Image content recognition system and method based on artificial intelligence

Abstract

The invention discloses an image content recognition system and method based on artificial intelligence, and relates to the technical field of data analysis, and the method comprises the steps: carrying out the recognition through collecting the identity and image information of a person, recording a failure reason, calculating the occurrence frequency of the failure reason, and determining a failure level; historical identification failure records are collected, the occurrence frequency of abnormal data is calculated, and target data are determined; setting a training period, dividing a day into a plurality of characteristic time periods, calculating failure cause frequency in each time period, and determining a target time period; counting the number of identification records in a target time period of a certain day, calculating a failure frequency and determining an optimization period; and calculating the real-time failure frequency in the optimization period. If so, reminding a worker; if yes, the optimization period is adjusted, and therefore the utilization efficiency of system resources is maximized while the recognition accuracy is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of data analysis, and in particular to an image content recognition system and method based on artificial intelligence. Background Art

[0002] With the rapid development of artificial intelligence technology, image content recognition has become a core demand in many fields. As a cutting-edge research direction in image processing, virtual human image recognition is widely used in identity verification, short video creation, social platforms and other fields. In recent years, with the rapid development of generative adversarial networks and deep learning technologies, the quality of virtual human image generation has been greatly improved. This technological advancement has brought new opportunities for virtual human recognition and comparison, but also posed new technical challenges. The high fidelity and rich details of virtual human images have impacted traditional recognition methods, requiring more accurate and efficient image processing algorithm support. With the continuous advancement of image generation technology, the accuracy of virtual human recognition technology may be affected, so the recognition model needs to be optimized and updated regularly. However, too frequent optimization may lead to high consumption of computing resources, and too long optimization intervals may reduce the accuracy of identifying virtual humans. Therefore, it is necessary to explore appropriate optimization strategies to maximize the utilization efficiency of system resources while improving recognition accuracy. Summary of the invention

[0003] The purpose of the present invention is to provide an image content recognition system and method based on artificial intelligence to solve the problems raised in the prior art.

[0004] In order to solve the above technical problems, the present invention provides the following technical solutions: an image content recognition method based on artificial intelligence, the method comprising: Step S100: by collecting the identity information of the personnel, collecting the image information of the personnel, identifying the personnel, forming an image recognition system, collecting the failure reasons in the historical recognition failure records, calculating the occurrence frequency of a certain failure reason, and determining a failure reason set of a certain failure level; Step S200: obtaining a set of historical recognition failure records of a certain failure cause, collecting abnormal data in the historical recognition failure records, calculating the frequency of occurrence of the abnormal data, and determining target data of a certain failure cause; Step S300: setting a training cycle, dividing a day into several characteristic time periods, obtaining a set of recognition failure records in a certain characteristic time period, calculating the frequency of failure reasons in the characteristic time period, and determining a target time period for a certain failure reason; Step S400: collecting the number of all recognition records in a target time period of a certain failure reason on a certain day within the training cycle, calculating the failure frequency of a certain failure reason on a certain day, determining the optimization period of a certain failure reason, and calculating the optimization period of the image recognition system; Step S500: Calculate the real-time failure frequency of the failure reasons within the optimization period of the image recognition system. If the real-time failure frequency is lower than the failure frequency threshold, remind the staff. If the real-time failure frequency is higher than the failure frequency threshold, calculate the adjusted optimization period.

