Intelligent inspection management system based on face recognition technology

Abstract

The invention discloses an intelligent inspection management system based on a face recognition technology, relates to the technical field of intelligent management, and solves the technical problem that in the prior art, the situation that an inspector possibly replaces a post in the inspection process is not considered, and the inspector is difficult to manage. The method comprises the following steps: acquiring facial feature information of inspection personnel through a face recognition technology, and constructing an identity verification library of the inspection personnel; acquiring inspection time, and scheduling inspection personnel according to the inspection time; managing attendance information of the inspection personnel according to the scheduling information; the identity of the inspector is verified before the inspector performs inspection work; the rating of the inspection personnel is adjusted regularly; acquiring an abnormal image uploaded by the inspection personnel, and analyzing according to the abnormal image to obtain an abnormal reason; and the abnormity is processed according to the abnormity reason, and the processing result is fed back to the management personnel, so that the identity of the inspection personnel can be verified, and the condition of post replacement or post lack is avoided.

Description

Technical Field

[0001] The present invention belongs to the field of intelligent management and relates to intelligent inspection management technology, in particular to an intelligent inspection management system based on face recognition technology. Background Art

[0002] Intelligent inspection management refers to the use of preset inspection routes and checkpoints, combined with GPS positioning, RFID identification, Internet of Things, big data and other technologies, to conduct automated inspections, detection and analysis of equipment or facilities, so as to achieve comprehensive and accurate management of inspection work; facial recognition technology can be prepared to identify the identity of inspection personnel, avoid illegal acts such as impersonation and substitution, thereby improving the safety of inspections; the intelligent inspection management system can automatically collect data generated during the inspection process, including the inspection personnel's location information, inspection time and inspection sequence, etc., to facilitate management personnel to conduct subsequent analysis and processing, thereby improving management efficiency; the system can generate inspection reports based on the inspection situation, and can provide management personnel with a comprehensive work overview, which helps managers understand the overall situation of the inspection work and make more scientific and reasonable decisions.

[0003] The prior art (patent document application with publication number CN115631547A) discloses an intelligent fire inspection management system and method, which includes digitally representing various collected data and integrating various data of the inspection object during the fire inspection after processing to obtain the inspection coefficient to evaluate its fire status as a whole; by integrating the inspection weight of the inspection object with the inspection results of different fires, reliable data support can be provided for the differentiated analysis of the fire status of the inspection object, and by real-time data collection of different inspection objects in the area and corresponding data analysis, it is judged whether the fire status of different types of inspection objects is abnormal and the degree of abnormality, and by conducting targeted tracing and adjustment of the abnormalities found through comparative analysis, the overall fire safety inspection effect in the area can be improved; the prior art improves the inspection effect of fire safety by conducting targeted tracing and adjustment of the abnormalities; however, the prior art improves the inspection effect by conducting targeted analysis of the abnormalities, but does not consider that during the inspection process, the inspection personnel may be replaced, making it difficult to manage the inspection personnel.

[0004] The present invention provides an intelligent inspection management system based on face recognition technology to solve the above technical problems. Summary of the invention

[0005] The present invention aims to solve at least one of the technical problems existing in the prior art; to this end, the present invention proposes an intelligent patrol management system based on face recognition technology, which is used to solve the technical problem that the prior art improves the patrol effect by performing targeted analysis of anomalies, but does not take into account the situation that patrol personnel may be replaced during the patrol process, making it difficult to manage the patrol personnel.

[0006] To achieve the above-mentioned purpose, the first aspect of the present invention provides an intelligent inspection management system based on face recognition technology, comprising: a management verification module, and a face collection module and an exception handling module connected thereto;

[0007] Face collection module: collects the facial feature information of patrol personnel through face recognition technology, and builds the identity authentication library of patrol personnel based on the facial feature information;

[0008] Management verification module: obtain inspection time, schedule inspection personnel according to inspection time; manage inspection personnel's attendance information according to scheduling information; verify the identity of inspection personnel through face recognition technology before inspection; regularly adjust the rating of inspection personnel;

[0009] Abnormal processing module: obtain abnormal images uploaded by inspection personnel, analyze the abnormal images to obtain the abnormal causes; handle the abnormalities according to the abnormal causes, and feed back the processing results to the management personnel.

