Search

CN-115205201-B - Archive exception detection method, archive exception detection device, archive exception detection processing equipment, archive exception detection storage medium and archive exception detection chip

CN115205201BCN 115205201 BCN115205201 BCN 115205201BCN-115205201-B

Abstract

The application relates to a file anomaly detection method, a device, a processing device, a storage medium and a chip, wherein the method comprises the steps of obtaining at least one characteristic image of a portrait file, wherein the characteristic image comprises a face image and/or a human body image, the similarity between the face image and the centroid of the portrait file is greater than a preset similarity threshold value; determining a first similarity according to the similarity between the at least one characteristic image and the centroid respectively, and/or determining a second similarity according to the similarity between the at least one characteristic image; and determining the anomaly probability of the portrait file according to the time interval from the last file anomaly detection, the first similarity and/or the second similarity. The archive anomaly detection method provided by the application can judge whether the portrait archive is abnormal without extracting all the information in the portrait archive, thereby improving the archive anomaly detection efficiency and saving a large amount of detection time.

Inventors

  • BU YUTIAN
  • CHEN LILI
  • ZHOU MINGWEI

Assignees

  • 浙江大华技术股份有限公司

Dates

Publication Date
20260508
Application Date
20220518

Claims (10)

  1. 1. A method for detecting archive anomalies, the method comprising: Acquiring at least one characteristic image of a portrait file, wherein the characteristic image comprises a face image and/or a human body image, and the similarity between the face image and the mass center of the portrait file is greater than a preset similarity threshold value; Determining a first similarity according to the similarity between the at least one characteristic image and the centroid respectively, and/or determining a second similarity according to the similarity between the at least one characteristic image; Determining the anomaly probability of the portrait file according to the time interval from the last file anomaly detection, the first similarity and/or the second similarity; the determining the anomaly probability of the portrait file according to the time interval from the last file anomaly detection, the first similarity and/or the second similarity comprises the following steps: determining the anomaly probability of the portrait file according to the positive correlation between the time interval from the last file anomaly detection and the anomaly probability of the portrait file and the negative correlation between the first similarity and/or the second similarity and the anomaly probability of the portrait file; the determining the anomaly probability of the portrait file according to the positive correlation between the time interval from the last file anomaly detection and the anomaly probability of the portrait file and the negative correlation between the first similarity and/or the second similarity and the anomaly probability of the portrait file comprises the following steps: determining a weighted value of the first similarity and the second similarity; And determining the anomaly probability of the portrait file according to the ratio of the weighted value to the time interval of the last file anomaly detection.
  2. 2. The method according to claim 1, wherein said determining a first similarity from the similarity between the at least one feature image and the centroid, respectively, and/or determining a second similarity from the similarity between the at least one feature image comprises: respectively determining a plurality of first candidate similarities between the at least one feature image and the centroid, and determining the first candidate similarity meeting a first preset requirement as a first similarity, wherein the first preset requirement comprises the smallest similarity in the plurality of first candidate similarities; And/or respectively determining second candidate similarities among the at least one characteristic image, and determining the second candidate similarities meeting a second preset requirement as second similarities, wherein the second preset requirement comprises the smallest similarity among the plurality of second candidate similarities.
  3. 3. The method of claim 1, further comprising, after determining the probability of anomaly of the portrait archive based on the time interval from the last archive anomaly detection, the first similarity, and/or the second similarity: and re-archiving the portrait files with the abnormal probability larger than the preset abnormal probability threshold.
  4. 4. A method according to claim 3, wherein the re-archiving the portrait files having the anomaly probabilities greater than a preset anomaly probability threshold comprises: determining at least one target portrait file with the abnormality probability larger than a preset abnormality probability threshold; respectively determining the confidence coefficient of the at least one target portrait file, wherein the confidence coefficient is used for representing the association degree between the human body image in the target portrait file and the human face image in the portrait file; and preferentially re-archiving the target portrait files with the confidence coefficient smaller than a preset confidence coefficient threshold value.
  5. 5. The method of claim 1, wherein determining the anomaly probability for the portrait archive based on the time interval from the last archive anomaly detection, the first similarity, and/or the second similarity comprises: Acquiring historical abnormal probability of the portrait file; And determining the anomaly probability of the portrait file according to the time interval from the last file anomaly detection, the first similarity and/or the second similarity and the historical anomaly probability.
  6. 6. The method according to claim 1, wherein said determining a first similarity from the similarity between the at least one feature image and the centroid, respectively, and/or determining a second similarity from the similarity between the at least one feature image comprises: Acquiring at least one face feature image and at least one body feature image in the at least one feature image; Determining a first similarity of the face according to the similarity between the at least one face feature image and the mass center of the face, and/or determining a second similarity of the face according to the similarity between the at least one face feature image; determining a first human body similarity according to the similarity between the at least one human body characteristic image and the mass center of the human body, and/or determining a second human body similarity according to the similarity between the at least one human body characteristic image; According to a preset rule, the smaller one of the first similarity of the human face and the first similarity of the human body is selected to be determined as the first similarity, and the smaller one of the second similarity of the human face and the second similarity of the human body is selected to be determined as the second similarity.
  7. 7. An archive exception detection device, the device comprising: The characteristic image acquisition module is used for acquiring at least one characteristic image of the portrait file, wherein the characteristic image comprises a face image and/or a human body image, and the similarity between the face image and the mass center of the portrait file is greater than a preset similarity threshold; a similarity determining module, configured to determine a first similarity according to a similarity between the at least one feature image and the centroid, and/or determine a second similarity according to a similarity between the at least one feature image; The anomaly probability determining module is used for determining the anomaly probability of the portrait file according to the time interval, the first similarity and/or the second similarity from the last file anomaly detection; The anomaly probability determining module is further configured to determine an anomaly probability of the portrait file according to a positive correlation between a time interval from a last file anomaly detection and an anomaly probability of the portrait file and a negative correlation between the first similarity and/or the second similarity and the anomaly probability of the portrait file; And the anomaly probability determining module is also used for determining the weighted value of the first similarity and the second similarity, and determining the anomaly probability of the portrait file according to the ratio of the weighted value to the time interval from the last file anomaly detection.
  8. 8. A processing device comprising a memory and a processor, the memory storing a computer program, characterized in that the processor implements the steps of the archive exception detection method of any one of claims 1 to 6 when executing the computer program.
  9. 9. A computer readable storage medium having stored thereon computer program instructions, which when executed by a processor, implement the steps of the archive anomaly detection method of any one of claims 1 to 6.
  10. 10. A chip comprising at least one processor for executing computer program instructions stored in a memory to perform the steps of the archive exception detection method of any one of claims 1 to 6.

