Granted Patent B2

US 12,437,413 · App. 18/673,627 · Granted Oct 7, 2025

Inventors: Yifan Hu (Shenzhen, CN); Yefeng Zheng (Shenzhen, CN)

Assignee: TENCENT TECHNOLOGY (SHENZHEN) COMPANY LIMITED

G06T7/11G06N5/022G06T7/136G06T2207/20081

Abstract

Claims (76)

1. A method for determining a target image region of a target object in a target image, the method comprising:

obtaining, by a device comprising a memory storing instructions and a processor in communication with the memory, an original mask and an image segmentation model, the image segmentation model comprising a first unit model and a second unit model;

downsampling, by the device, the original mask based on a pooling layer in the first unit model to obtain a downsampled mask;

extracting, by the device, region convolution feature information of a target image based on a convolution pooling layer in the second unit model and the downsampled mask;

updating, by the device, the original mask according to the region convolution feature information; and

in response to the updated original mask satisfying an error convergence condition, determining, by the device, a target image region of a target object in the target image according to the updated original mask,

wherein the target object is a lesion object; the target image is a biological tissue image; and the target image region is a lesion image region.

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

obtaining a sample image comprising the target object, and obtaining a sample original mask and a sample image segmentation model, the sample image segmentation model comprising the first unit model and a sample second unit model;

downsampling the sample original mask based on the pooling layer in the first unit model to obtain a sample downsampled mask; and

generating a prediction mask based on the sample second unit model and the sample downsampled mask.

3. The method according to claim 2 , further comprising:

determining a sample mask according to an image region of the target object in the sample image; and

training the sample second unit model according to the sample image, the prediction mask, and the sample mask to obtain the image segmentation model.

4. The method according to claim 3 , wherein the training the sample second unit model according to the sample image, the prediction mask, and the sample mask to obtain the image segmentation model comprises:

generating a brightness loss according to the sample image, the prediction mask, and the sample mask;

generating a segmentation loss according to the prediction mask and the sample mask;

combining the brightness loss and the segmentation loss into a target loss;

determining a parameter value of a model variable parameter in the sample second unit model based on a gradient descent rule and the target loss, and updating the model variable parameter in the sample second unit model according to the parameter value; and

using, in a case that training counts reach a training count threshold, the sample second unit model with an updated model variable parameter as the second unit model, and combining the first unit model and the second unit model into the image segmentation model.

5. The method according to claim 1 , wherein:

the convolution pooling layer in the second unit model comprises a first convolution pooling layer and a second convolution pooling layer; and

the downsampled mask comprises a first original mask corresponding to the first convolution pooling layer and a second original mask corresponding to the second convolution pooling layer.

6. The method according to claim 5 , wherein the extracting the region convolution feature information of the target image based on the convolution pooling layer in the second unit model and the downsampled mask comprises:

performing convolution and pooling on the target image based on the first convolution pooling layer and the first original mask to obtain first convolution feature information, and

performing convolution and pooling on the first convolution feature information based on the second convolution pooling layer and the second original mask to obtain the region convolution feature information of the target image.

7. An apparatus for determining a target image region of a target object in a target image, the apparatus comprising:

a memory storing instructions; and

a processor in communication with the memory, wherein, when the processor executes the instructions, the processor is configured to cause the apparatus to perform:

obtaining an original mask and an image segmentation model, the image segmentation model comprising a first unit model and a second unit model;

downsampling the original mask based on a pooling layer in the first unit model to obtain a downsampled mask;

extracting region convolution feature information of a target image based on a convolution pooling layer in the second unit model and the downsampled mask;

updating the original mask according to the region convolution feature information; and

in response to the updated original mask satisfying an error convergence condition, determining a target image region of a target object in the target image according to the updated original mask,

wherein the target object is a lesion object; the target image is a biological tissue image; and the target image region is a lesion image region.

8. The apparatus according to claim 7 , wherein when the processor executes the instructions, the processor is configured to further cause the apparatus to perform:

obtaining a sample image comprising the target object, and obtaining a sample original mask and a sample image segmentation model, the sample image segmentation model comprising the first unit model and a sample second unit model;

downsampling the sample original mask based on the pooling layer in the first unit model to obtain a sample downsampled mask; and

generating a prediction mask based on the sample second unit model and the sample downsampled mask.

9. The apparatus according to claim 8 , wherein when the processor executes the instructions, the processor is configured to further cause the apparatus to perform:

determining a sample mask according to an image region of the target object in the sample image; and

training the sample second unit model according to the sample image, the prediction mask, and the sample mask to obtain the image segmentation model.