[0010] Preferably, scheduling inspection personnel according to inspection time includes:

[0011] Retrieve the inspection time, divide the inspection time into several time periods evenly, and obtain the number of inspection shifts; determine the number of inspection personnel in the corresponding inspection time period according to the number of inspection shifts;

[0012] Obtain the ratings of the inspectors, and classify the inspectors according to their ratings; select inspectors from each type of inspectors to form an inspection team based on the number of inspectors; and schedule the inspection team according to the inspection time period.

[0013] The present invention divides the inspection personnel into inspection groups according to the inspection time and the rating of the inspection personnel, and arranges the inspection groups according to the inspection time periods, which can ensure the quality of the inspection groups and reasonably arrange the shifts of the inspection personnel, which is conducive to ensuring the normal progress of the inspection work.

[0014] Preferably, the management of the inspection personnel's attendance information according to the shift information includes:

[0015] Retrieve the shift information, and test the punch-in records of the corresponding patrol personnel according to the shift information to obtain the test results;

[0016] Determine whether there are any inspectors who have not clocked in; if yes, analyze the reasons for not clocking in; if no, continue to check the clocking in records of the inspectors.

[0017] The present invention manages the attendance information of patrol inspectors according to the shift information, can monitor the work of patrol inspectors, and avoids accidents caused by patrol inspectors failing to arrive at their posts in time.

[0018] Preferably, the analysis of the reasons for not clocking in includes:

[0019] Retrieve the patrol personnel information of the personnel who have not clocked in to determine whether the patrol personnel are on duty; if yes, change the corresponding patrol personnel's clock-in record to clocked in; if no, analyze the patrol personnel's leave information;

[0020] Determine whether the patrol personnel has asked for leave. If yes, change the punch-in record of the corresponding patrol personnel to asking for leave; if not, change the punch-in record of the corresponding patrol personnel to absent; integrate the punched-in, leave and absence into the attendance information of the corresponding patrol personnel.

[0021] The present invention analyzes the reasons why the patrol personnel fail to punch in. When the patrol personnel forget to punch in or ask for leave, the punching record is adjusted, which can avoid the situation where the patrol personnel fail to punch in due to other reasons. When the patrol personnel do not ask for leave and do not punch in, the punching record is changed to absence. This is conducive to detecting the working status of the patrol personnel and adjusting the patrol personnel in time.

[0022] Preferably, the identity verification of the inspection personnel by face recognition technology includes:

[0023] Obtain the clock-in image of the patrol personnel and extract the facial feature information in the clock-in image; match the clock-in image with the image in the identity verification library according to the facial feature information to obtain a matching result; wherein the matching result includes a matching success or a matching failure;

[0024] Determine whether the matching result is a successful match; if yes, determine that the identity authentication of the patrol personnel has passed; if no, determine that the identity authentication of the patrol personnel has failed, generate an alarm signal, and notify the management personnel; integrate the identity authentication passed and the identity authentication failed into the verification information.

[0025] The present invention analyzes the inspection personnel's punch-in image, extracts facial feature information of the inspection personnel's punch-in image, and matches the facial feature information with the image in the identity verification library. When the match fails, the inspection personnel's identity verification fails, which can avoid the inspection personnel from taking turns on duty and is conducive to improving the management of the inspection personnel.

[0026] Preferably, the regular adjustment of the rating of the inspectors includes:

[0027] Regularly obtain the attendance information and verification information of patrol personnel; mark the number of leave requests and absences in the attendance information as JC and QC respectively; mark the number of identity verification failures as YC;

[0028] The adjustment coefficient of the corresponding inspection personnel is calculated by the formula TX=α×(JC+QC)+β×exp(YC); the rating of the inspection personnel is adjusted according to the adjustment coefficient; wherein α and β are proportional coefficients greater than 0.

[0029] The present invention calculates the inspection personnel's adjustment coefficient according to the inspection personnel's attendance information and verification information, can comprehensively consider the inspection personnel's attendance information and verification information, and provides a data basis for subsequent adjustment of the inspection personnel's rating.

