Detecting GAN generated Fake Images using Co-occurrence Matrices

By Lakshmanan Nataraj, Tajuddin Manhar Mohammed, Bangalore S. Manjunath, S. Chandrasekaran, Arjuna Flenner, Jawadul H. Bappy, Amit K. Roy-Chowdhury. 2019

368 citations recorded by the source corpus.

The advent of Generative Adversarial Networks (GANs) has brought about completely novel ways of transforming and manipulating pixels in digital images. GAN based techniques such as Image-to-Image translations, DeepFakes, and other automated methods have become increasingly popular in creating fake images. In this paper, we propose a novel approach to detect GAN generated fake images using a combination of co-occurrence matrices and deep learning. We extract co-occurrence matrices on three color channels in the pixel domain and train a model using a deep convolutional neural network (CNN) framework. Experimental results on two diverse and challenging GAN datasets comprising more than 56,000 images based on unpaired image-to-image translations (cycleGAN [1]) and facial attributes/expressions (StarGAN [2]) show that our approach is promising and achieves more than 99% classification accuracy in both datasets. Further, our approach also generalizes well and achieves good results when trained on one dataset and tested on the other.

Subfields: Computer Vision and Pattern Recognition, Generative Models, Safety, Robustness & Interpretability

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Computer Vision and Pattern Recognition

In Generative Models

In Safety, Robustness & Interpretability

Citations in the corpus

CNN-Generated Images Are Surprisingly Easy to Spot… for Now (2019, methodology): “with comparisons to [44, 26, 12], respectively.”
Media Forensics and DeepFakes: An Overview (2020, background): “Several approaches also adopt deep learning solutions in a supervised setting [177], [178].”
Towards Universal Fake Image Detectors that Generalize Across Generative Models (2023, methodology): “We follow existing works [10, 25, 34, 49, 52] and report both average precision (AP) and classification accuracy.”
Global Texture Enhancement for Fake Face Detection in the Wild (2020, methodology): “However, our method outperforms the ResNet baseline by around 10% and [27] by around 15% regarding accuracy in all settings.”
Incremental learning for the detection and classification of GAN-generated images (2019, methodology): “Finally, we also considered the network proposed in [15] that uses the co-occurrence matrix as input.”
DeepFakes and Beyond: A Survey of Face Manipulation and Fake Detection (2020, background): “Nataraj et al. proposed in [31] a detection system inspired by steganalysis and natural image statistics.”
FakeSpotter: A Simple Baseline for Spotting AI-Synthesized Fake Faces (2019, methodology): “Recently, some research [6,7,9,18] borrow ideas from the traditional digital image forensics to detect GAN-synthesized fake faces/images.”
Source Generator Attribution via Inversion (2019, methodology): “A related problem is the detection of whether a given image was generated by any GAN-type generator, which has successfully been demonstrated by several groups using various meth-ods [18, 22, 7, 6, 14, 20].”
Detecting and Simulating Artifacts in GAN Fake Images (2019, methodology): “Since both the JPEG compression and image resize destroy the up-sampling artifact, the model trained with images without post-processing does not generalize to the post-processed images, also reported in [15].”
Deepfakes generation and detection: state-of-the-art, open challenges, countermeasures, and way forward (2021, methodology): “Deep Learning Nataraj et al. [162] CNN Deep features + co-occurrence matrices Accuracy=99.49% ▪ cycleGAN ▪ Works with static images only.”
GANprintR: Improved Fakes and Evaluation of the State of the Art in Face Manipulation Detection (2019, methodology): “2) Steganalysis [38]: the method by Nataraj et al.”
Leveraging Frequency Analysis for Deep Fake Image Recognition (2020, methodology): “…et. al. provide a comparison of different steganalysis and CNN-based methods (Marra et al., 2018), several approaches use CNNs in the image domain (Mo et al., 2018; Yu et al., 2019a; Tariq et al., 2019), others use statistics in the image domain (McCloskey & Albright, 2018; Nataraj et al., 2019).”
Detecting CNN-Generated Facial Images in Real-World Scenarios (2020, methodology): “For pre-processing techniques we use high-pass filters, co-occurrence matrices, and color transformations, since these have recently been demonstrated to work well in CNN-generated image detection [13, 33, 24].”
Detection of AI-Generated Synthetic Faces (2022, methodology): “In this setting, even a simple approach like a shallow CNN can reach almost perfect performance [2, 4, 18, 35, 42].”
DIRE for Diffusion-Generated Image Detection (2023, background): “Earlier researchers focus on detecting generated images leveraging hand-crafted features, such as color cues [28], saturation cues [29], blending artifacts [22], co-occurrence features [30].”