[0005] Furthermore, step S100 includes: Step S101: Acquire the identity information of the person, collect the feature data of the identity information in the information database, arrange image acquisition equipment on site, collect the image information of the person, perform preprocessing and feature extraction on the collected image data, and generate a recognition record at the same time, match the extracted feature data with the feature data of the identity information, if the match fails, set the feature data that failed to match as abnormal data, and mark the reason for the failure, so as to form an image recognition system; Step S102: Obtain a set of historical recognition failure records, collect the failure cause in a certain historical recognition failure record, summarize the historical recognition failure records with the same failure cause, collect the number of historical recognition failure records with a certain failure cause, and calculate the occurrence frequency of the failure cause according to the following formula: ; Where A represents the occurrence frequency of the failure cause, S represents the number of historical recognition failure records of the failure cause, and S' represents the total number of historical recognition failure records of all failure causes; Step S103: Summarize the occurrence frequencies of all failure reasons, divide the occurrence frequencies of failure reasons into several failure levels, collect the occurrence frequency range corresponding to a certain failure level, summarize the failure reasons that meet a certain failure level, and obtain a failure reason set of a certain failure level; In the image recognition system, the reasons for failure include image quality problems, occlusion problems, posture problems, scene interference, etc. Feature data includes facial feature points such as the position and proportion of eyes, nose, and mouth, skin texture features such as pore distribution and skin color information, and dynamic features such as blinking and mouth movement; Through feature extraction and matching mechanisms, the image recognition system can accurately capture the identity characteristics of people and conduct systematic analysis of failures, thereby effectively reducing the occurrence of misidentification and missed identification; The system records and marks the feature data of matching failures, classifies and performs frequency analysis on different failure causes, provides data support for subsequent optimization, and improves the comprehensiveness and accuracy of abnormal data detection.

[0006] Further, step S200 includes: Step S201: In a set of failure reasons at a certain failure level, a set of historical recognition failure records for a certain failure reason is obtained, abnormal data in a certain historical recognition failure record is collected, historical recognition failure records with the same abnormal data are summarized, and the number of historical recognition failure records is collected, and the frequency of abnormal data occurrence is calculated according to the following formula: ; Wherein, B represents the frequency of abnormal data occurrence, C represents the number of historical recognition failure records of abnormal data, and C' represents the number of historical recognition failure records of failure causes; Step S202: setting a frequency threshold of abnormal data occurrence, setting abnormal data whose frequency exceeds the frequency threshold as target data, and obtaining target data of a certain failure reason; Setting the frequency threshold and filtering out the target data can help the system automatically filter out high-frequency anomalies and reduce unnecessary data interference, thereby achieving targeted optimization and adjustment and improving the system's adaptive capabilities; Summarizing and classifying abnormal data helps the system filter out noise data, improve the utilization of historical data, ensure the high quality of subsequent training data, and provide more reliable data support for subsequent analysis.

[0007] Furthermore, step S300 includes: Step S301: select a number of consecutive days as a training cycle, divide a day into a number of characteristic time periods, and obtain the time range corresponding to the characteristic time period; Step S302: Obtain a set of recognition failure records within a training cycle, collect the recognition time point in a certain recognition failure record, obtain the characteristic time period corresponding to the recognition time point, and then obtain the characteristic time period of the recognition failure record, summarize the recognition failure records belonging to the same characteristic time period, and obtain a set of recognition failure records of a certain characteristic time period; Step S303: in the set of recognition failure records in a certain characteristic time period, the failure cause of a certain recognition failure record is obtained, the recognition failure records with the same failure cause are summarized, and abnormal data in a certain recognition failure record in the set of recognition failure records with a certain failure cause is extracted. If the abnormal data is the target data of the failure cause, the recognition failure record is marked as the target record; Step S304: Collect the target number of a certain failure reason marked as a target record in a certain characteristic time period, summarize the target number of a certain failure reason in a certain characteristic time period in the training cycle, and calculate the frequency of the failure reason in the characteristic time period according to the following formula: ; Where D represents the frequency of the failure cause in the characteristic time period, E represents the target number of failure causes in the characteristic time period, and E' represents the target total number of failure causes; Step S305: setting a frequency threshold of the failure cause, marking a characteristic time period in which the frequency of the failure cause exceeds the frequency threshold as a target time period, and obtaining a target time period for a certain failure cause; Dividing recognition failure records into characteristic time periods by time can accurately locate the high-incidence period of failures within a specific time period, reveal the potential correlation between time and failure causes, and provide a basis for formulating more efficient optimization strategies; By summarizing the set of recognition failure records within the characteristic time period and analyzing the failure frequency in a specific time period, the time distribution weight of the training data can be adjusted to improve the model's adaptability to different time characteristics. Time period-based failure analysis can further refine the system optimization direction, concentrate resources on high-frequency failure periods, avoid unnecessary full-time optimization, and reduce resource waste.

