Image annotation method, device and system

Patent metadata

Field	Value
Publication number	US-12106465-B2
Application number	US-202117628364-A
Country	US
Kind code	B2
Filing date	Mar 26, 2021
Priority date	Mar 31, 2020
Publication date	Oct 1, 2024
Grant date	Oct 1, 2024

Abstract

An image marking method, apparatus and system, which relates to the technical field of image processing. The present disclosure includes, when the working mode of the first client is a first mode, receiving a first marking task assigned by a second client, on the condition that the image marking approach is the first marking approach, according to a neural network model, determining a first marking result corresponding to the first original image; on the condition that the image marking approach is the second marking approach, according to an unsupervised algorithm model, determining a second marking result corresponding to the first original image; on the condition that the image marking approach is the third marking approach, receiving a third marking result inputted by a user into the first original image; and sending a target marking result to the second client.

First claim

The invention claimed is:

1. An image marking method, wherein the method is applied to a first client, and the method comprises:
determining a working mode of the first client;
on the condition that the working mode is a first mode, receiving a first marking task assigned by a second client, wherein the first marking task comprises performing defect marking to a first original image, and an image marking approach for performing the defect marking to the first original image comprises a first marking approach, a second marking approach and a third marking approach;
on the condition that the image marking approach is the first marking approach, according to a neural network model, determining a first marking result corresponding to the first original image by using a neural network model;
on the condition that the image marking approach is the second marking approach, according to an unsupervised algorithm model, determining a second marking result corresponding to the first original image;
on the condition that the image marking approach is the third marking approach, receiving a third marking result inputted by a user into the first original image; and
sending a target marking result to the second client, wherein the target marking result comprises any one of the first marking result, the second marking result and the third marking result.

2. The method according to claim 1 , wherein the step of, on the condition that the image marking approach is the first marking approach, according to the neural network model, determining the first marking result corresponding to the first original image comprises:
on the condition that the image marking approach is the first marking approach, acquiring the first marking result obtained after a server performs defect marking to the first original image in the server through the neural network model selected by the user.

3. The method according to claim 2 , wherein the neural network model is a target detection-algorithm model, and the server stores a plurality of categories of target detection-algorithm models; and
each of the categories of the target detection-algorithm models is obtained by training according to a plurality of sample images of a same category and a marking result of each of the sample images acquired in advance, wherein a quantity of the sample images is less than a predetermined quantity.

4. The method according to claim 1 , wherein the step of, on the condition that the image marking approach is the second marking approach, according to the unsupervised algorithm model, determining the second marking result corresponding to the first original image comprises:
on the condition that the image marking approach is the second marking approach, according to the unsupervised algorithm model, performing defect marking to the first original image in a server, to obtain the second marking result.

5. The method according to claim 1 , wherein the unsupervised algorithm model is a fast-Fourier-transform-algorithm model, and the fast-Fourier-transform-algorithm model is stored in the first client.

6. The method according to claim 1 , wherein before the step of sending the target marking result to the second client, the method further comprises:
displaying in an image marking interface of the first client the target marking result corresponding to the first original image; and
on the condition that an adjusting operation to the target marking result by the user is received, adjusting the target marking result.

7. The method according to claim 6 , wherein after the step of, on the condition that the adjusting operation to the target marking result by the user is received, adjusting the target marking result, the method further comprises:
on the condition that an eighth operation to a saving option in the image marking interface by the user is received, saving the target marking result obtained after the adjustment.

8. The method according to claim 6 , wherein after the step of displaying in the image marking interface of the first client the target marking result corresponding to the first original image, the method further comprises:
on the condition that a ninth operation to a marking option in the image marking interface by the user is received, marking the target marking result, to obtain a marking result to be processed.

9. The method according to claim 8 , wherein the step of sending the target marking result to the second client comprises:
on the condition that a tenth operation to a submitting option in the image marking interface by the user is received, determining whether the marking result to be processed exists in the first original image in the first marking task;
on the condition that the marking result to be processed does not exist, sending the target marking result to the second client; and
on the condition that the marking result to be processed exists, displaying a prompt box in the image marking interface, to prompt the user to adjust or save the marking result to be processed.

10. The method according to claim 1 , wherein the target marking result comprises a defect category and a defect position of the first original image.

11. The method according to claim 1 , wherein after the step of determining the working mode of the first client, the method further comprises:
on the condition that the working mode is a second mode, according to a characteristic of distribution of patterns in a second original image in a server, assigning a second marking task to a third client, wherein the second marking task comprises performing defect marking to the second original image, and an image marking approach for performing the defect marking to the second original image comprises a first marking approach, a second marking approach and a third marking approach; and
receiving the target marking result obtained after the third client marks the second original image according to the image marking approach.

12. The method according to claim 11 , wherein the step of, according to the characteristic of the distribution of the patterns in the second original image in the server, assigning the second marking task to the third client comprises:
on the condition that the patterns in the second original image are distributed in an array, determining that the image marking approach corresponding to the second original image is the second marking approach;
on the condition that the patterns in the second original image are not distributed in an array and differences between the second original images are small, determining that the image marking approach corresponding to the second original image is the first marking approach; and
on the condition that differences between the second original images are large and an image background of the second original image is complicated, determining that the image marking approach corresponding to the second original image is the third marking approach.

13. The method according to claim 11 , wherein after the step of receiving the target marking result obtained after the third client marks the second original image according to the image marking approach, the method further comprises:
receiving an examination result with respect to the target marking result inputted by the user into the first client;
on the condition that the examination result has no error, sending the target marking result to the server; and
on the condition that the examination result has an error, returning the target marking result to the third client, and re-performing the defect marking.

14. The method according to claim 11 , wherein the step of determining the w

Assignees

• Boe Technology Group Co Ltd

Inventors

• Wang Yaoping
• Zhang Meijuan
• Liu Yongzhang
• Li Zhaoyue
• Chai Dong
• Wang Hong

Classifications

• G06N3/0464

  Convolutional networks [CNN, ConvNet] · CPC title

• G06N3/09

  Supervised learning · CPC title

• G06T2207/30204

  Marker · CPC title

• G06T2207/20084

  Artificial neural networks [ANN] · CPC title

• G06T2207/20081

  Training; Learning · CPC title