Method and apparatus with target tracking

Patent metadata

Field	Value
Publication number	US-12423835-B2
Application number	US-202218084003-A
Country	US
Kind code	B2
Filing date	Dec 19, 2022
Priority date	Dec 17, 2021
Publication date	Sep 23, 2025
Grant date	Sep 23, 2025

Abstract

A processor-implemented method with target tracking includes: generating a first target tracking result based on a search region of a current frame image; determining a scale feature of the first target tracking result; predicting a scale of a target in the search region based on the scale feature of the first target tracking result; and generating a second target tracking result by adjusting the first target tracking result based on a scale predicting result.

First claim

What is claimed is:

1. A processor-implemented method with target tracking, the method comprising:
generating a first target tracking result based on a search region of a current frame image;
determining a scale feature of the first target tracking result;
predicting a scale of a target in the search region based on the scale feature of the first target tracking result; and
generating a second target tracking result by adjusting the first target tracking result based on a scale predicting result, and
determining whether target tracking has succeeded based on the scale feature of the first target tracking result.

2. The method of claim 1 , wherein the search region is either one of an entire region of the current frame image and a region larger than a target tracking result of a previous frame image.

3. The method of claim 1 , wherein the search region is determined based on a target tracking result of a previous frame image.

4. The method of claim 1 , wherein
the first target tracking result comprises a first tracking bounding box, and
the second target tracking result comprises a second tracking bounding box.

5. The method of claim 1 , wherein the determining of whether the target tracking has succeeded based on the scale feature of the first target tracking result comprises:
determining an appearance feature of the first target tracking result; and
determining whether the target tracking has succeeded based on the appearance feature of the first target tracking result and the scale feature of the first target tracking result.

6. The method of claim 5 , further comprising:
adjusting the first target tracking result based on the appearance feature of the first target tracking result.

7. The method of claim 1 , wherein
the determining of the scale feature of the first target tracking result comprises:
determining a multi-scale template region of interest (ROI) feature;
determining an ROI feature of the first target tracking result; and
determining the scale feature of the first target tracking result based on the multi-scale template ROI feature and the ROI feature of the first target tracking result, and
the ROI feature of the first target tracking result comprises ROI features of respective scales.

8. The method of claim 7 , wherein the determining of the scale feature of the first target tracking result based on the multi-scale template ROI feature and the ROI feature of the first target tracking result comprises:
performing feature alignment of each of the ROI features of the respective scales comprised in the ROI feature of the first target tracking result, based on an appearance feature of the first target tracking result; and
determining the scale feature of the first target tracking result based on the feature-aligned ROI features of the respective scales.

9. The method of claim 7 , wherein the determining of the scale feature of the first target tracking result based on the multi-scale template ROI feature and the ROI feature of the first target tracking result comprises:
determining the scale feature of the first target tracking result by calculating a correlation between the multi-scale template ROI feature and the ROI feature of the first target tracking result.

10. The method of claim 9 , wherein the determining of the scale feature of the first target tracking result by calculating the correlation between the multi-scale template ROI feature and the ROI feature of the first target tracking result comprises:
calculating each correlation between the ROI feature of each scale from the ROI feature of the first target tracking result and a scale of the multi-scale template ROI feature.

11. The method of claim 7 , wherein
the ROI feature of the first target tracking result comprises a single-scale ROI feature, and
the scale feature of the first target tracking result comprises a one-dimensional scale feature.

12. The method of claim 7 , wherein
the ROI feature of the first target tracking result comprises a multi-scale ROI feature, and
the scale feature of the first target tracking result comprises a two-dimensional scale feature.

13. An apparatus with target tracking, the apparatus comprising:
a processor configured to:
determine a first target tracking result based on a search region of a current frame image;
obtain a scale feature of the first target tracking result;
predict a scale of a target in the search region based on the scale feature of the first target tracking result;
obtain a second target tracking result by adjusting the first target tracking result based on a scale predicting result, and
determine whether target tracking has succeeded based on the scale feature of the first target tracking result.

14. The apparatus of claim 13 , wherein the search region is either one of an entire region of the current frame image and a region larger than a target tracking result of a previous frame image.

15. The apparatus of claim 13 , wherein the search region is determined based on a target tracking result of a previous frame image.

16. The apparatus of claim 13 , wherein, for the determining of whether the target tracking has succeeded based on the scale feature of the first target tracking result, the processor is configured to:
determine an appearance feature of the first target tracking result; and
determine whether the target tracking has succeeded based on the appearance feature of the first target tracking result and the scale feature of the first target tracking result.

17. The apparatus of claim 13 , wherein
for the determining of the scale feature of the first target tracking result, the processor is configured to:
determine a multi-scale template region of interest (ROI) feature;
determine an ROI feature of the first target tracking result; and
determine the scale feature of the first target tracking result based on the multi-scale template ROI feature and the ROI feature of the first target tracking result, and
the ROI feature of the first target tracking result comprises ROI features of respective scales.

18. The apparatus of claim 17 , wherein, for the determining of the scale feature of the first target tracking result based on the multi-scale template ROI feature and the ROI feature of the first target tracking result, the processor is configured to:
determine the scale feature of the first target tracking result by calculating a correlation between the multi-scale template ROI feature and the ROI feature of the first target tracking result.

19. A processor-implemented method with target tracking, the method comprising:
determining a multi-scale template region of interest (ROI) feature based on a first target tracking result generated based on a search region of a frame image;
determining an ROI feature of the first target tracking result;
determining a scale feature of the first target tracking result based on the multi-scale template ROI feature and the ROI feature of the first target tracking result;
predicting a scale of a target in the search region based on the scale feature of the first target tracking result; and
generating a second target tracking result by adjusting the first target tracking result based on a result of the predicting.

Assignees

• Samsung Electronics Co Ltd

Inventors

• Yu Jiaqian
• Pan Siyang
• Park Changbeom
• Lee Hyunjeong
• Yoo Byung In
• Wang Qiang
• Chen Yiwei

Classifications

• G06T2207/20016

  Hierarchical, coarse-to-fine, multiscale or multiresolution image processing; Pyramid transform · CPC title

• G06T2207/20084

  Artificial neural networks [ANN] · CPC title

• G06T7/248 Primary

  involving reference images or patches · CPC title