Defect detection method and apparatus, and computer-readable storage medium
View Patent ↗Provided are a defect detection method and apparatus, and a computer-readable storage medium. Specifically, the method includes: obtaining a to-be-detected image; obtaining a feature map of the to-be-detected image based on the to-be-detected image, where the feature map of the to-be-detected image includes a feature map of spatial position coordinate information; and performing defect detection on the to-be-detected image based on the feature map of the to-be-detected image. By modifying a neural network structure of defect detection and extracting the feature map of spatial position coordinate information during the detection, this application makes the neural network for use of defect detection sensitive to a spatial position, thereby enhancing sensitivity of a detection neural network to the spatial position, and in turn, increasing accuracy of detecting some specific defect types by the detection neural network, and increasing accuracy of defect detection.
1. A defect detection method, characterized in that the method comprises:
obtaining a to-be-detected image;
obtaining a feature map of the to-be-detected image based on the to-be-detected image, wherein the feature map of the to-be-detected image comprises a feature map of spatial position coordinate information; and
performing defect detection on the to-be-detected image based on the feature map of the to-be-detected image; and, wherein the feature map of the to-be-detected image further comprises an image information feature map, and the obtaining a feature map of the to-be-detected image based on the to-be-detected image comprises:
obtaining the image information feature map based on the to-be-detected image; and
generating the feature map of the spatial position coordinate information based on the image information feature map, wherein a number of dimensions of the feature map of the spatial position coordinate information is identical to a number of dimensions of the image information feature map; and, wherein generating the feature map of the spatial position coordinate information based on the image information feature map comprises:
generating a linear value corresponding to the spatial position coordinate information;
generating a first coordinate network based on the linear value; and
increasing a number of dimensions of the first coordinate network based on the image information feature map, so as to generate the feature map of the spatial position coordinate information.
2. The method according to claim 1 , characterized in that the performing defect detection on the to-be-detected image based on the feature map of the to-be-detected image comprises:
performing, based on the feature map of the to-be-detected image, defect detection on the to-be-detected image by using a filter model, wherein the filter model comprises a filter configured to process the feature map of the spatial position coordinate information.
3. The method according to claim 2 , characterized in that the filter is a filter of a 1 st convolutional layer.
4. The method according to claim 1 , characterized in that the feature map of the spatial position coordinate information comprises at least one of a feature map of coordinate information in an x-axis direction, a feature map of coordinate information in a y-axis direction, or a feature map of coordinate information in a z-axis direction.
5. The method according to claim 1 , characterized in that the method is used for defect detection of tabs and/or electrode plates.
6. The method according to claim 5 , characterized in that when the method is used for defect detection of tabs, the defect detection of tabs comprises detection of defects of a tab fold feature.
7. The method according to claim 1 , characterized in that the obtaining a feature map of the to-be-detected image based on the to-be-detected image comprises:
inputting the to-be-detected image into a neural network; and
extracting, through a backbone network of the neural network, a defect feature vector and extracting a coordinate information feature corresponding to the defect feature vector, so as to obtain the feature map of the to-be-detected image.
8. A defect detection apparatus, characterized in that the apparatus comprises:
an obtaining unit, configured to obtain a to-be-detected image; and
a processing unit, configured to obtain a feature map of the to-be-detected image based on the to-be-detected image, wherein the feature map of the to-be-detected image comprises a feature map of spatial position coordinate information, wherein
the processing unit is further configured to perform defect detection on the to-be-detected image based on the feature map of the to-be-detected image; and, wherein the feature map of the to-be-detected image further comprises an image information feature map, and the obtaining a feature map of the to-be-detected image based on the to-be-detected image comprises:
obtaining the image information feature map based on the to-be-detected image; and
generating the feature map of the spatial position coordinate information based on the image information feature map, wherein a number of dimensions of the feature map of the spatial position coordinate information is identical to a number of dimensions of the image information feature map; and, wherein generating the feature map of the spatial position coordinate information based on the image information feature map comprises:
generating a linear value corresponding to the spatial position coordinate information;
generating a first coordinate network based on the linear value; and
increasing a number of dimensions of the first coordinate network based on the image information feature map, so as to generate the feature map of the spatial position coordinate information.
9. A defect detection apparatus, characterized in that the apparatus comprises a processor and a memory, the memory is configured to store a program, and the processor is configured to call the program from the memory and run the program to perform a defect detection method, wherein the method comprises:
obtaining a to-be-detected image;
obtaining a feature map of the to-be-detected image based on the to-be-detected image, wherein the feature map of the to-be-detected image comprises a feature map of spatial position coordinate information; and
performing defect detection on the to-be-detected image based on the feature map of the to-be-detected image; and, wherein the feature map of the to-be-detected image further comprises an image information feature map, and the obtaining a feature map of the to-be-detected image based on the to-be-detected image comprises:
obtaining the image information feature map based on the to-be-detected image; and
generating the feature map of the spatial position coordinate information based on the image information feature map, wherein a number of dimensions of the feature map of the spatial position coordinate information is identical to a number of dimensions of the image information feature map; and, wherein generating the feature map of the spatial position coordinate information based on the image information feature map comprises:
generating a linear value corresponding to the spatial position coordinate information;
generating a first coordinate network based on the linear value; and
increasing a number of dimensions of the first coordinate network based on the image information feature map, so as to generate the feature map of the spatial position coordinate information.
10. A non-transitory computer-readable storage medium, characterized in that the storage medium comprises a computer program, and, when executed on a computer, the computer program causes the computer to perform a defect detection method, wherein the method comprises:
obtaining a to-be-detected image;
obtaining a feature map of the to-be-detected image based on the to-be-detected image, wherein the feature map of the to-be-detected image comprises a feature map of spatial position coordinate information; and
performing defect detection on the to-be-detected image based on the feature map of the to-be-detected image; and, wherein the feature map of the to-be-detected image further comprises an image information feature map, and the obtaining a feature map of the to-be-detected image based on the to-be-detected image comprises:
obtaining the image information feature map based on the to-be-detected image; and
generating the feature map of the spatial position coordinate information based on the image information feature map, wherein a number of dimensions of the feature map of the spatial position coordinate information is identical to a number of dimensions of the image information feature map; and, wherein generating the feature map of the spatial position coordinate information based on the image information feature map comprises:
generating a linear value corresponding to the spatial position coordinate information;
generating a first coordinate network based on the linear value; and
increasing a number of dimensions of the first coordinate network based on the image information feature map, so as to generate the feature map of the spatial position coordinate information.