Detection of Deep Network Generated Images Using Disparities in Color Components

Code for the journal("Detection of Deep Network Generated Images Using Disparities in Color Components")

---

Main files:

1. 'featureGeneration.py': The 588-D feature feature generation code for the real and DNG image datasets as inputs.
2. 'features.py': callable function container.
3. 'svmTrainTest.py': Model for one class classification of the feature set generated.
4. 'modelRc64.sav': Pre-trained model on the 64x64 image feature set.
5. 'modelRc128.sav': Pre-trained model on the 128x128 image feature set.

---

Link to the paper:

[1] Haodong Li, Bin Li, Shunquan Tan, and Jiwu Huang, “Detection of deep network generated images using disparities in color components”, arXiv preprint arXiv:1808.07276, 2018. (Link to the paper)