[0030] Preferably, the step of adjusting the rating of the inspection personnel according to the adjustment coefficient includes:

[0031] Retrieve the adjustment coefficient and corresponding rating of the corresponding patrol personnel; determine whether the patrol personnel's adjustment coefficient is greater than the first-level threshold. If yes, adjust the patrol personnel's rating down by one level; if no, compare the adjustment coefficient with the second-level threshold;

[0032] Determine whether the adjustment coefficient is less than the secondary threshold. If yes, the rating of the patrol inspector is adjusted up by one level; if no, the rating of the corresponding patrol inspector is not adjusted. The primary threshold and the secondary threshold are set based on experience, and the primary threshold is greater than the secondary threshold.

[0033] It should be noted that the inspection personnel's rating is divided into first, second and third grades, among which first grade is the highest and third grade is the lowest. When the inspection personnel is in the third grade and needs to be adjusted down by one level, the inspection personnel's grade will not be adjusted, and the inspection personnel will be punished accordingly; when the inspection personnel is in the first grade and needs to be adjusted up by one level, the inspection personnel's grade will not be adjusted, and the inspection personnel will be rewarded accordingly.

[0034] The present invention compares the inspection personnel's adjustment coefficient with the first-level threshold. When the adjustment coefficient is greater than the first-level threshold, the inspection personnel's rating is adjusted downward by one level; when the adjustment coefficient is less than the second-level threshold, the inspection personnel's rating is adjusted upward by one level. The inspection personnel's rating can be adjusted according to the inspection personnel's staged performance, which is conducive to stimulating the inspection personnel's work enthusiasm.

[0035] Preferably, obtaining the abnormal cause according to the abnormal image analysis includes:

[0036] Retrieving an abnormal image, and preprocessing the abnormal image to obtain an input image; wherein the preprocessing includes format conversion, grayscale conversion, and denoising;

[0037] Call the anomaly analysis model, input the input image into the anomaly analysis model, and obtain the anomaly label; match the corresponding anomaly cause based on the anomaly label; wherein the anomaly analysis model is built based on the artificial intelligence model; the anomaly label is set to a positive integer.

[0038] The present invention obtains an input image after preprocessing the abnormal image, analyzes the input image through a trained abnormal analysis model, obtains an abnormal label, and matches the corresponding abnormal cause, so as to analyze the abnormalities occurring during the inspection process and handle the abnormal situation in time.

[0039] Preferably, the abnormal analysis model is constructed based on an artificial intelligence model, including:

[0040] Acquire an abnormal data set; wherein the abnormal data set includes standard input data consistent with the content attributes of the input image, and standard output data consistent with the content attributes of the abnormal label;

[0041] The abnormal data set is divided into a training set and a validation set, the training set is used to train the artificial intelligence model, and the validation set is used to evaluate the parameters of the trained artificial intelligence model; when the parameters of the artificial intelligence model are greater than the parameter threshold, the artificial intelligence model is marked as an abnormal analysis model; otherwise, the artificial intelligence model is retrained; wherein the artificial intelligence model includes a convolutional neural network model or a BP neural network model.

[0042] Preferably, the processing of the exception according to the cause of the exception and feeding back the processing result to the management personnel include:

[0043] Retrieve the exception cause and obtain the exception handling library; match the exception cause with the exception handling library, obtain the corresponding treatment measures for the exception cause, and handle the exception according to the treatment measures;

[0044] After a set time period, the abnormal points are tested and the test results are fed back to the management personnel; the test results include normal or abnormal.

[0045] The present invention handles abnormal situations according to the causes of the abnormal situations, and after a set time period, detects the points where the abnormalities occur, feeds back the detection results to the management personnel, and can evaluate the abnormal handling measures, which is conducive to ensuring that the abnormal situations are resolved.

[0046] Compared with the prior art, the present invention has the following beneficial effects:

[0047] 1. The present invention divides the inspection personnel into inspection groups according to the inspection time and the rating of the inspection personnel, and arranges the inspection groups according to the inspection time period, which can ensure the quality of the inspection groups and reasonably arrange the shifts of the inspection personnel, which is conducive to ensuring the normal progress of the inspection work; the inspection personnel's attendance information is managed according to the scheduling information, and the inspection personnel's work can be monitored to avoid accidents caused by the inspection personnel's failure to arrive at the post in time; the reasons for the inspection personnel's failure to punch in are analyzed, and when the inspection personnel forget to punch in or take leave, the punching record is adjusted, which can avoid the situation that the inspection personnel fail to punch in due to other reasons. When the inspection personnel do not take leave and fail to punch in, the punching record is changed to absence; it is conducive to detecting the working status of the inspection personnel and adjusting the inspection personnel in time.

