The United States Patent and Trademark Office (USPTO) has granted Dell Products L.P. patent US12586152B2, titled 'Method, electronic device, and computer program product for training image processing model.' The patent, effective March 24, 2026, covers a novel method for training image processing models. Key aspects include obtaining folding weights from convolutional layers of a pre-trained generator via a folding operation and embedding this generator into the image processing model. The training process utilizes pairs of sample images with different resolutions, specifically including a first sample image with a lower resolution and a second sample image with a higher resolution.

This patent grant represents a new intellectual property for Dell related to AI and machine learning in image processing. While it does not impose direct compliance obligations on other entities, it signifies innovation in the field and may influence future product development and licensing strategies for companies involved in AI-driven image analysis. Companies developing or utilizing similar image processing models should be aware of this patent's existence and scope, particularly concerning the described training methodologies involving resolution variations and generator embedding.