[0008] Furthermore, step S400 includes: Step S401: On a certain day in the training cycle, the number of all recognition records in the target time period of a certain failure reason is collected, and the failure frequency of a certain failure reason on a certain day is calculated according to the following formula: ; Among them, F ir It is expressed as the failure frequency of the target time period for the rth failure reason on the i-th day, H ir It is represented as the number of recognition failure records of the target time period of the rth failure reason in the i-th day, H ir ' represents the number of all identified records in the target time period for the rth failure reason on the i-th day; Step S402: sorting the failure frequencies of a certain failure cause within the training cycle in chronological order, setting a failure frequency threshold, marking the time when the failure frequency is lower than the failure frequency threshold as abnormal time, and obtaining an abnormal time set of a certain failure cause; Step S403: sorting the abnormal time set of a certain failure cause in chronological order, obtaining the bit sequence corresponding to each abnormal time, setting the abnormal time number threshold, and when the bit sequence corresponding to the abnormal time reaches the abnormal time number threshold, setting the abnormal time as the optimization period of the failure cause; Step S404: Summarize the optimization cycles of all failure reasons in a certain failure level, and calculate the optimization cycle of the image recognition system according to the following formula: ; Where T represents the optimization period of the image recognition system, t ae It is expressed as the optimization period of the e-th failure reason in the a-th failure level, f a Expressed as the total number of failure reasons in the ath failure level, α a The weight of the ath failure level is expressed as b, the total number of failure levels is expressed as b, and the optimization cycle of the image recognition system is obtained; Through dynamic calculation and sorting of failure frequency and abnormal time, the time period that needs to be optimized can be accurately identified to avoid global fixed optimization and improve the flexibility and efficiency of system adjustment; Limit the optimization cycle to periods of high failure frequency, and adjust the optimization strategies for different failure levels based on weight factors to avoid resource dispersion and ensure that optimization can maximize system performance.

[0009] Furthermore, step S500 includes: Step S501: within a target time period of a certain failure reason, a real-time recognition record set is obtained, the number of real-time recognition records within the target time period is collected, and the real-time failure reason of a certain real-time recognition failure record in the real-time recognition record set is extracted. If the real-time failure reason is the failure reason, the real-time recognition failure record is marked, the number of marked real-time recognition failure records within the target time period is collected, and the real-time failure frequency of the failure reason within the optimization period of the image recognition system is calculated. If the real-time failure frequency is lower than the failure frequency threshold, the staff is reminded to optimize the image recognition system. If the real-time failure frequency is higher than the failure frequency threshold, step S502 is executed; Step S502: Summarize the real-time failure frequencies of all failure reasons in a certain failure level, and calculate the adjusted optimization period according to the following formula: ; Among them, T' represents the adjusted optimization period, R ae It is expressed as the real-time failure frequency of the e-th failure reason in the a-th failure level. The real-time failure frequency of the failure reason is calculated according to the calculated adjusted optimization period; Based on the analysis and summary of real-time failure frequency, the optimization cycle is adjusted dynamically to ensure that the optimization plan can respond to new failure modes in a timely manner and improve the flexibility of the optimization strategy; Introducing the adjusted optimization cycle into the system's optimization plan realizes the evolution from static rules to dynamic adjustments, enabling the system to adapt to a changing operating environment. Using real-time failure data to refine the planning of the optimization cycle helps identify key issues in the failure level, ensuring that the optimization goals are more focused and reducing ineffective adjustments.

