Systems and methods for tracking and detecting a target object
AssigneeQUALCOMM Incoporated
InventorsDashan Gao, Xin Zhong, Yingyong Qi, Ning Bi
Life of this patent
Estimating the term…
Abstract
A method for detecting and tracking a target object is described. The method includes performing motion-based tracking for a current video frame by comparing a previous video frame and the current video frame. The method also includes selectively performing object detection in the current video frame based on a tracked parameter.
Claims60
A method for detecting and tracking a target object, comprising:
- performing motion-based tracking by comparing a previous video frame and a current video frame;
- determining a tracking error between the previous video frame and the current video frame;
- using the tracking error to obtain a tracking confidence value that corresponds to a likelihood that the target object is in the current video frame;
- selectively performing object detection in the current video frame based on the tracking confidence value, wherein selectively performing object detection comprises performing object detection on the current video frame when the tracking confidence value is less than or equal to a tracking threshold;
- determining a detection confidence value for the current video frame;
- determining a combined detection and tracking confidence value by selecting a larger value from among the tracking confidence value and the detection confidence value;
- and
- determining whether to narrow or expand a search range to detect the target object in a next video frame based on the combined detection and tracking confidence value.
An apparatus for detecting and tracking a target object, comprising:
- means for performing motion-based tracking by comparing a previous video frame and a current video frame;
- means for determining a tracking error between the previous video frame and the current video frame;
- means for using the tracking error to obtain a tracking confidence value that corresponds to a likelihood that the target object is in the current video frame;
- means for selectively performing object detection in the current video frame based on the tracking confidence value, wherein the means for selectively performing object detection comprises means for performing object detection on the current video frame when the tracking confidence value is less than or equal to a tracking threshold;
- means for determining a detection confidence value for the current video frame;
- means for determining a combined detection and tracking confidence value by selecting a larger value from among the tracking confidence value and the detection confidence value;
- and
- means for determining whether to narrow or expand a search range to detect the target object in a next video frame based on the combined detection and tracking confidence value.
An electronic device for detecting and tracking a target object, comprising:
- a processor;
- memory in electronic communication with the processor;
- instructions stored in memory, the instructions being executable to:
- perform motion-based tracking by comparing a previous video frame and a current video frame;
- determine a tracking error between the previous video frame and the current video frame;
- use the tracking error to obtain a tracking confidence value that corresponds to a likelihood that the target object is in the current video frame;
- selectively perform object detection in the current video frame based on the tracking confidence value, wherein the instructions being executable to selectively perform object detection comprise instructions being executable to perform object detection on the current video frame when the tracking confidence value is less than or equal to a tracking threshold;
- determine a detection confidence value for the current video frame;
- determine a combined detection and tracking confidence value by selecting a larger value from among the tracking confidence value and the detection confidence value;
- and
- determine whether to narrow or expand a search range to detect the target object in a next video frame based on the combined detection and tracking confidence value.
A computer-program product for detecting and tracking a target object, comprising a non-transitory computer-readable medium having instructions thereon, the instructions comprising:
- code for causing an electronic device to perform motion-based tracking by comparing a previous video frame and a current video frame;
- code for determining a tracking error between the previous video frame and the current video frame;
- code for using the tracking error to obtain a tracking confidence value that corresponds to a likelihood that the target object is in the current video frame;
- code for causing the electronic device to selectively perform object detection in the current video frame based on the tracking confidence value, wherein the code for causing the electronic device to selectively perform object detection comprises code for causing the electronic device to perform object detection on the current video frame when the tracking confidence value is less than or equal to a tracking threshold;
- code for causing the electronic device to determine a detection confidence value for the current video frame;
- code for causing the electronic device to determine a combined detection and tracking confidence value by selecting a larger value from among the tracking confidence value and the detection confidence value;
- and
- code for causing the electronic device to determine whether to narrow or expand a search range to detect the target object in a next video frame based on the combined detection and tracking confidence value.
Dependencies are inferred from the claim text.