CN-116069626-B - Validity test method and system of data generator and electronic equipment

CN116069626BCN 116069626 BCN116069626 BCN 116069626BCN-116069626-B

Abstract

The invention provides a method, a system and electronic equipment for testing effectiveness of a data generator, wherein the method comprises the steps of obtaining test targets, test backgrounds and a plurality of matching rules for a plurality of tests, obtaining fourth time-space characteristic data based on physics about the matching combination, forming a first unmatched combination with the targets not conforming to the matching rules and a second unmatched combination with the backgrounds not conforming to the matching rules, respectively obtaining the corresponding first time-space characteristic data and the second time-space characteristic data, obtaining third time-space characteristic data through the data generator, comparing the third time-space characteristic data with the fourth time-space characteristic data, obtaining sub-scores of the effectiveness of the data generator, obtaining a plurality of sub-scores based on the matching rules, and forming a total score of the effectiveness of the data generator. To assist in evaluating whether the data generator is capable of being applied to a desired data augmentation scene.

Inventors

GAN HAIYONG
LIU WENDE
FENG GUOJIN
HE YINGWEI

Assignees

中国计量科学研究院

Dates

Publication Date: 20260512
Application Date: 20221215

Claims (10)

1. A method of testing the validity of a data generator, wherein the data generator is capable of synthesizing virtual third spatiotemporal feature data based on physical first spatiotemporal feature data and second spatiotemporal feature data, wherein the first spatiotemporal feature data comprises a first object and a first background, the second spatiotemporal feature data comprises a second object and a second background, and the third spatiotemporal feature data comprises the first object and the second background, the method comprising: Acquiring a plurality of test targets, test backgrounds and a plurality of matching rules for testing; Based on the matching rule, placing the test target in the matched test background to form a matching combination, and acquiring fourth time-space characteristic data based on physics about the matching combination; based on the matching rules, the test targets and the test backgrounds are respectively replaced with other test targets and other test backgrounds to form a first unmatched combination of which the targets do not accord with the matching rules and a second unmatched combination of which the backgrounds do not accord with the matching rules; Based on the first mismatch combination and the second mismatch combination, respectively acquiring corresponding first time-space characteristic data and second time-space characteristic data; Acquiring, by the data generator, the third spatiotemporal feature data based on the first spatiotemporal feature data and the second spatiotemporal feature data; Comparing the third time-space characteristic data with the fourth time-space characteristic data to obtain a sub-score of the effectiveness of the data generator; Acquiring a plurality of sub-scores based on a plurality of the matching rules; Based on the plurality of sub-scores, a total score for the effectiveness of the data generator is formed.
2. The method of claim 1, wherein the target and/or the background is composed of a plurality of objects.
3. The method of claim 1, wherein the test context in the matched set, the first unmatched set, and the second unmatched set comprises the test object in a set in space-time coordinates.
4. The method of claim 1, wherein obtaining a plurality of sub-scores based on a plurality of the matching rules comprises: For different matching rules, when the corresponding third space-time characteristic data and the corresponding fourth space-time characteristic data have the same similarity degree through comparison, the obtained sub-scores may not be the same.
5. The method of validity testing of a data generator of claim 4, said method further comprising: Classifying the test targets and the test backgrounds, and respectively giving different scores according to different categories; and obtaining a sub-score of the effectiveness of the data generator based on the scores given by the test background and the test target in the matching rule and the result of comparing the third time-space characteristic data and the fourth time-space characteristic data.
6. The method according to claim 5, wherein, for the matching rule, the more the part of the features contained in the test background that are identical to the features contained in the test target, the smaller the score corresponding to the matching rule, and conversely, the fewer the part of the features contained in the test background that are identical to the features contained in the test target, the larger the score corresponding to the matching rule.
7. The method of claim 5, wherein forming an overall score for the effectiveness of the data generator based on the plurality of sub-scores comprises: Obtaining a plurality of sub-scores under each category for the test targets or the test backgrounds of different categories, and forming a total score for the effectiveness of the data generator under each category based on the plurality of sub-scores under each category; And/or the number of the groups of groups, And obtaining a plurality of sub-scores under each category for the test targets or the test backgrounds of different categories, and forming a total score for the effectiveness of the data generator under all categories based on the plurality of sub-scores under each category.
8. A system for testing the validity of a data generator, said data generator capable of synthesizing virtual third spatiotemporal feature data based on physical first spatiotemporal feature data comprising a first object and a first background and second spatiotemporal feature data comprising a second object and a second background, said third spatiotemporal feature data comprising said first object and second background, said system comprising: The acquisition module is used for acquiring a plurality of test targets, test backgrounds and a plurality of matching rules for testing; The real data acquisition module is used for placing the test target in the matched test background to form a matching combination based on the matching rule, and acquiring fourth time-space characteristic data based on physics about the matching combination; The cross combination module is used for respectively replacing the test target and the test background with other test targets and other test backgrounds based on the matching rules to form a first unmatched combination of which the targets do not accord with the matching rules and a second unmatched combination of which the backgrounds do not accord with the matching rules; the input data acquisition module is used for respectively acquiring the corresponding first time-space characteristic data and the corresponding second time-space characteristic data based on the first mismatch combination and the second mismatch combination; The data generation module is used for acquiring the third space-time characteristic data through the data generator based on the first space-time characteristic data and the second space-time characteristic data; A comparison module for comparing the third space-time characteristic data with the fourth space-time characteristic data to obtain a sub-score of the effectiveness of the data generator; The circulation module is used for acquiring a plurality of sub scores based on a plurality of matching rules; And a scoring module for forming an overall score for the effectiveness of the data generator based on the plurality of sub-scores.
9. An electronic device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, characterized in that the processor implements the steps of the method for testing the validity of a data generator according to any one of claims 1-7 when the program is executed by the processor.
10. A non-transitory computer readable storage medium, having stored thereon a computer program, characterized in that the computer program, when executed by a processor, implements the steps of the method of validity testing of a data generator according to any one of claims 1-7.