Description

Archive exception detection method, archive exception detection device, archive exception detection processing equipment, archive exception detection storage medium and archive exception detection chip Technical Field The present application relates to the field of computer technologies, and in particular, to a method, an apparatus, a processing device, a storage medium, and a chip for detecting file abnormality. Background With the development of mobile internet technology and the continuous innovation of face recognition technology, various monitoring devices are widely applied to the security field to form a portrait monitoring system. The human image monitoring system based on the face recognition technology can cluster human images to form a human image file. In the actual human image clustering process, due to the influence of severe weather, foreign matter shielding, equipment damage, algorithm defects and other factors, the finally obtained human image file may have wrong files, such as the images of the Lifour in the Zhang three human image file. However, the prior art lacks a mechanism for detecting a person image file error file. Therefore, there is a need in the related art for an efficient file anomaly detection method. Disclosure of Invention Based on the above, the present application provides a method, an apparatus, a processing device, a storage medium and a chip for detecting file abnormality, so as to at least solve the problem that the related art lacks a mechanism for detecting a person file error file, thereby improving the efficiency of file abnormality detection. In a first aspect, an embodiment of the present application provides a method for detecting archive exception, where the method includes: Acquiring at least one characteristic image of a portrait file, wherein the characteristic image comprises a face image and/or a human body image, and the similarity between the face image and the mass center of the portrait file is greater than a preset similarity threshold value; Determining a first similarity according to the similarity between the at least one characteristic image and the centroid respectively, and/or determining a second similarity according to the similarity between the at least one characteristic image; and determining the anomaly probability of the portrait file according to the time interval from the last file anomaly detection, the first similarity and/or the second similarity. According to the file abnormality detection method provided by the embodiment of the application, in the process of abnormality detection of the human image file, the abnormality probability of the human image file can be determined by utilizing the similarity between the mass center of the human image file and the characteristic image and the time interval from the last file abnormality detection. The feature image may be a face image and/or a human body image with a similarity greater than a preset similarity threshold with the centroid of the portrait file, so that the feature image may represent the most feature information of the portrait file, and thus a more accurate reference may be provided for subsequent similarity calculation, thereby improving the accuracy of file anomaly detection. In addition, the archive anomaly detection method provided by the application can judge whether the portrait archive is anomalous without extracting all the information in the portrait archive, thereby improving the archive anomaly detection efficiency and saving a large amount of detection time. Optionally, in an embodiment of the present application, the determining the first similarity according to the similarity between the at least one feature image and the centroid, and/or determining the second similarity according to the similarity between the at least one feature image includes: respectively determining a plurality of first candidate similarities between the at least one feature image and the centroid, and determining the first candidate similarity meeting a first preset requirement as a first similarity, wherein the first preset requirement comprises the smallest similarity in the plurality of first candidate similarities; And/or respectively determining second candidate similarity between the at least one feature image, and determining the second candidate similarity smaller than a second preset threshold value as the second similarity, wherein the second preset requirement comprises the minimum similarity in the plurality of second candidate similarity. Optionally, in an embodiment of the present application, the determining the anomaly probability of the portrait file according to the time interval from the last archive anomaly detection, the first similarity and/or the second similarity includes: And determining the anomaly probability of the portrait file according to the positive correlation relationship between the time interval from the last file anomaly detection and the anomaly p