansary / pyCMFD

Copy Move Image Forgery Detection with Segmentation Models

pyCMFD

Copy Move Image Forgery Detection with Segmentation Models

Version: 0.0.3  
Python : 3.6.9
Author : Md. Nazmuddoha Ansary 
         Shakir Hossain  
         Mohammad Bin Monjil  
         Habibur Rahman
         Shahriar Prince  

Version and Requirements

• Python == 3.6.9
• pip3 install -r requirements.txt

Data Set

• Evaluation:
  • Casia
  • CoMoFoD
• Training:
  • MICC
  • COVERAGE
  • Grip
  • CVIP
  • FFCMF
• Synthetic data Generation
  • COCO

Processing

• The folder structure after Dividing CVIP and MICC is as follows

  ​       ├── CASIA
  ​       │   ├── Au
  ​       │   ├── GT_Mask
  ​       │   └── Tp
  ​       ├── coco
  ​       │   ├── annotations
  ​       │   └── train2014
  ​       ├── CoMoFoD
  ​       ├── COVERAGE
  ​       │   ├── image
  ​       │   ├── label
  ​       │   └── mask
  ​       ├── CVIP0
  ​       ├── CVIP12
  ​       │   ├── im1
  ​       │   ├── im10
  ​       │   ├── im11
  ​       │   ├── im12
  ​       │   ├── im13
  ​       │   ├── im14
  ​       │   ├── im15
  ​       │   ├── im16
  ​       │   ├── im17
  ​       │   ├── im18
  ​       │   ├── im19
  ​       │   ├── im2
  ​       │   ├── im20
  ​       │   ├── im3
  ​       │   ├── im4
  ​       │   ├── im5
  ​       │   ├── im6
  ​       │   ├── im7
  ​       │   ├── im8
  ​       │   └── im9
  ​       ├── F2000
  ​       ├── F220
  ​       ├── F600
  ​       ├── FFCMF
  ​       └── GRIP
  

• dataset.py holds all the code for processing the datasets

• data.ipynb creates the Train and Eval Dataset (in TFRECORD and Image format locally)

ENVIRONMENT

    OS          : Ubuntu 18.04.3 LTS (64-bit) Bionic Beaver     
    Memory      : 7.7 GiB  
    Processor   : Intel® Core™ i5-8250U CPU @ 1.60GHz × 8    
    Graphics    : Intel® UHD Graphics 620 (Kabylake GT2)  
    Gnome       : 3.28.2