[0010] In order to better implement the above method, an image content recognition system based on artificial intelligence is also proposed, which includes an image recognition system module, a target data module, a target time period module, an optimization cycle module of the image recognition system, and a real-time adjustment module; Image recognition system module: collects personnel identity information, collects personnel image information, identifies personnel, forms an image recognition system, collects failure reasons in historical recognition failure records, calculates the frequency of occurrence of a certain failure reason, and determines the failure reason set of a certain failure level; Target data module: obtains a set of historical recognition failure records of a certain failure cause, collects abnormal data in the historical recognition failure records, calculates the frequency of abnormal data occurrence, and determines the target data of a certain failure cause; Target time period module: Set the training cycle, divide a day into several characteristic time periods, obtain the set of recognition failure records in a certain characteristic time period, calculate the frequency of failure reasons in the characteristic time period, and determine the target time period for a certain failure reason; Optimization cycle module of image recognition system: collects the number of all recognition records in the target time period of a certain failure reason on a certain day within the training cycle, calculates the failure frequency of a certain failure reason on a certain day, determines the optimization cycle of a certain failure reason, and calculates the optimization cycle of the image recognition system; Real-time adjustment module: calculates the real-time failure frequency of the failure reasons within the optimization period of the image recognition system. If the real-time failure frequency is lower than the failure frequency threshold, the staff is reminded. If the real-time failure frequency is higher than the failure frequency threshold, the adjusted optimization period is calculated.

[0011] Furthermore, the image recognition system module includes an image recognition system building unit and a failure level determination unit: Establishing an image recognition system unit: obtaining the identity information of the personnel, collecting the feature data of the identity information in the information database, collecting the image information of the personnel by arranging image acquisition equipment on site, and preprocessing and extracting features of the collected image data, and generating an identification record at the same time, matching the extracted feature data with the feature data of the identity information, and if the match fails, setting the feature data that failed to match as abnormal data, and marking the reason for the failure, to form an image recognition system; Determine the failure level unit: obtain a set of historical recognition failure records, collect the failure reasons in a historical recognition failure record, summarize the historical recognition failure records with the same failure reason, collect the number of historical recognition failure records with a certain failure reason, calculate the occurrence frequency of the failure reason, summarize the occurrence frequency of all failure reasons, divide the occurrence frequency of the failure reason into several failure levels, collect the occurrence frequency range corresponding to a certain failure level, summarize the failure reasons that meet a certain failure level, and obtain a failure cause set for a certain failure level.

[0012] Furthermore, the target time period module includes a target recording unit and a target time period determination unit: Determine the target record unit: select several consecutive days as the training cycle, divide the time of one day into several characteristic time periods, obtain the time range corresponding to the characteristic time period, obtain the set of recognition failure records within the training cycle, collect the recognition time point in a certain recognition failure record, obtain the characteristic time period corresponding to the recognition time point, and then obtain the characteristic time period of the recognition failure record, summarize the recognition failure records belonging to the same characteristic time period, obtain the recognition failure record set of a certain characteristic time period, obtain the failure cause of a certain recognition failure record in the recognition failure record set of a certain characteristic time period, summarize the recognition failure records with the same failure cause, and extract abnormal data from a certain recognition failure record in the recognition failure record set of a certain failure cause, if the abnormal data is the target data of the failure cause, mark the recognition failure record as the target record; Determine the target time period unit: collect the target number of a certain failure cause marked as a target record in a certain feature time period, summarize the target number of a certain failure cause in a certain feature time period within the training cycle, calculate the frequency of the failure cause in the feature time period, set the frequency threshold of the failure cause, mark the feature time period in which the frequency of the failure cause exceeds the frequency threshold as the target time period, and obtain the target time period for a certain failure cause.

[0013] Furthermore, the optimization cycle module of the image recognition system includes an abnormal time determination unit and an optimization cycle calculation unit of the image recognition system: Determine the abnormal time unit: on a certain day in the training cycle, collect the number of all identification records in the target time period of a certain failure reason, calculate the failure frequency of a certain failure reason on a certain day, sort the failure frequency of a certain failure reason in the training cycle in chronological order, set a failure frequency threshold, and mark the time when the failure frequency is lower than the failure frequency threshold as abnormal time; Calculate the optimization cycle unit of the image recognition system: sort the abnormal time set of a certain failure cause in chronological order, obtain the bit sequence corresponding to each abnormal time, set the abnormal time number threshold, and when the bit sequence corresponding to the abnormal time reaches the abnormal time number threshold, set the abnormal time as the optimization cycle of the failure cause, summarize the optimization cycles of all failure causes in a certain failure level, calculate the optimization cycle of the image recognition system, and obtain the optimization cycle of the image recognition system.

