Method, apparatus, and non-transitory computer readable medium for detecting welding defects of workpieces

1. A method for detecting welding defects of workpieces comprising:

obtaining a first image from an image capturing element;

extracting welding information from the first image;

transmitting the welding information to a logic processing element to obtain welding defects;

wherein the logic processing element is formed by evolution of self-learning of historical welding information; and

generating display information of the welding defects and displaying the welding defects in visual form and/or characteristics values of the welding defects;

wherein the welding defects comprises at least one of a quantify of missing welding, a quantify of lacking welding, a ratio of over welding, a ratio of welding displacement, a ratio of welding defects, a ratio of welding holes, and a displacement degree of the first workpiece;

wherein the generating display information of the welding defects and displaying the welding defects in visual form and/or characteristics values of the welding defects comprises:

generating display information of the welding defects of at least one of the quantify of missing welding, the quantify of lacking welding, the ratio of over welding, the ratio of welding displacement, the ratio of welding defects, the ratio of welding holes, and the displacement degree of the first workpiece, and displaying the welding defects in visual form and/or characteristics values of the welding defects corresponding to the display information.

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

generating a second image according to the first image and the display information.

3. The method according to claim 2 , wherein the first image comprises a first region, the welding information comprises a first welding information, the first welding information is extracted based on the first region, the display information comprises first display information indicated to the welding defects corresponding to the first welding information;

wherein the generating the second image according to the first image and the display information comprises:

determining the first region corresponding to the first display information; and

adding the first display information to a display region of the first region and thereby generating the second image.

4. The method according to claim 1 , wherein the welding information comprises at least one of a first contour of a first workpiece, a second contour of a second workpiece, welding positions, welding defect points, and a welding center, the first workpiece and the second workpiece are welded together;

wherein the transmitting the welding information to a logic processing element to obtain welding defects comprises:

extracting at least one of the first contour, the second contour, the welding positions, the welding defect points, and the welding center according to the first image; and

transmitting the at least one of the first contour, the second contour, the welding positions, the welding defect points, and the welding center to the logic processing element to obtain the welding defects.

5. The method according to claim 1 , wherein the historical welding information comprises at least one historical image marked by historical welding defects;

wherein the logic processing element formed by evolution of self-learning of historical welding information comprises:

extending the historical images to form a training set;

transmitting the training set to an initial model to generate welding characteristics;

transmitting the welding characteristics to a classifier to generate classified results;

transmitting the historical welding defect information and the classified results to an evaluation model to generate evaluation results;

determining whether the evaluation results match a predetermined condition; and

forming the logic processing element according to the initial model and the evaluation results when the evaluation results match the predetermined condition;

wherein the predetermined condition comprises convergence of loss function and a predetermined value of identification precision.

6. The method according to claim 5 , wherein the initial model is a semantic incision model and comprises a convolutional layer and a pooling layer;

wherein transmitting the training set to an initial model to generate welding characteristics comprises:

transmitting the training set to the convolutional layer and generating convolutional results; and

transmitting the convolutional results to the pooling layer and generating the welding characteristics.

7. The method according to claim 1 , wherein the transmitting the welding information to a logic processing element to obtain welding defects comprises:

transmitting the welding information to the logic processing element to obtain welding characteristics of the welding information; and

transmitting the welding characteristics to a classifier to generate the welding defects.

8. The method according to claim 1 , wherein the welding information comprises a first welding position and a second welding position;

wherein the transmitting the welding information to a logic processing element to obtain welding defects comprises:

transmitting the welding information to the logic processing element to obtain welding defects; wherein the welding defects comprises a first welding defect formed at the first welding position and a second welding defect formed at the second welding position; and

generating a first welding defect display information corresponding to the first welding defect and a second welding defect display information corresponding to the second welding defect, and displaying the welding defects in visual form and/or characteristics values of the welding defects according to the first welding defect display information and the second welding defect display information.

9. A apparatus for detecting welding defects of workpieces comprising:

at least one processor; and

at least one memory coupled to the at least one processor and storing program instructions;

the memory and the program instructions configured to, with the at least one processor, cause the apparatus to perform:

obtaining a first image from an image capturing element;

extracting welding information from the first image;

transmitting the welding information to a logic processing element to obtain welding defects;

wherein the logic processing element is formed by evolution of self-learning of historical welding information; and

generating display information of the welding defects and displaying the welding defects in visual form and/or characteristics values of the welding defects;

wherein the welding defects comprises at least one of a quantify of missing welding, a quantify of lacking welding, a ratio of over welding, a ratio of welding displacement, a ratio of welding defects, a ratio of welding holes, and a displacement degree of the first workpiece;

wherein the generating display information of the welding defects and displaying the welding defects in visual form and/or characteristics values of the welding defects comprises:

generating display information of the welding defects of at least one of the quantify of missing welding, the quantify of lacking welding, the ratio of over welding, the ratio of welding displacement, the ratio of welding defects, the ratio of welding holes, and the displacement degree of the first workpiece, and displaying the welding defects in visual form and/or characteristics values of the welding defects corresponding to the display information.

