Detection method, detection device and storage medium

1. A detection method, comprising:

obtaining a to-be-detected image;

obtaining a plurality of confidence levels corresponding to a plurality of detection items, which correspond to a plurality types of defects in the to-be-detected image; and

determining a detection result of the to-be-detected image according to the plurality of confidence levels and a plurality of detection thresholds, wherein the plurality of detection thresholds are corresponding to the plurality of detection items and are a non-inferior solution of objective functions of an overkill rate and a missed detection rate, the overkill rate is the ratio of qualified ones detected as defective, and the missed detection rate is the ratio of defective ones detected as qualified,

wherein before the determining a detection result of the to-be-detected image according to the plurality of confidence levels and a plurality of detection thresholds, the detection method further comprises:

obtaining first test data and second test data, wherein the first test data is qualified data, and the second test data is unqualified data; and

obtaining the plurality of detection thresholds according to the first test data and the second test data,

wherein the obtaining the plurality of detection thresholds according to the first test data and the second test data comprises:

obtaining a plurality of prediction confidence levels and prediction results according to the first test data and the second test data, the plurality of prediction confidence levels being corresponding to the plurality of detection items; and

obtaining the plurality of detection thresholds according to the plurality of prediction confidence levels and the prediction results, and

wherein the obtaining the plurality of detection thresholds according to the plurality of prediction confidence levels and the prediction results comprises:

determining a primary population of a genetic algorithm model according to the plurality of prediction confidence levels and the prediction result;

obtaining a non-inferior solution of the objective functions of the overkill rate and the missed detection rate by performing iteration according to the primary population based on the objective functions of the overkill rate and the missed detection rate; and

determining the plurality of detection thresholds according to the non-inferior solution.

2. The detection method according to claim 1 , wherein independent variables of the objective functions of the overkill rate and the missed detection rate are corresponding to the plurality of detection items, and dependent variables of the objective functions of the overkill rate and the missed detection rate are the overkill rate and the missed detection rate.

3. The detection method according to claim 2 , wherein the determining the plurality of detection thresholds according to the non-inferior solution comprises:

determining a set of detection thresholds according to the non-inferior solution, the set of detection thresholds including multiple groups of the non-inferior solutions; and

determining the plurality of detection thresholds according to the non-inferior solution corresponding to a minimum missed detection rate in the set of detection thresholds.

4. The detection method according to claim 3 , wherein the non-inferior solution is a solution when the objective functions of the overkill rate and the missed detection rate are in the Pareto optimal state.

5. The detection method according to claim 1 , wherein the obtaining a non-inferior solution of the objective functions of the overkill rate and the missed detection rate by performing iteration according to the primary population based on the objective functions of the overkill rate and the missed detection rate comprises:

obtaining a non-inferior solution of the objective functions of the overkill rate and the missed detection rate by performing iteration on the primary population through an elitist non-dominated sorting genetic algorithm.

6. The detection method according to claim 1 , wherein the first test data is overkill data, and the overkill data is the data in which qualified ones are detected as defective in the first test data.

7. The detection method according to claim 1 , wherein the quantity of the first test data is the same as the quantity of the second test data.

8. The detection method according to claim 1 , wherein the determining a detection result of the to-be-detected image according to the plurality of confidence levels and a plurality of detection thresholds comprises:

determining the detection result as unqualified in a case where the plurality of confidence levels are greater than at least one of the detection thresholds; or

determining the detection result as qualified in a case where the plurality of confidence levels are less than or equal to each of the detection thresholds.

9. A detection device, comprising circuitry configured to:

obtain a to-be-detected image;

obtain a plurality of confidence levels corresponding to a plurality of detection items, which correspond to a plurality types of defects in the to-be-detected image; and

determine a detection result of the to-be-detected image according to the plurality of confidence levels and the plurality of detection thresholds, wherein the plurality of detection thresholds are corresponding to the plurality of detection items and are a non-inferior solution of objective functions of an overkill rate and a missed detection rate, the overkill rate is the ratio of qualified ones detected as defective, and the missed detection rate is the ratio of defective ones detected as qualified,

wherein, before determining the detection result of the to-be-detected image according to the plurality of confidence levels and the plurality of detection thresholds, the circuitry is further configured to:

obtain first test data and second test data, and obtain a plurality of detection thresholds according to the first test data and the second test data, wherein the first test data is qualified data, and the second test data is unqualified data:

wherein, to obtain the plurality of detection thresholds according to the first test data and the second test data, the circuitry is configured to:

obtain a plurality of prediction confidence levels and prediction results according to the first test data and the second test data, the plurality of prediction confidence levels being corresponding to the plurality of detection items; and

obtain the plurality of detection thresholds according to the plurality of prediction confidence levels and the prediction results,

wherein, to obtain the plurality of detection thresholds according to the plurality of prediction confidence levels and the prediction results, the circuitry is configured to:

determine a primary population of a genetic algorithm model according to the plurality of prediction confidence levels and the prediction result;

obtain a non-inferior solution of the objective functions of the overkill rate and the missed detection rate by performing iteration according to the primary population based on the objective functions of the overkill rate and the missed detection rate; and

determine the plurality of detection thresholds according to the non-inferior solution.

10. The detection device according to claim 9 , wherein independent variables of the objective functions of the overkill rate and the missed detection rate are corresponding to the plurality of detection items, and dependent variables of the objective functions of the overkill rate and the missed detection rate are the overkill rate and the missed detection rate.

11. The detection device according to claim 10 , wherein the circuitry is configured to:

determine a set of detection thresholds according to the non-inferior solution, the set of detection thresholds including multiple groups of the non-inferior solutions; and

determine the plurality of detection thresholds according to the non-inferior solution corresponding to a minimum missed detection rate in the set of detection thresholds.

12. The detection device according to claim 11 , wherein the non-inferior solution is a solution when the objective functions of the overkill rate and the missed detection rate are in the Pareto optimal state.

13. The detection device according to claim 9 , wherein the circuitry is configured to: obtain a non-inferior solution of the objective functions of the overkill rate and the missed detection rate by performing iteration on the primary population through an elitist non-dominated sorting genetic algorithm.

14. The detection device according to claim 9 , wherein the first test data is overkill data, and the overkill data is the data in which qualified ones are detected as defective in the first test data.

15. The detection device according to claim 9 , wherein the quantity of the first test data is the same as the quantity of the second test data.

16. The detection device according to claim 9 , wherein the circuitry is configured to:

determine the detection result as unqualified in a case where the plurality of confidence levels are greater than at least one of the detection thresholds; or

determining the detection result as qualified in a case where the plurality of confidence levels are less than or equal to each of the detection thresholds.

17. A detection device, comprising:

a memory, configured to store computer executable instructions; and

a processor, configured to access the memory and execute the computer executable instructions so as to perform operations in the method according to claim 1 .

18. A non-transitory storage medium configured to store a computer program, wherein when the computer program is executed by a computing device, the computing device is caused to implement the method according to claim 1 .