IP Library › Granted Patent US 12682611

Granted Patent B2

US 12682611 · App. 18/199,872 · Granted Jul 14, 2026

Image acquisition model training method and apparatus, image detection method and apparatus, and device

Inventors: Chenchen Qin (Shenzhen, CN); Dasheng Wu (Shenzhen, CN); Jianhua Yao (Shenzhen, CN)

Assignee: TENCENT TECHNOLOGY (SHENZHEN) COMPANY LIMITED

G06V10/774 G06V10/48 G06V10/7715

View Patent ↗

Loading inventors, assignments & file history…

Monitor This Case

Get email alerts when status or documents change.

Order Certified Copies

Most orders are placed with the USPTO same day — all within 24 business hours.

Order via The Patent Place →

Pre-filled with this patent's details

Quick Facts

Patent No.: US 12682611
App. No.: 18/199,872
Granted: Jul 14, 2026
Kind: B2

External Links

USPTO Patent Center↗ Full Patent PDF↗

Abstract

An image acquisition model training method is contemplated. The method includes: acquiring a label image pair of a sample image; acquiring a predicted image pair of the sample image according to a first network model; adjusting the first network model based on the label image pair and the predicted image pair to obtain a second network model, so as to reduce a difference, obtained according to the second network model, between the predicted image pair and the label image pair; and determining the second network model as an image acquisition model in response to the second network model satisfying a training termination condition. By the above method, a heat map of an image in Hough space can be determined more quickly, and an image detection result can be determined based on the heat map of the image in the Hough space, thereby improving the accuracy of subsequent analysis processing.

Claims (73)

1 . An image acquisition model training method, comprising:

acquiring, by an electronic device, a label image pair of a sample image, the label image pair comprising a first label image and a second label image, the first label image being a heat map of the sample image in image space, and the second label image being a heat map of the sample image in Hough space;

acquiring, by the electronic device, a predicted image pair of the sample image according to a first network model, the predicted image pair comprising a first predicted image and a second predicted image, the first predicted image being a heat map of the sample image obtained by the first network model in the image space, and the second predicted image being a heat map of the sample image obtained by the first network model in the Hough space, further comprising:

acquiring, by the electronic device, a first image spatial feature of the sample image in the image space, the first image spatial feature comprising at least two sub-image spatial features;

determining, by the electronic device, the first predicted image based on the at least two sub-image spatial features;

determining, by the electronic device, a first Hough spatial feature of the sample image in the Hough space based on the first image spatial feature; and

determining, by the electronic device, the second predicted image based on the first Hough spatial feature;

adjusting, by the electronic device, the first network model based on the label image pair and the predicted image pair to obtain a second network model, so as to reduce a difference, obtained according to the second network model, between the predicted image pair and the label image pair; and

determining, by the electronic device, the second network model as an image acquisition model in response to the second network model satisfying a training termination condition.

2 . The method according to claim 1 , wherein the determining, by the electronic device, the first Hough spatial feature of the sample image in the Hough space based on the first image spatial feature comprises:

performing rectification, by the electronic device, on the first image spatial feature to obtain a rectified first image spatial feature; and

determining, by the electronic device, the first Hough spatial feature based on the rectified first image spatial feature.

3 . The method according to claim 1 , further comprising:

determining, by the electronic device, the second image spatial feature of the sample image in the image space based on the first Hough spatial feature and the first image spatial feature;

determining, by the electronic device, a third predicted image based on the second image spatial feature, the third predicted image being a heat map of the sample image obtained by the first network model in the image space;

determining, by the electronic device, a second Hough spatial feature of the sample image in the Hough space based on the second image spatial feature;

determining, by the electronic device, a fourth predicted image based on the second Hough spatial feature, the fourth predicted image being a heat map of the sample image obtained by the first network model in the Hough space;

adjusting, by the electronic device, the first network model based on the label image pair, the predicted image pair, the third predicted image, and the fourth predicted image to obtain the second network model, so as to reduce the difference, obtained according to the second network model, between the predicted image pair and the label image pair, and reduce the difference between an image pair composed of the third predicted image and the fourth predicted image and the label image pair.

4 . The method according to claim 1 , wherein the determining, by the electronic device, the first Hough spatial feature based on the first image spatial feature comprises:

for a sub-image spatial feature of the at least two sub-image spatial features, determining, by the electronic device, a sub-Hough spatial feature corresponding to the sub-image spatial feature, the first Hough spatial feature comprising the sub-Hough spatial feature corresponding to the sub-image spatial feature; and

the determining, by the electronic device, the second predicted image based on the first Hough spatial feature comprises:

determining, by the electronic device, the second predicted image based on the sub-Hough spatial feature corresponding to each sub-image spatial feature.

