Back to results
US 12,608,952 B2Granted
Selective object tracking using neural networks
AssigneeNVIDIA Corporation
InventorsPankaj Ratnakar Kadtan, Vishesh Gupta
Life of this patent
Estimating the term…
Abstract
Apparatuses, systems, and techniques to track object movements in a video. In at least one embodiment, a rate, at which one or more neural networks is to identify one or more objects within the video, may be adjusted based, at least in part on one or more user-selected portions of a video frame.
Claims20
A processor comprising one or more circuits to;
- determine one or more regions of interest of a frame associated with one or more streams of videos;
- perform, using one or more neural networks, object detection to identify one or more objects within one or more images corresponding to at least one region of interest of the one or more regions of interest, wherein the one or more images are consecutively sampled from the one or more streams according to an inference rate associated with the at least one region of interest;
- and
- cause the inference rate to be dynamically adjusted based, at least in part on whether an object movement is detected within the at least one region of interest based on the object detection from the one or more sampled images.
1 more element
A system comprising:
- one or more processors to:
- determine one or more regions of interest of a frame associated with one or more streams of videos;
- perform, using one or more neural networks, object detection to identify one or more objects within one or more images corresponding to at least one region of interest of the one or more regions of interest, wherein the one or more images are consecutively sampled from the one or more streams according to an inference rate associated with the at least one region of interest;
- and
- cause the inference rate to be dynamically adjusted based, at least in part on whether an object movement is detected within the at least one region of interest based on the object detection from the one or more sampled images;
- and
- one or more memories to store the one or more neural networks.
4 more elements
A method comprising:
- determining one or more regions of interest of a frame associated with one or more streams of videos;
- performing, using one or more neural networks, object detection to identify one or more objects within one or more images corresponding to at least one region of interest of the one or more regions of interest, wherein the one or more images are consecutively sampled from the one or more streams according to an inference rate associated with the at least one region of interest;
- and
- causing the inference rate to be dynamically adjusted based, at least in part on whether an object movement is detected within the at least one region of interest based on the object detection from the one or more sampled images.
1 more element
Dependencies are inferred from the claim text.