Generative image congealing
Embodiments are disclosed for generative image congealing which provides an unsupervised learning technique that learns transformations of real data to improve the image quality of GANs trained using that image data. In particular, in one or more embodiments, the disclosed systems and methods comprise generating, by a spatial transformer network, an aligned real image for a real image from an unaligned real dataset, providing, by the spatial transformer network, the aligned real image to an adversarial discrimination network to determine if the aligned real image resembles aligned synthetic images generated by a generator network, and training, by a training manager, the spatial transformer network to learn updated transformations based on the determination of the adversarial discrimination network.
1. A computer-implemented method comprising:
generating, by a spatial transformer network, an aligned real dataset from an unaligned real dataset, wherein the spatial transformer network is trained to learn transformations to generate aligned images based on a comparison of generated aligned images and synthetic aligned images using an adversarial discrimination network;
providing, by the spatial transformer network, at least one real image from the aligned real dataset to the adversarial discrimination network to determine if the at least one real image resembles synthetic images generated by a generator network; and
training, by a training manager, the generator network to learn to generate the synthetic images based on the determination of the adversarial discrimination network.
2. The computer-implemented method of claim 1 , wherein the spatial transformer network is trained by:
generating, by the spatial transformer network, an aligned real image for a real image from the unaligned real dataset;
providing, by the spatial transformer network, the aligned real image to the adversarial discrimination network to determine if the aligned real image resembles aligned synthetic images generated by the generator network; and
training, by the training manager, the spatial transformer network to learn updated transformations based on the determination of the adversarial discrimination network.
3. The computer-implemented method of claim 2 , wherein the aligned synthetic images are generated by the generator network upon receiving input vectors that have been biased to increase fidelity of the aligned synthetic images.
4. The computer-implemented method of claim 2 , further comprising:
applying, by the spatial transformer network, the updated transformations learned during training to the unaligned real dataset to generate the aligned real dataset.
5. The computer-implemented method of claim 2 , wherein the updated transformations include one or more geometric transformations or crops.
6. The computer-implemented method of claim 1 , further comprising:
generating, by the spatial transformer network, a second aligned real image for a second real image from the unaligned real dataset;
providing, by the spatial transformer network, the second aligned real image to the adversarial discrimination network to determine if the second aligned real image resembles new synthetic images generated by the generator network, wherein the new synthetic images are generated by the generator network using different input parameters than used to generate the aligned synthetic images; and
training, by the training manager, the spatial transformer network to learn further updated transformations based on the determination of the adversarial discrimination network.
7. The computer-implemented method of claim 1 , further comprising:
deploying, by an artificial intelligence management system, the generator network to a digital design system.
8. A system, comprising:
a computing device implementing a generative image congealing system, the generative image congealing system comprising:
a spatial transformer network to:
generate an aligned real dataset from an unaligned real dataset, wherein the spatial transformer network is trained to learn transformations to generate aligned images based on a comparison of generated aligned images and synthetic aligned images using an adversarial discrimination network; and
provide at least one real image from the aligned real dataset to the adversarial discrimination network to determine if the at least one real image resembles synthetic images generated by a generator network; and
a training manager to train the generator network to learn to generate the synthetic images based on the determination of the adversarial discrimination network.
9. The system of claim 8 , wherein:
the spatial transformer network is further to:
generate an aligned real image for a real image from the unaligned real dataset; and
provide the aligned real image to the adversarial discrimination network to determine if the aligned real image resembles aligned synthetic images generated by the generator network; and
the training manager is further to train the spatial transformer network to learn updated transformations based on the determination of the adversarial discrimination network.
10. The system of claim 9 , wherein the aligned synthetic images are generated by the generator network upon receiving input vectors that have been biased to increase fidelity of the aligned synthetic images.
11. The system of claim 9 , wherein the spatial transformer network is further to:
apply the updated transformations learned during training to the unaligned real dataset to generate an aligned real dataset.
12. The system of claim 9 , wherein the updated transformations include one or more geometric transformations or crops.
13. The system of claim 8 , further comprising:
wherein the spatial transformer network is further to:
generate a second aligned real image for a second real image from the unaligned real dataset; and
provide the second aligned real image to the adversarial discrimination network to determine if the second aligned real image resembles new synthetic images generated by the generator network, wherein the new synthetic images are generated by the generator network using different input parameters than used to generate the aligned synthetic images; and
wherein the training manager is further to train the spatial transformer network to learn further updated transformations based on the determination of the adversarial discrimination network.
14. The system of claim 8 , further comprising:
an artificial intelligence management system to deploy the generator network to a digital design system.
15. A system comprising:
means for generating an aligned real dataset from an unaligned real dataset by a spatial transformer network, wherein the spatial transformer network is trained to learn transformations to generate aligned images based on a comparison of generated aligned images and synthetic aligned images using an adversarial discrimination network;
means for providing at least one real image from the aligned real dataset to the adversarial discrimination network to determine if the at least one real image resembles synthetic images generated by a generator network; and
means for training the generator network to learn to generate the synthetic images based on the determination of the adversarial discrimination network.
16. The system of claim 15 , further comprising:
means for generating an aligned real image for a real image from the unaligned real dataset;
means for providing the aligned real image to the adversarial discrimination network to determine if the aligned real image resembles aligned synthetic images generated by the generator network; and
means for training the spatial transformer network to learn updated transformations based on the determination of the adversarial discrimination network.
17. The system of claim 16 , wherein the aligned synthetic images are generated by the generator network upon receiving input vectors that have been biased to increase fidelity of the aligned synthetic images.
18. The system of claim 16 , further comprising:
means for applying the updated transformations learned during training to the unaligned real dataset to generate an aligned real dataset.
19. The system of claim 15 , further comprising:
means for generating a second aligned real image for a second real image from the unaligned real dataset;
means for providing the second aligned real image to the adversarial discrimination network to determine if the second aligned real image resembles new synthetic images generated by the generator network, wherein the new synthetic images are generated by the generator network using different input parameters than used to generate the aligned synthetic images; and
means for training the spatial transformer network to learn further updated transformations based on the determination of the adversarial discrimination network.
20. The system of claim 15 , further comprising:
means for deploying the generator network to a digital design system.