5 . The method according to claim 1 , wherein

the determining, by the electronic device, the first Hough spatial feature based on the first image spatial feature comprises:

for a sub-image spatial feature other than the first sub-image spatial feature of the at least two sub-image spatial features, determining a second fused feature corresponding to the sub-image spatial feature based on the sub-image spatial feature and a previous sub-image spatial feature of the sub-image spatial feature; and

determining, by the electronic device, a sub-Hough spatial feature corresponding to the sub-image spatial feature based on the second fused feature corresponding to the sub-image spatial feature and the previous sub-image spatial feature.

6 . An electronic device, comprising a processor and a memory, the memory storing at least one program code, and at least one program code being loaded and executed by the processor to enable the electronic device to implement an image acquisition model training method including:

acquiring a label image pair of a sample image, the label image pair comprising a first label image and a second label image, the first label image being a heat map of the sample image in image space, and the second label image being a heat map of the sample image in Hough space;

acquiring, by the electronic device, a first image spatial feature of the sample image in the image space, the first image spatial feature comprising at least two sub-image spatial features;

determining, by the electronic device, the first predicted image based on the at least two sub-image spatial features;

determining, by the electronic device, a first Hough spatial feature of the sample image in the Hough space based on the first image spatial feature; and

determining, by the electronic device, the second predicted image based on the first Hough spatial feature;

adjusting the first network model based on the label image pair and the predicted image pair to obtain a second network model, so as to reduce a difference, obtained according to the second network model, between the predicted image pair and the label image pair; and

determining the second network model as an image acquisition model in response to the second network model satisfying a training termination condition.

7 . The electronic device according to claim 6 , wherein the determining the first Hough spatial feature of the sample image in the Hough space based on the first image spatial feature comprises:

performing rectification on the first image spatial feature to obtain a rectified first image spatial feature; and

determining the first Hough spatial feature based on the rectified first image spatial feature.

8 . The electronic device according to claim 6 , wherein the method further comprises:

determining the second image spatial feature of the sample image in the image space based on the first Hough spatial feature and the first image spatial feature;

determining a third predicted image based on the second image spatial feature, the third predicted image being a heat map of the sample image obtained by the first network model in the image space;

determining a second Hough spatial feature of the sample image in the Hough space based on the second image spatial feature;

determining a fourth predicted image based on the second Hough spatial feature, the fourth predicted image being a heat map of the sample image obtained by the first network model in the Hough space;

adjusting the first network model based on the label image pair, the predicted image pair, the third predicted image, and the fourth predicted image to obtain the second network model, so as to reduce the difference, obtained according to the second network model, between the predicted image pair and the label image pair, and reduce the difference between an image pair composed of the third predicted image and the fourth predicted image and the label image pair.

9 . The electronic device according to claim 6 , wherein the determining the first Hough spatial feature based on the first image spatial feature comprises:

determining the second predicted image based on the sub-Hough spatial feature corresponding to each sub-image spatial feature.

10 . The electronic device according to claim 6 , wherein

the determining the first Hough spatial feature based on the first image spatial feature comprises:

11 . A non-transitory computer-readable storage medium, storing at least one program code, and the at least one program code being loaded and executed by a processor of an electronic device to enable the electronic device to implement an image acquisition model training method including:

acquiring, by the electronic device, a first image spatial feature of the sample image in the image space, the first image spatial feature comprising at least two sub-image spatial features;

determining, by the electronic device, the first predicted image based on the at least two sub-image spatial features;

determining, by the electronic device, a first Hough spatial feature of the sample image in the Hough space based on the first image spatial feature; and

determining, by the electronic device, the second predicted image based on the first Hough spatial feature;

determining the second network model as an image acquisition model in response to the second network model satisfying a training termination condition.

12 . The non-transitory computer-readable storage medium according to claim 11 , wherein the determining the first Hough spatial feature of the sample image in the Hough space based on the first image spatial feature comprises:

performing rectification on the first image spatial feature to obtain a rectified first image spatial feature; and

determining the first Hough spatial feature based on the rectified first image spatial feature.

13 . The non-transitory computer-readable storage medium according to claim 11 , wherein the method further comprises:

determining the second image spatial feature of the sample image in the image space based on the first Hough spatial feature and the first image spatial feature;

determining a third predicted image based on the second image spatial feature, the third predicted image being a heat map of the sample image obtained by the first network model in the image space;

determining a second Hough spatial feature of the sample image in the Hough space based on the second image spatial feature;

determining a fourth predicted image based on the second Hough spatial feature, the fourth predicted image being a heat map of the sample image obtained by the first network model in the Hough space;

14 . The non-transitory computer-readable storage medium according to claim 11 , wherein

the determining the first Hough spatial feature based on the first image spatial feature comprises:

determining a sub-Hough spatial feature corresponding to the sub-image spatial feature based on the second fused feature corresponding to the sub-image spatial feature and the previous sub-image spatial feature.