[0014] Compared with the prior art, the present invention has the following beneficial effects: Failure causes are managed in different levels according to frequency, and the target time period and real-time failure frequency are monitored to intelligently adjust the optimization cycle, effectively optimize resource allocation, and improve system operation efficiency. By setting characteristic time periods and training cycles, the failure frequency in each time period is accurately calculated, and high-frequency failure time periods are identified, thereby optimizing the system's recognition performance in different time periods. By combining real-time data with historical data and calculating the adjusted optimization cycle, the system can dynamically adjust the optimization strategy according to the latest failure mode to achieve a balance between system stability and flexibility. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] Figure 1 A schematic diagram of a flow chart of an image content recognition method based on artificial intelligence of the present invention; Figure 2 The figure is a schematic diagram of the structure of an image content recognition system based on artificial intelligence of the present invention. DETAILED DESCRIPTION

[0016] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not 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.

[0017] See also Figure 1 and Figure 2 The present invention provides a technical solution: an image content recognition method based on artificial intelligence, the method comprising: Step S100: by collecting the identity information of the personnel, collecting the image information of the personnel, identifying the personnel, forming an image recognition system, collecting the failure reasons in the historical recognition failure records, calculating the occurrence frequency of a certain failure reason, and determining a failure reason set of a certain failure level; Wherein, step S100 includes: Step S101: Acquire the identity information of the person, collect the feature data of the identity information in the information database, arrange image acquisition equipment on site, collect the image information of the person, perform preprocessing and feature extraction on the collected image data, and generate a recognition record at the same time, match the extracted feature data with the feature data of the identity information, if the match fails, set the feature data that failed to match as abnormal data, and mark the reason for the failure, so as to form an image recognition system; Step S102: Obtain a set of historical recognition failure records, collect the failure cause in a certain historical recognition failure record, summarize the historical recognition failure records with the same failure cause, collect the number of historical recognition failure records with a certain failure cause, and calculate the occurrence frequency of the failure cause according to the following formula: ; Where A represents the occurrence frequency of the failure cause, S represents the number of historical recognition failure records of the failure cause, and S' represents the total number of historical recognition failure records of all failure causes; Step S103: Summarize the occurrence frequencies of all failure causes, divide the occurrence frequencies of failure causes into several failure levels, collect the occurrence frequency range corresponding to a certain failure level, summarize the failure causes that meet a certain failure level, and obtain a failure cause set of a certain failure level.

[0018] Step S200: obtaining a set of historical recognition failure records of a certain failure cause, collecting abnormal data in the historical recognition failure records, calculating the frequency of occurrence of the abnormal data, and determining target data of a certain failure cause; Wherein, step S200 includes: Step S201: In a set of failure reasons at a certain failure level, a set of historical recognition failure records for a certain failure reason is obtained, abnormal data in a certain historical recognition failure record is collected, historical recognition failure records with the same abnormal data are summarized, and the number of historical recognition failure records is collected, and the frequency of abnormal data occurrence is calculated according to the following formula: ; Wherein, B represents the frequency of abnormal data occurrence, C represents the number of historical recognition failure records of abnormal data, and C' represents the number of historical recognition failure records of failure causes; Step S202: setting a frequency threshold of abnormal data occurrence, setting abnormal data whose frequency exceeds the frequency threshold as target data, and obtaining target data of a certain failure reason; For example, there are a total of 500 historical recognition failure records for the first failure cause, and the number of historical recognition failures for the first abnormal data is 150. The frequency of abnormal data occurrence is 30%. The frequency threshold of abnormal data occurrence is set to 20%, and the first abnormal data is the target data for the first failure cause.

