/Patent Tracker
NvidiaPatent applicationAI & softwareInterest 6/10

Nvidia patents method to keep object detection accurate in bad weather

Patent drawing
Patent drawing from the official filing.

Nvidia describes a system that continuously updates computer vision models by selecting images that match specific environmental conditions. The system adds digital noise to these images to simulate real-world imperfections like rain or fog, then uses them to retrain the model so it can still identify objects clearly.

Why it mattersThis approach helps self-driving cars and security systems maintain accuracy when visibility drops due to weather or lighting changes.

In the filing’s own words

Apparatuses, systems, and techniques for continuous training of an object detection and/or classification model. A set of images each depicting one or more objects in an environment is obtained. Of the set of images, a selection is made of a respective image satisfying one or more image criteria associated with training a machine learning model to perform one or more object detection operations with respect to images depicting objects in the environment having a respective environmental condition. One or more augmented training images are generated by augmenting the respective image to include one or more noise characteristics associated with the respective environmental condition. Training data is provided for updating one or more parameters of the machine learning model to perform the one or more object detection operations with respect to images depicting the respective environmental condition, the training data including the one or more augmented training images.

What exactly they claim (claim 1)

1. A method comprising: obtaining a plurality of images each depicting one or more objects in an environment; selecting, of the plurality of images, a respective image satisfying one or more image criteria associated with training a machine learning model to perform one or more object detection operations with respect to images depicting objects in the environment having a respective environmental condition; generating one or more augmented training images by augmenting the respective image to include one or more noise characteristics associated with the respective environmental condition; and providing training data for updating one or more parameters of the machine learning model to perform the one or more object detection operations with respect to images depicting the respective environmental condition, the training data comprising the one or more augmented training images.

Facts

Company
Nvidia (NVIDIA Corporation)
Official title
CONTINUOUS TRAINING OF AN OBJECT DETECTION AND CLASSIFICATION MODEL FOR VARYING ENVIRONMENTAL CONDITIONS
Publication number
US20260330821A1
Application number
19724717
Filed
Jun 29, 2026
Published
Oct 22, 2026
Inventors
3
Claims
20

Classified under

G06V 10/82 Physics & computingG06T 5/70 Physics & computingG06V 10/774 Physics & computingG06V 20/20 Physics & computing
Read the full filing (official PDF) ↗ Open in the Patent Tracker

More from Nvidia in AI & software

Summaries are simplified from the filing’s own abstract and first claim. A patent filing is not a product announcement.
Patent data and drawings: U.S. Patent and Trademark Office (USPTO). Nebula is not affiliated with or endorsed by the USPTO.