CN-121980615-A - Sensitive information detection method, electronic device, storage medium, and program product

CN121980615ACN 121980615 ACN121980615 ACN 121980615ACN-121980615-A

Abstract

The embodiment of the invention provides a sensitive information detection method, electronic equipment, a storage medium and a program product, and relates to the technical field of financial science and technology; the method comprises the steps of detecting a target service picture based on detection object description information to obtain a main object detection result, inquiring a service information base to determine auxiliary object description information of a main object, detecting the target service picture based on the auxiliary object description information to obtain an auxiliary object detection result, constructing an object relation matrix based on the main object detection result and the auxiliary object detection result, and identifying sensitive information in the target service picture based on the object relation matrix. The invention can improve the accuracy and reliability of sensitive information detection in complex service scenes.

Inventors

KANG XIAOHUI
WANG ZHENGXIANG

Assignees

中国工商银行股份有限公司

Dates

Publication Date: 20260505
Application Date: 20260202

Claims (11)

1. A method for detecting sensitive information, comprising: Acquiring detection object description information and a target service picture; detecting the target business picture based on the detection object description information to obtain a main object detection result; querying a business information base to determine auxiliary object description information of the main object; Detecting the target business picture based on the auxiliary object description information to obtain an auxiliary object detection result; constructing an object relation matrix based on the main object detection result and the auxiliary object detection result; And identifying the sensitive information in the target business picture based on the object relation matrix.
2. The method of claim 1, wherein obtaining the target business picture comprises: acquiring a service video file, and analyzing the service video file to obtain candidate service pictures; and calculating the structural similarity value between the candidate business pictures adjacent in time sequence, screening the candidate business pictures by using the structural similarity value, and determining the target business picture.
3. The method of claim 1, wherein constructing an object relationship matrix based on the primary object detection result and the secondary object detection result comprises: Constructing an object group according to the main object detection result and the auxiliary object detection result, wherein the object group comprises the relationship between each object and the relationship between each object and other objects; the object relationship matrix is constructed based on the set of objects.
4. The method of claim 3, wherein the object detection results comprise at least one of a category label, a bounding box position, and a bounding box size.
5. The method of claim 4, wherein identifying sensitive information in the target business picture based on the object-relationship matrix comprises: converting the object relation matrix into an object relation vector, and inputting the object relation vector into a first sensitive information identification model to obtain the confidence that the main object region contains sensitive information; And when the confidence coefficient is higher than a first confidence coefficient threshold value, determining that the main object region in the target business picture contains sensitive information.
6. The method of claim 5, further comprising, after inputting the object relation vector into a first sensitive information identification model to obtain a confidence that the primary object region contains sensitive information: When the confidence coefficient is not lower than a second confidence coefficient threshold value and is not higher than the first confidence coefficient threshold value, performing region clipping on the target business picture according to the main object detection result so as to obtain the main object region; wherein the first confidence threshold is greater than the second confidence threshold; Querying a service information base to determine auxiliary action description information of the main object; inputting the main object region and the auxiliary action description information into a similarity calculation model, so that the similarity calculation model outputs semantic similarity values of the main object region and the auxiliary action description information; And when the semantic similarity value is higher than a first semantic similarity threshold value, determining that a main object area in the target business picture contains sensitive information.
7. The method of claim 6, further comprising, after inputting the main object region and the auxiliary action description information into a similarity calculation model such that the similarity calculation model outputs semantic similarity values of the main object region and the auxiliary action description information: When the semantic similarity value is not lower than a second semantic similarity threshold value and is not higher than the first semantic similarity threshold value, inputting the main object region, the detection object description information and auxiliary discrimination prompt words of the main object into a second sensitive information identification model so as to determine whether the main object region in the target business picture contains sensitive information or not; wherein the first semantic similarity threshold is greater than the second semantic similarity threshold.
8. The method of claim 2, further comprising, after identifying sensitive information in the target business picture based on the object-relationship matrix: When the main object area in the target service picture is determined to contain sensitive information, inquiring the detection object description information in a preset sensitive information matching table to obtain a sensitive grade corresponding to the detection object description information; And desensitizing the business video file according to the sensitivity level.
9. An electronic device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, wherein the processor implements the sensitive information detection method according to any one of claims 1 to 8 when the program is executed by the processor.
10. A computer-readable storage medium, on which a computer program is stored, characterized in that the program, when being executed by a processor, implements the sensitive information detection method according to any one of claims 1 to 8.
11. A computer program product, characterized in that the computer program product comprises a computer program which, when executed by a processor, implements the sensitive information detection method according to any one of claims 1 to 8.

Description

Sensitive information detection method, electronic device, storage medium, and program product Technical Field The embodiment of the invention relates to the technical field of financial science and technology, in particular to a sensitive information detection method, electronic equipment, a storage medium and a program product. Background Along with the wide application of digital and video monitoring of financial business, a large amount of images or videos related to sensitive contents such as customer identity information, bank card numbers, bill certificates, transaction interfaces and the like can be generated in scenes such as banking outlets, counters, business halls and the like. On the premise of ensuring service traceability and compliance, effective detection and protection of the sensitive information become an important problem in the fields of financial information safety and privacy protection. The existing sensitive information detection method generally depends on a target detection model of a fixed category or a predefined label, and when a new sensitive object appears in a service scene or a detection target needs to be expanded, the existing scheme is difficult to directly expand, so that the detection range is limited. Meanwhile, the detection methods detect through only single sensitive information, and under complex service scenes, the sensitive information is easy to miss judgment or misjudge, so that the recognition accuracy and the adaptability are limited. Therefore, a method for detecting the sensitive information in the picture or the video in the complex service scene and ensuring the accuracy and the reliability of the sensitive information detection is needed. Disclosure of Invention The embodiment of the invention provides a sensitive information detection method, electronic equipment, a storage medium and a program product, which can detect sensitive information in a complex service scene and ensure the accuracy and reliability of the sensitive information detection. In a first aspect, a method for detecting sensitive information provided by an embodiment of the present invention includes: Acquiring detection object description information and a target service picture; detecting the target business picture based on the detection object description information to obtain a main object detection result; querying a business information base to determine auxiliary object description information of the main object; detecting the target business picture based on the auxiliary object description information to obtain an auxiliary object detection result; constructing an object relation matrix based on the main object detection result and the auxiliary object detection result; and identifying the sensitive information in the target business picture based on the object relation matrix. In a second aspect, an embodiment of the present invention provides a sensitive information detection apparatus, including: The acquisition module is used for acquiring the description information of the detection object and the target service picture; the first detection module is used for detecting the target business picture based on the detection object description information to obtain a main object detection result; The query module is used for querying the business information base to determine auxiliary object description information of the main object; The second detection module is used for detecting the target business picture based on the auxiliary object description information to obtain an auxiliary object detection result; The matrix construction module is used for constructing an object relation matrix based on the main object detection result and the auxiliary object detection result; and the identification module is used for identifying the sensitive information in the target business picture based on the object relation matrix. In a third aspect, an electronic device provided in an embodiment of the present invention includes a memory, a processor, and a computer program stored in the memory and executable on the processor, where the processor implements a method for detecting sensitive information according to any embodiment of the present invention when the processor executes the program. In a fourth aspect, embodiments of the present invention provide a computer readable storage medium having stored thereon a computer program which, when executed by a processor, implements a sensitive information detection method according to any of the embodiments of the present invention. In a fifth aspect, embodiments of the present invention provide a computer program product comprising a computer program which, when executed by a processor, implements a sensitive information detection method according to any of the embodiments of the present invention. According to the embodiment of the invention, the main object detection result is obtained by acquiring the detection object description inf