10. The apparatus according to claim 9 , wherein when the processor is configured to cause the apparatus to perform training the sample second unit model according to the sample image, the prediction mask, and the sample mask to obtain the image segmentation model, the processor is configured to cause the apparatus to perform:

generating a brightness loss according to the sample image, the prediction mask, and the sample mask;

generating a segmentation loss according to the prediction mask and the sample mask;

combining the brightness loss and the segmentation loss into a target loss;

determining a parameter value of a model variable parameter in the sample second unit model based on a gradient descent rule and the target loss, and updating the model variable parameter in the sample second unit model according to the parameter value; and

using, in a case that training counts reach a training count threshold, the sample second unit model with an updated model variable parameter as the second unit model, and combining the first unit model and the second unit model into the image segmentation model.

11. The apparatus according to claim 7 , wherein:

the convolution pooling layer in the second unit model comprises a first convolution pooling layer and a second convolution pooling layer; and

the downsampled mask comprises a first original mask corresponding to the first convolution pooling layer and a second original mask corresponding to the second convolution pooling layer.

12. The apparatus according to claim 11 , wherein when the processor is configured to cause the apparatus to perform extracting the region convolution feature information of the target image based on the convolution pooling layer in the second unit model and the downsampled mask, the processor is configured to cause the apparatus to perform:

performing convolution and pooling on the target image based on the first convolution pooling layer and the first original mask to obtain first convolution feature information, and performing convolution and pooling on the first convolution feature information based on the second convolution pooling layer and the second original mask to obtain the region convolution feature information of the target image.

13. A non-transitory computer-readable storage medium, storing computer-readable instructions, wherein, the computer-readable instructions, when executed by a processor, are configured to cause the processor to perform:

obtaining an original mask and an image segmentation model, the image segmentation model comprising a first unit model and a second unit model;

downsampling the original mask based on a pooling layer in the first unit model to obtain a downsampled mask;

extracting region convolution feature information of a target image based on a convolution pooling layer in the second unit model and the downsampled mask;

updating the original mask according to the region convolution feature information; and

in response to the updated original mask satisfying an error convergence condition, determining a target image region of a target object in the target image according to the updated original mask,

wherein the target object is a lesion object; the target image is a biological tissue image; and the target image region is a lesion image region.

14. The non-transitory computer-readable storage medium according to claim 13 , wherein, when the computer-readable instructions are executed by the processor, the computer-readable instructions are configured to cause the processor to perform:

obtaining a sample image comprising the target object, and obtaining a sample original mask and a sample image segmentation model, the sample image segmentation model comprising the first unit model and a sample second unit model;

downsampling the sample original mask based on the pooling layer in the first unit model to obtain a sample downsampled mask; and

generating a prediction mask based on the sample second unit model and the sample downsampled mask.

15. The non-transitory computer-readable storage medium according to claim 14 , wherein, when the computer-readable instructions are executed by the processor, the computer-readable instructions are configured to cause the processor to perform:

determining a sample mask according to an image region of the target object in the sample image; and

training the sample second unit model according to the sample image, the prediction mask, and the sample mask to obtain the image segmentation model.

16. The non-transitory computer-readable storage medium according to claim 15 , wherein, when the computer-readable instructions are configured to cause the processor to perform training the sample second unit model according to the sample image, the prediction mask, and the sample mask to obtain the image segmentation model, the computer-readable instructions are configured to cause the processor to perform:

generating a brightness loss according to the sample image, the prediction mask, and the sample mask;

generating a segmentation loss according to the prediction mask and the sample mask;

combining the brightness loss and the segmentation loss into a target loss;

determining a parameter value of a model variable parameter in the sample second unit model based on a gradient descent rule and the target loss, and updating the model variable parameter in the sample second unit model according to the parameter value; and

using, in a case that training counts reach a training count threshold, the sample second unit model with an updated model variable parameter as the second unit model, and combining the first unit model and the second unit model into the image segmentation model.

17. The non-transitory computer-readable storage medium according to claim 13 , wherein:

the convolution pooling layer in the second unit model comprises a first convolution pooling layer and a second convolution pooling layer; and

the downsampled mask comprises a first original mask corresponding to the first convolution pooling layer and a second original mask corresponding to the second convolution pooling layer.

Priority Claims (1)

Continuity (3)

References Cited (19)