[0048] 2. The present invention collects facial feature information of patrol inspectors to build an identity authentication library; analyzes the clock-in images of patrol inspectors, extracts facial feature information of the clock-in images of patrol inspectors, and matches the facial feature information with the images in the identity authentication library. When the match fails, the identity authentication of the patrol inspector fails, which can avoid the situation where patrol inspectors have to take turns to replace each other, and is conducive to improving the management of patrol inspectors; the adjustment coefficient of the patrol inspectors is calculated according to the attendance information and verification information of the patrol inspectors, which can comprehensively consider the attendance information and verification information of the patrol inspectors, and provide a data basis for the subsequent adjustment of the rating of the patrol inspectors; the adjustment coefficient of the patrol inspectors is compared with the first-level threshold value, and when the adjustment coefficient is greater than the first-level threshold value, the rating of the patrol inspectors is adjusted downward by one level; when the adjustment coefficient is less than the second-level threshold value, the rating of the patrol inspectors is adjusted upward by one level, and the rating of the patrol inspectors can be adjusted according to the periodic performance of the patrol inspectors, which is conducive to stimulating the enthusiasm of the patrol inspectors to work.

[0049] 3. The present invention obtains an input image after preprocessing the abnormal image, analyzes the input image through a trained abnormal analysis model, obtains an abnormal label, and matches the corresponding abnormal cause. It can analyze the abnormalities occurring during the inspection process and can handle the abnormal situation in time; handle the abnormal situation according to the abnormal cause, and after a set time period, detect the abnormal point, and feed back the detection result to the management personnel, so as to evaluate the abnormal handling measures, which is conducive to ensuring that the abnormal situation is resolved. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0051] Figure 1 It is a schematic diagram of the specific steps of the present invention;

[0052] Figure 2 A schematic diagram of the inspection personnel attendance information management steps of the present invention;

[0053] Figure 3 A schematic diagram of the steps of identity verification and rating adjustment of patrol personnel of the present invention;

[0054] Figure 4 Schematic diagram of the abnormality detection processing steps of the present invention. DETAILED DESCRIPTION

[0055] The technical solution of the present invention will be clearly and completely described below in conjunction with the embodiments. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0056] See also Figure 1 , the first aspect of the present invention provides an intelligent inspection management system based on face recognition technology, including: a management verification module, and a face collection module and an exception processing module connected thereto;

[0057] Face collection module: collects the facial feature information of patrol personnel through face recognition technology, and builds the identity authentication library of patrol personnel based on the facial feature information;

[0058] Management verification module: obtain inspection time, schedule inspection personnel according to inspection time; manage inspection personnel's attendance information according to scheduling information; verify the identity of inspection personnel through face recognition technology before inspection; regularly adjust the rating of inspection personnel;

[0059] Abnormal processing module: obtain abnormal images uploaded by inspection personnel, analyze the abnormal images to obtain the abnormal causes; handle the abnormalities according to the abnormal causes, and feed back the processing results to the management personnel.

[0060] See also Figure 2, obtain the inspection time, divide the inspection time into several time periods on average, and obtain the number of inspection shifts; determine the number of inspectors in the corresponding inspection time period according to the number of inspection shifts; obtain the ratings of the inspectors, and classify the inspectors according to their ratings; select inspectors from each type of inspectors to form an inspection team according to the number of inspectors; schedule the inspection team according to the inspection time period; check the punch-in records of the corresponding inspectors according to the scheduling information to obtain the test results; determine whether there are inspectors who have not punched in; if yes, analyze the reasons for not punching in; if not, continue to check the punch-in records of the inspectors.

[0061] For example: suppose that area A is inspected from 8:00 to 20:00; the inspection time is divided into three sections of 4 hours: 8:00-12:00; 12:00-16:00; and 16:00-20:00; the ratings of the inspectors are obtained, and one inspector is selected from each of the three ratings to form an inspection team; each inspection team consists of 3 people, and inspection team A will conduct inspections from 8:00 to 12:00, inspection team B will conduct inspections from 12:00 to 16:00, and inspection team C will conduct inspections from 16:00 to 20:00.

