Neural network high resolution color information to project image data
Transformer-based neural networks with hierarchical processing stages and color upsampling techniques efficiently transform low-resolution images into high-resolution images, optimizing resource use and improving image quality and frame rates.
WO2026148205A1PCT designated stage Publication Date: 2026-07-09NVIDIA CORP
Patent Information
- Authority / Receiving Office
- WO · WO
- Patent Type
- Applications
- Current Assignee / Owner
- NVIDIA CORP
- Filing Date
- 2026-01-02
- Publication Date
- 2026-07-09
AI Technical Summary
Technical Problem
Existing neural networks for image upscaling consume significant memory, time, and computing resources, making them inefficient for transforming low-resolution images into high-resolution images.
Method used
The use of transformer-based neural networks with hierarchical processing stages and color upsampling techniques, including optical flow and sampling methods, to generate high-resolution image data while optimizing resource utilization.
Benefits of technology
This approach reduces GPU and CPU resource utilization, enhances image quality, and increases frame rates by generating additional frames with reduced computational overhead.
✦ Generated by Eureka AI based on patent content.
Smart Images
Figure US2026010046_09072026_PF_FP_ABST
Abstract
Apparatuses, systems, and techniques to update lower resolution images. In at least one embodiment, color information from one or more upsampled images may be obtained so that the color information from the one or more upsampled images may be caused to be applied to one or more subsequent lower resolution images.
Need to check novelty before this filing date? Find Prior Art