Generating images using neural networks

Generating confidence values ​​through neural networks and mixing images based on these values ​​solves the problem of artifacts during image mixing and improves display quality.

CN120031992APending Publication Date: 2025-05-23NVIDIA CORP
View PDF 0 Cites 1 Cited by

Patent Information

Application Number
CN202411610046.5
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Priority Date
2023-11-21
Filing Date
2024-11-12
Publication Date
2025-05-23

AI Technical Summary

Technical Problem

During the image mixing process, especially in video conferencing, visual artifact problems are prone to occur, resulting in poor display results.

Method used

The occurrence of artifacts is reduced by generating confidence values ​​using a neural network and mixing based on images with these confidence values ​​below the first threshold.

Benefits of technology

Effectively reduces artifact problems during image mixing and improves the quality and authenticity of the display.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN120031992A_ABST
    Figure CN120031992A_ABST
Patent Text Reader

Abstract

The invention relates to generating images using neural networks. Apparatus, systems, and techniques for mixing two or more images based on confidence values for objects in the two or more images. In at least one embodiment, one or more confidence values in one or more images are generated using one or more neural networks, such as for blending two or more images to be displayed.
Need to check novelty before this filing date? Find Prior Art

Claims

1. A processor, comprising: One or more circuits for blending two or more images based at least in part on one or more confidence values ​​below a first threshold using one or more neural networks.

2. The processor according to claim 1, wherein: The one or more circuits are configured to generate the one or more confidence values ​​using the one or more neural networks, the one or more confidence values ​​being configured to indicate whether one or more pixels in the two or more images correspond to an object.

3. The processor according to claim 1, wherein: The one or more circuits are configured to use the one or more neural networks to avoid generating the one or more confidence values ​​within a range that causes artifacts when blending the two or more images.

4. The processor according to claim 1, wherein: The first threshold is the highest confidence value that causes artifacts to appear.

5. The processor according to claim 1, wherein: The two or more images include at least a first image having an object and a second image having a background.

6. The processor of claim 1, wherein: The one or more neural networks are used to generate different confidence values ​​for different display devices.

7. The processor according to claim 1, wherein: The one or more circuits are configured to cause one or more computing devices to display the two or more images in a video conference.

8. A system comprising: One or more processors for blending two or more images based at least in part on one or more confidence values ​​below a first threshold using one or more neural networks.

9. The system according to claim 8, wherein: The one or more neural networks are trained to adjust one or more weights and / or one or more biases to generate the one or more confidence values ​​within a range that avoids artifacts when blending the two or more images.

10. The system according to claim 8, wherein: The one or more processors are configured to generate the one or more confidence values ​​using the one or more neural networks, the one or more confidence values ​​being configured to delineate an object in a first image of the two or more images.

11. The system according to claim 8, wherein: The one or more processors are configured to identify the first threshold based at least in part on a display device.

12. The system according to claim 8, wherein: The two or more images include at least: a first image including a person; and a second image including a background photo.

13. The system according to claim 8, wherein: The one or more processors are configured to adjust a threshold based at least in part on a display device configuration.

14. The system according to claim 8, wherein: The one or more processors are used to cause the robotic device to perform one or more grasping operations using the blended two or more images.

15. A method comprising: The two or more images are blended using one or more neural networks based at least in part on one or more confidence values ​​below a first threshold.

16. The method according to claim 15, wherein: The one or more neural networks are trained to generate the one or more confidence values ​​based at least in part on a configuration of the display device.

17. The method according to claim 15, wherein: The one or more neural networks are configured to identify the first threshold as a maximum confidence value, wherein when the maximum confidence value is exceeded, an artifact is caused to appear in the two or more images.

18. The method according to claim 15, wherein: The two or more images include a first image depicting an object obtained from a camera of a user device and a second image obtained from a database including one or more photographs.

19. The method according to claim 15, wherein: The one or more neural networks are used to generate a range of confidence values ​​for the display device based at least in part on a type of the display device.

20. The method of claim 15, further comprising: The blended two or more images are caused to be displayed by a device of the autonomous vehicle.

Citation Information

Cited By

  • Serialization and deserialization method of IPU acceleration service grid based on FPGA

    CN121560383A