10. The apparatus according to claim 9 , wherein the memory and the program instructions configured to, with the at least one processor, cause the apparatus further to perform:

generating a second image according to the first image and the display information.

11. The apparatus according to claim 10 , wherein the first image comprises a first region, the welding information comprises a first welding information, the first welding information is extracted based on the first region, the display information comprises first display information indicated to the welding defects corresponding to the first welding information;

wherein the generating the second image according to the first image and the display information comprises:

determining the first region corresponding to the first display information; and

adding the first display information to a display region of the first region and thereby generating the second image.

12. The apparatus according to claim 9 , wherein the welding information comprises at least one of a first contour of a first workpiece, a second contour of a second workpiece, welding positions, welding defect points, and a welding center, the first workpiece and the second workpiece are welded together;

wherein the transmitting the welding information to a logic processing element to obtain welding defects comprises:

extracting at least one of the first contour, the second contour, the welding positions, the welding defect points, and the welding center according to the first image; and

transmitting the at least one of the first contour, the second contour, the welding positions, the welding defect points, and the welding center to the logic processing element to obtain the welding defects.

13. The apparatus according to claim 9 , wherein the historical welding information comprises at least one historical image marked by historical welding defects;

wherein the logic processing element formed by evolution of self-learning of historical welding information comprises:

extending the historical images to form a training set;

transmitting the training set to an initial model to generate welding characteristics;

transmitting the welding characteristics to a classifier to generate classified results;

transmitting the historical welding defect information and the classified results to an evaluation model to generate evaluation results;

determining whether the evaluation results match a predetermined condition; and

forming the logic processing element according to the initial model and the evaluation results when the evaluation results match the predetermined condition;

wherein the predetermined condition comprises convergence of loss function and a predetermined value of identification precision.

14. The apparatus according to claim 13 , wherein the initial model is a semantic incision model and comprises a convolutional layer and a pooling layer;

wherein transmitting the training set to an initial model to generate welding characteristics comprises:

transmitting the training set to the convolutional layer and generating convolutional results; and

transmitting the convolutional results to the pooling layer and generating the welding characteristics.

15. The apparatus according to claim 9 , wherein the transmitting the welding information to a logic processing element to obtain welding defects comprises:

transmitting the welding information to the logic processing element to obtain welding characteristics of the welding information; and

transmitting the welding characteristics to a classifier to generate the welding defects.

16. The apparatus according to claim 9 , wherein the welding information comprises a first welding position and a second welding position;

wherein the transmitting the welding information to a logic processing element to obtain welding defects comprises:

transmitting the welding information to the logic processing element to obtain welding defects; wherein the welding defects comprises a first welding defect formed at the first welding position and a second welding defect formed at the second welding position; and

generating a first welding defect display information corresponding to the first welding defect and a second welding defect display information corresponding to the second welding defect, and displaying the welding defects in visual form and/or characteristics values of the welding defects according to the first welding display information and the second welding defect display information.

17. A non-transitory computer readable medium comprising program instructions for causing an apparatus to perform a method comprising:

obtaining a first image from an image capturing element;

extracting welding information from the first image;

transmitting the welding information to a logic processing element to obtain welding defects;

wherein the logic processing element is formed by evolution of self-learning of historical welding information; and

generating display information of the welding defects and displaying the welding defects in visual form and/or characteristics values of the welding defects;

wherein the welding defects comprises at least one of a quantify of missing welding, a quantify of lacking welding, a ratio of over welding, a ratio of welding displacement, a ratio of welding defects, a ratio of welding holes, and a displacement degree of the first workpiece;

wherein the generating display information of the welding defects and displaying the welding defects in visual form and/or characteristics values of the welding defects comprises:

generating display information of the welding defects of at least one of the quantify of missing welding, the quantify of lacking welding, the ratio of over welding, the ratio of welding displacement, the ratio of welding defects, the ratio of welding holes, and the displacement degree of the first workpiece, and displaying the welding defects in visual form and/or characteristics values of the welding defects corresponding to the display information.