[0062] The clock-in records of the inspectors were checked at the corresponding time, and analysis showed that there was one person who did not clock in in both inspection team B and inspection team C.

[0063] Retrieve the patrol personnel information of the personnel who have not punched in, and determine whether the patrol personnel are on duty; if yes, change the punch-in record of the corresponding patrol personnel to punched in; if not, analyze the leave information of the patrol personnel; determine whether the patrol personnel have asked for leave, if yes, change the punch-in record of the corresponding patrol personnel to leave; if not, change the punch-in record of the corresponding patrol personnel to absence; integrate the punched in, leave and absence into the attendance information of the corresponding patrol personnel.

[0064] For example: suppose that the non-punching information of the inspectors in inspection group B and inspection group C is analyzed, and it is found that the inspectors in inspection group B are not on duty, but have leave records, and the punch-in records of the inspectors are marked as leave; the inspectors in inspection group C are not on duty and have no leave records, so the punch-in records of the inspectors are marked as absent.

[0065] See also Figure 3, collect the facial feature information of the patrol personnel through face recognition technology, build the identity authentication library of the patrol personnel according to the facial feature information; obtain the clock-in image of the patrol personnel, and extract the facial feature information in the clock-in image; match the clock-in image with the image in the identity authentication library according to the facial feature information to obtain a matching result; wherein the matching result includes a successful match or a failed match; judge whether the matching result is a successful match; if yes, it is determined that the identity authentication of the patrol personnel has passed; if not, it is determined that the identity authentication of the patrol personnel has failed, an alarm signal is generated, and the management personnel are notified; the identity authentication passed and the identity authentication failed are integrated into the verification information.

[0066] For example, assuming that there are three inspectors of inspection team A with punch-in images, the punch-in images are matched with the identity verification database. If the punch-in image of one inspector fails to match, the information of the inspector is sent to the management personnel.

[0067] Regularly obtain the inspection personnel's attendance information and verification information; mark the number of leave requests and absences in the attendance information as JC and QC respectively; mark the number of identity verification failures as YC; calculate the adjustment coefficient of the corresponding inspection personnel through the formula TX = α × (JC + QC) + β × exp (YC); adjust the inspection personnel's rating according to the adjustment coefficient; where α and β are proportional coefficients greater than 0;

[0068] Retrieve the adjustment coefficient and corresponding rating of the corresponding patrol personnel; determine whether the patrol personnel's adjustment coefficient is greater than the first-level threshold, if so, adjust the patrol personnel's rating downward by one level; if not, compare the adjustment coefficient with the second-level threshold; determine whether the adjustment coefficient is less than the second-level threshold, if so, adjust the patrol personnel's rating upward by one level; if not, do not adjust the corresponding patrol personnel's rating; wherein, the first-level threshold and the second-level threshold are set based on experience, and the first-level threshold is greater than the second-level threshold.

[0069] For example, suppose that the adjustment coefficients of three patrol inspectors are calculated, α = 1; β = 1; the specific steps are as follows:

[0070] Patrolman 1: Number of leave requests JC = 3, number of absences QC = 2; number of identity verification failures YC = 5; then the adjustment coefficient TX = 153.41; if it is greater than the first-level threshold of 10, the rating of the patrolman will be reduced from second-level to third-level;

[0071] Patrolman 2: Number of leave requests JC = 1, number of absences QC = 0; number of identity verification failures YC = 0; then the adjustment coefficient TX = 1; if it is less than the secondary threshold 3, the rating of the patrolman will be raised from the second grade to the first grade;

[0072] Patrolman 3: Number of leave requests JC = 2, number of absences QC = 3; number of identity verification failures YC = 2; then the adjustment coefficient TX = 7.38; then no adjustment will be made to the rating of this patrolman.

[0073] See also Figure 4 , obtain the abnormal image uploaded by the inspection personnel, and obtain the input image after preprocessing the abnormal image; wherein the preprocessing includes format conversion, grayscale, and denoising; call the abnormal analysis model, input the input image into the abnormal analysis model, and obtain the abnormal label; match the corresponding abnormal cause based on the abnormal label; wherein the abnormal analysis model is built based on the artificial intelligence model; the abnormal label is set to a positive integer;

[0074] Obtain an exception handling library; match the cause of the exception with the exception handling library, obtain the corresponding treatment measures for the cause of the exception, and handle the exception according to the treatment measures; after setting a time period, detect the points where the exception occurs, and feed back the detection results to the management personnel; the detection results include normal or abnormal.

