Detecting digital image manipulations
Systems, devices, methods and instructions are described for detecting GAN generated images. On embodiment involves receiving an images, generating co-occurrence matrices on color channels of the image, generating analysis of the image by using a convolutional neural network trained to analyze image features of the images based on the generated co-occurrence matrices and determining whether the image is a GAN generated image based on the analysis.
1. A method comprising:
receiving, using one or more processors, an image that comprises a red channel, a green channel, and a blue channel;
generating, using the one or more processors, a plurality of co-occurrence matrices, the plurality of co-occurrence matrices comprising a first co-occurrence matrix based on co-occurring image pixel values of the red channel of the image, a second co-occurrence matrix based on co-occurring image pixels values of the green channel of the image, and a third co-occurrence matrix based on co-occurring image pixel values of the blue channel of the image;
generating, using the one or more processors, analysis of the image based on the generated plurality of co-occurrence matrices by using a convolutional neural network trained to detect generative adversarial network (GAN) generated images; and
determining using the one or more processors, whether the image is a generative adversarial network (GAN) generated image based on the analysis.
2. The method of claim 1 , further comprising storing, using the one or more processors, the determination in association with the image.
3. The method of claim 1 , wherein the convolutional neural network is a multi-layer deep convolutional neural network.
4. The method of claim 1 , wherein the convolutional neural network is trained on a first dataset comprising unpaired image-to-image translations of objects generated using a cycle-consistent GAN framework.
5. The method of claim 1 , wherein the convolutional neural network is trained on a second dataset comprising authentic images and GAN generated images.
6. The method of claim 1 , wherein the convolutional neural network is trained on compressed images.
7. A computer system comprising:
a processor; and
a memory storing instructions that, when executed by the processor, configure the computer system to perform operations comprising:
receiving an image that comprises a red channel, a green channel, and a blue channel;
generating a plurality of co-occurrence matrices, the plurality of co-occurrence matrices comprising a first co-occurrence matrix based on co-occurring image-pixel values of the red channel of the image, a second co-occurrence matrix based on co-occurring image pixels values of the green channel of the image, and a third co-occurrence matrix based on co-occurring image pixel values of the blue channel of the image;
generating analysis of the image based on the generated plurality of co-occurrence matrices by using a convolutional neural network trained to detect generative adversarial network (GAN) generated images; and
determining whether the image is a generative adversarial network (GAN) generated image based on the analysis.
8. The computer system of claim 7 , wherein the operations further comprise:
storing the determination in association with the image.
9. The computer system of claim 7 , wherein the convolutional neural network is a multi-layer deep convolutional neural network.
10. The computer system of claim 7 , wherein the convolutional neural network is trained on a first dataset comprising unpaired image-to-image translations of objects generated using a cycle-consistent GAN framework.
11. The computer system of claim 7 , wherein the convolutional neural network is trained on a second dataset comprising authentic images and GAN generated images.
12. The computer system of claim 7 , wherein the convolutional neural network is trained on compressed images.
13. A non-transitory computer-readable storage medium, the computer-readable storage medium including instructions that when executed by a computer, cause the computer to perform operations comprising:
receiving an image that comprises a red channel, a green channel, and a blue channel;
generating a plurality of co-occurrence matrices, the plurality of co-occurrence matrices comprising a first co-occurrence matrix based on co-occurring image pixel values of the red channel of the image, a second co-occurrence matrix based on co-occurring image pixels values of the green channel of the image, and a third co-occurrence matrix based on co-occurring image pixel values of the blue channel of the image;
generating analysis of the image based on the generated plurality of co-occurrence matrices by using a convolutional neural network trained to detect generative adversarial network (GAN) generated images; and
determining whether the image is a generative adversarial network (GAN) generated image based on the analysis.
14. The computer-readable storage medium of claim 13 , further comprising storing the determination in association with the image.
15. The computer-readable storage medium of claim 13 , wherein the convolutional neural network is a multi-laver deep convolutional neural network.
16. The computer-readable storage medium of claim 13 , wherein the convolutional neural network is trained on a first dataset comprising unpaired image-to-image translations of objects generated use a cycle-consistent GAN framework.
17. The computer-readable storage medium of claim 13 , wherein the convolutional neural network is trained on a second dataset comprising authentic images and GAN generated images.