Video super-resolution using an artificial neural network

1. A video processing system comprising:

a computing platform including a hardware processor and a system memory;

a software code including an artificial neural network (ANN) stored in the system memory;

the hardware processor configured to execute the software code to:

receive a first video sequence having a first display resolution;

produce a second video sequence based on the first video sequence using the ANN, the second video sequence having a second display resolution higher than the first display resolution;

wherein the ANN is configured to provide sequential frames of the second video sequence that are temporally stable and consistent in color to reduce visual flicker and color shifting in the second video sequence.

2. The video processing system of claim 1 , wherein the hardware processor is further configured to execute the software code to train the ANN through use of a plurality of individual still images, but not a video sequence.

3. The video processing system of claim 1 , wherein the hardware processor is further configured to execute the software code to display the second video sequence on a display.

4. The video processing system of claim 1 , wherein the ANN comprises a generative adversarial network (GAN) including a generator module and a discriminator module.

5. The video processing system of claim 4 , wherein the GAN comprises an energy-based GAN (EBGAN).

6. The video processing system of claim 4 , wherein a generator loss corresponding to the generator module includes a content loss factor penalizing differences between down-sampled frames of the first video sequence having a third display resolution lower than the first display resolution and down-sampled corresponding frames of the second video sequence having the third display resolution.

7. The video processing system of claim 1 , wherein the second display resolution is up to approximately four times higher than the first display resolution.

8. The video processing system of claim 1 , wherein the second video sequence comprises one of 4K video and 8K video.

9. A method for use by a video processing system including a computing platform having a hardware processor and a system memory storing a software code including an artificial neural network (ANN), the method comprising:

receiving, using the hardware processor, a first video sequence having a first display resolution;

producing, using the hardware processor and the ANN, a second video to sequence based on the first video sequence, the second video sequence having a second display resolution higher than the first display resolution;

wherein the ANN is configured to provide sequential frames of the second video sequence that are temporally stable and consistent in color to reduce visual flicker and color shifting in the second video sequence.

10. The method of claim 9 , further comprising training the ANN, using the hardware processor, through use of a plurality of individual still images, but not a video sequence.

11. The method of claim 9 , further comprising displaying, using the hardware processor, the second video sequence on a display.

12. The method of claim 9 , wherein the ANN comprises a generative adversarial network (GAN) including a generator module and a discriminator module.

13. The method of claim 12 , wherein the GAN comprises an energy-based GAN (EBGAN).

14. The method of claim 12 , wherein a generator loss corresponding to the generator module includes a content loss factor penalizing differences between down-sampled frames of the first video sequence having a third display resolution lower than the first display resolution and down-sampled corresponding frames of the second video sequence having the third display resolution.

15. The method of claim 9 , wherein the second display resolution is up to approximately four times higher than the first display resolution.

16. The method of claim 9 , wherein the second video sequence comprises one of 4K video and 8K video.

17. A computer-readable non-transitory medium having stored thereon instructions and an artificial neural network (ANN), which when executed by a hardware processor, instantiate a method comprising:

receiving a first video sequence having a first display resolution;

producing a second video sequence based on the first video sequence using the ANN, the second video sequence having a second display resolution higher than the first display resolution;

wherein the ANN is configured to provide sequential frames of the second video sequence that are temporally stable and consistent in color to reduce visual flicker and color shifting in the second video sequence.

18. The computer-readable non-transitory medium of claim 17 , wherein the ANN comprises a generative adversarial network (GAN) including a generator module and a discriminator module.

19. The computer-readable non-transitory medium of claim 18 , wherein the GAN comprises an energy-based GAN (EBGAN).

20. The computer-readable non-transitory medium of claim 18 , wherein a generator loss corresponding to the generator module includes a content loss factor penalizing differences between down-sampled frames of the first video sequence having a third display resolution lower than the first display resolution and down-sampled corresponding frames of the second video sequence having the third display resolution.