[0075] It is worth noting that the anomaly analysis model is built based on artificial intelligence models, including:

[0076] Acquire an abnormal data set; wherein the abnormal data set includes standard input data consistent with the content attributes of the input image, and standard output data consistent with the content attributes of the abnormal label;

[0077] The abnormal data set is divided into a training set and a validation set, the training set is used to train the artificial intelligence model, and the validation set is used to evaluate the parameters of the trained artificial intelligence model; when the parameters of the artificial intelligence model are greater than the parameter threshold, the artificial intelligence model is marked as an abnormal analysis model; otherwise, the artificial intelligence model is retrained; wherein the artificial intelligence model includes a convolutional neural network model or a BP neural network model.

[0078] For example: there are two abnormal images collected, and the two abnormal images are pre-processed respectively to obtain two input images. The abnormal analysis model is called, and the input images are input into the abnormal analysis model to obtain abnormal labels, and the corresponding abnormal causes are matched to D and G respectively; the abnormal causes are matched with the abnormal processing library to obtain processing measures, and the corresponding processing measures are used to process abnormalities D and abnormality G respectively; after 2 hours, the processed abnormalities are detected, and the detection result of abnormality D is normal; the detection result of abnormality G is abnormal, and abnormality G is re-analyzed.

[0079] Part of the data in the above formula is calculated by removing the dimension and taking its numerical value. The formula is a formula closest to the actual situation obtained by software simulation of a large amount of collected data; the preset parameters and preset thresholds in the formula are set by technical personnel in this field according to actual conditions or obtained through simulation of a large amount of data.

[0080] The working principle of the present invention is as follows: the present invention collects facial feature information of patrol inspectors through face recognition technology, builds an identity authentication library of patrol inspectors according to the facial feature information; obtains patrol time, and schedules patrol inspectors according to the patrol time; manages the attendance information of patrol inspectors according to the scheduling information; before patrol inspectors conduct patrol work, they verify the identities of patrol inspectors through face recognition technology; regularly adjust the ratings of patrol inspectors; obtains abnormal images uploaded by patrol inspectors, and obtains the causes of abnormalities according to abnormal image analysis; processes abnormalities according to the causes of abnormalities, and feeds back the processing results to management personnel.

[0081] The above embodiments are only used to illustrate the technical method of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical method of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical method of the present invention.

Claims

1. Intelligent inspection management system based on face recognition technology, including: Management verification module, and the face collection module and exception handling module connected thereto; characterized in that: Face collection module: collects the facial feature information of patrol personnel through face recognition technology, and builds the identity authentication library of patrol personnel based on the facial feature information; Management verification module: obtain inspection time, schedule inspection personnel according to inspection time; manage inspection personnel's attendance information according to scheduling information; verify the identity of inspection personnel through face recognition technology before inspection; regularly adjust the rating of inspection personnel; Abnormal processing module: obtain abnormal images uploaded by inspection personnel, analyze the abnormal images to obtain the abnormal causes; handle the abnormalities according to the abnormal causes, and feed back the processing results to the management personnel.

2. The intelligent inspection management system based on face recognition technology according to claim 1 is characterized in that: The inspection personnel are scheduled according to the inspection time, including: Retrieve the inspection time, divide the inspection time into several time periods evenly, and obtain the number of inspection shifts; determine the number of inspection personnel in the corresponding inspection time period according to the number of inspection shifts; Obtain the ratings of the inspectors, and classify the inspectors according to their ratings; select inspectors from each type of inspectors to form an inspection team based on the number of inspectors; and schedule the inspection team according to the inspection time period.

3. The intelligent inspection management system based on face recognition technology according to claim 1 is characterized in that: The management of the inspection personnel's attendance information according to the shift information includes: Retrieve the shift information, and test the punch-in records of the corresponding patrol personnel according to the shift information to obtain the test results; Determine whether there are any inspectors who have not clocked in; if yes, analyze the reasons for not clocking in; if no, continue to check the clocking in records of the inspectors.