Description

Validity test method and system of data generator and electronic equipment Technical Field The invention relates to the technical field of virtual synthesis effectiveness evaluation, in particular to a effectiveness test method, a system and electronic equipment of a data generator. Background Based on the real data, the generation of new virtual data by means of virtual synthesis has wide application. For example, a new drawing is virtually composed based on a real drawing image, and for example, a new background target image is virtually composed based on a real target image. The task specific direction of virtual composition may not be consistent for different demand scenarios, but all are to meet the preset demand, and to the extent such meets the demand, the application is called effectiveness. Only to some extent, the virtually synthesized new data can be applied to a specific task. How to evaluate the effectiveness of the generated results of the aforementioned means of virtual synthesis is a problem to be studied. The information disclosed in this background section is only for enhancement of understanding of the general background of the invention and should not be taken as an acknowledgement or any form of suggestion that this information forms the prior art already known to a person skilled in the art. Disclosure of Invention Aiming at the problems existing in the prior art, the invention provides a method, a system and electronic equipment for testing the effectiveness of a data generator. The invention provides a validity testing method of a data generator, wherein the data generator can synthesize virtual third space-time characteristic data based on physical first space-time characteristic data and second space-time characteristic data, the first space-time characteristic data comprises a first target and a first background, the second space-time characteristic data comprises a second target and a second background, and the third space-time characteristic data comprises the first target and the second background, and the method comprises the following steps: Acquiring a plurality of test targets, test backgrounds and a plurality of matching rules for testing; Based on the matching rule, placing the test target in the matched test background to form a matching combination, and acquiring fourth time-space characteristic data based on physics about the matching combination; based on the matching rules, the test targets and the test backgrounds are respectively replaced with other test targets and other test backgrounds to form a first unmatched combination of which the targets do not accord with the matching rules and a second unmatched combination of which the backgrounds do not accord with the matching rules; Based on the first mismatch combination and the second mismatch combination, respectively acquiring corresponding first time-space characteristic data and second time-space characteristic data; Acquiring, by the data generator, the third spatiotemporal feature data based on the first spatiotemporal feature data and the second spatiotemporal feature data; Comparing the third time-space characteristic data with the fourth time-space characteristic data to obtain a sub-score of the effectiveness of the data generator; Acquiring a plurality of sub-scores based on a plurality of the matching rules; Based on the plurality of sub-scores, a total score for the effectiveness of the data generator is formed. According to the validity testing method of the data generator provided by the invention, the target and/or the background are formed by combining a plurality of objects. According to the validity testing method of the data generator, the test backgrounds in the matching combination, the first unmatched combination and the second unmatched combination comprise the test targets in the combination on space-time coordinates. According to the validity test method of the data generator provided by the invention, a plurality of sub scores are obtained based on a plurality of matching rules, and the method comprises the following steps: For different matching rules, when the corresponding third space-time characteristic data and the corresponding fourth space-time characteristic data have the same similarity degree through comparison, the obtained sub-scores may not be the same. According to the validity test method of the data generator provided by the invention, the method further comprises the following steps: Classifying the test targets and the test backgrounds, and respectively giving different scores according to different categories; and obtaining a sub-score of the effectiveness of the data generator based on the scores given by the test background and the test target in the matching rule and the result of comparing the third time-space characteristic data and the fourth time-space characteristic data. According to the validity testing method of the data generator provided by the invention, f