4. The intelligent inspection management system based on face recognition technology according to claim 3 is characterized in that: The reasons for not clocking in include: Retrieve the patrol personnel information of the personnel who have not clocked in to determine whether the patrol personnel are on duty; if yes, change the corresponding patrol personnel's clock-in record to clocked in; if no, analyze the patrol personnel's leave information; Determine whether the patrol personnel has asked for leave. If yes, change the punch-in record of the corresponding patrol personnel to asking for leave; if not, change the punch-in record of the corresponding patrol personnel to absent; integrate the punched-in, leave and absence into the attendance information of the corresponding patrol personnel.

5. The intelligent inspection management system based on face recognition technology according to claim 1 is characterized in that: The verification of the identity of the inspection personnel by face recognition technology includes: Obtain the clock-in image of the patrol personnel and extract the facial feature information in the clock-in image; match the clock-in image with the image in the identity verification library according to the facial feature information to obtain a matching result; wherein the matching result includes a matching success or a matching failure; Determine whether the matching result is a successful match; if yes, determine that the identity authentication of the patrol personnel has passed; if no, determine that the identity authentication of the patrol personnel has failed, generate an alarm signal, and notify the management personnel; integrate the identity authentication passed and the identity authentication failed into the verification information.

6. The intelligent inspection management system based on face recognition technology according to claim 1 is characterized in that: The regular adjustments to the ratings of inspectors include: Regularly obtain the attendance information and verification information of patrol personnel; mark the number of leave requests and absences in the attendance information as JC and QC respectively; mark the number of identity verification failures as YC; The adjustment coefficient of the corresponding inspection personnel is calculated by the formula TX=α×(JC+QC)+β×exp(YC); the rating of the inspection personnel is adjusted according to the adjustment coefficient; wherein α and β are proportional coefficients greater than 0.

7. The intelligent inspection management system based on face recognition technology according to claim 6 is characterized in that: The adjustment of the inspection personnel's rating according to the adjustment coefficient includes: Retrieve the adjustment coefficient and corresponding rating of the corresponding patrol personnel; determine whether the patrol personnel's adjustment coefficient is greater than the first-level threshold. If yes, adjust the patrol personnel's rating down by one level; if no, compare the adjustment coefficient with the second-level threshold; Determine whether the adjustment coefficient is less than the secondary threshold. If yes, the rating of the patrol inspector is adjusted up by one level; if no, the rating of the corresponding patrol inspector is not adjusted. The primary threshold and the secondary threshold are set based on experience, and the primary threshold is greater than the secondary threshold.

8. The intelligent inspection management system based on face recognition technology according to claim 1 is characterized in that: The abnormal cause obtained by analyzing the abnormal image includes: Retrieving an abnormal image, and preprocessing the abnormal image to obtain an input image; wherein the preprocessing includes format conversion, grayscale conversion, and denoising; Call the anomaly analysis model, input the input image into the anomaly analysis model, and obtain the anomaly label; match the corresponding anomaly cause based on the anomaly label; wherein the anomaly analysis model is built based on the artificial intelligence model; the anomaly label is set to a positive integer.

9. The intelligent inspection management system based on face recognition technology according to claim 8 is characterized in that: The abnormal analysis model is constructed based on an artificial intelligence model, including: Acquire an abnormal data set; wherein the abnormal data set includes standard input data consistent with the content attributes of the input image, and standard output data consistent with the content attributes of the abnormal label; The abnormal data set is divided into a training set and a validation set, the training set is used to train the artificial intelligence model, and the validation set is used to evaluate the parameters of the trained artificial intelligence model; when the parameters of the artificial intelligence model are greater than the parameter threshold, the artificial intelligence model is marked as an abnormal analysis model; otherwise, the artificial intelligence model is retrained; wherein the artificial intelligence model includes a convolutional neural network model or a BP neural network model.

10. The intelligent inspection management system based on face recognition technology according to claim 1 is characterized in that: The above-mentioned abnormality is handled according to the abnormality cause, and the processing result is fed back to the management personnel, including: Retrieve the exception cause and obtain the exception handling library; match the exception cause with the exception handling library, obtain the corresponding treatment measures for the exception cause, and handle the exception according to the treatment measures; After setting the time period, detect the abnormal points and feedback the test results to the management personnel; The test results include normal or abnormal.

Images