[0019] Step S300: setting a training cycle, dividing a day into several characteristic time periods, obtaining a set of recognition failure records in a certain characteristic time period, calculating the frequency of failure reasons in the characteristic time period, and determining a target time period for a certain failure reason; Wherein, step S300 includes: Step S301: select a number of consecutive days as a training cycle, divide a day into a number of characteristic time periods, and obtain the time range corresponding to the characteristic time period; Step S302: Obtain a set of recognition failure records within a training cycle, collect the recognition time point in a certain recognition failure record, obtain the characteristic time period corresponding to the recognition time point, and then obtain the characteristic time period of the recognition failure record, summarize the recognition failure records belonging to the same characteristic time period, and obtain a set of recognition failure records of a certain characteristic time period; Step S303: in the set of recognition failure records in a certain characteristic time period, the failure cause of a certain recognition failure record is obtained, the recognition failure records with the same failure cause are summarized, and abnormal data in a certain recognition failure record in the set of recognition failure records with a certain failure cause is extracted. If the abnormal data is the target data of the failure cause, the recognition failure record is marked as the target record; Step S304: Collect the target number of a certain failure reason marked as a target record in a certain characteristic time period, summarize the target number of a certain failure reason in a certain characteristic time period in the training cycle, and calculate the frequency of the failure reason in the characteristic time period according to the following formula: ; Where D represents the frequency of the failure cause in the characteristic time period, E represents the target number of failure causes in the characteristic time period, and E' represents the target total number of failure causes; Step S305: setting a frequency threshold of the failure cause, marking a characteristic time period in which the frequency of the failure cause exceeds the frequency threshold as a target time period, and obtaining a target time period for a certain failure cause.

[0020] Step S400: collecting the number of all recognition records in a target time period of a certain failure reason on a certain day within the training cycle, calculating the failure frequency of a certain failure reason on a certain day, determining the optimization period of a certain failure reason, and calculating the optimization period of the image recognition system; Wherein, step S400 includes: Step S401: On a certain day in the training cycle, the number of all recognition records in the target time period of a certain failure reason is collected, and the failure frequency of a certain failure reason on a certain day is calculated according to the following formula: ; Among them, F ir It is expressed as the failure frequency of the target time period for the rth failure reason on the i-th day, H ir It is represented as the number of recognition failure records of the target time period of the rth failure reason in the i-th day, H ir ' represents the number of all identified records in the target time period for the rth failure reason on the i-th day; Step S402: sorting the failure frequencies of a certain failure cause within the training cycle in chronological order, setting a failure frequency threshold, marking the time when the failure frequency is lower than the failure frequency threshold as abnormal time, and obtaining an abnormal time set of a certain failure cause; Step S403: sorting the abnormal time set of a certain failure cause in chronological order, obtaining the bit sequence corresponding to each abnormal time, setting the abnormal time number threshold, and when the bit sequence corresponding to the abnormal time reaches the abnormal time number threshold, setting the abnormal time as the optimization period of the failure cause; Step S404: Summarize the optimization cycles of all failure reasons in a certain failure level, and calculate the optimization cycle of the image recognition system according to the following formula: ; Where T represents the optimization period of the image recognition system, t ae It is expressed as the optimization period of the e-th failure reason in the a-th failure level, f a Expressed as the total number of failure reasons in the ath failure level, α a The weight of the ath failure level is expressed as b, the total number of failure levels is expressed as b, and the optimization cycle of the image recognition system is obtained; For example, the optimization period for failure reason 1 in the first failure level is 3 days, and the optimization period for failure reason 2 is 5 days; the optimization period for failure reason 3 in the second failure level is 7 days. The calculated optimization period for the image recognition system is 4.9 days.

[0021] Step S500: Calculate the real-time failure frequency of the failure reasons within the optimization period of the image recognition system. If the real-time failure frequency is lower than the failure frequency threshold, remind the staff. If the real-time failure frequency is higher than the failure frequency threshold, calculate the adjusted optimization period. Wherein, step S500 includes: Step S501: within a target time period of a certain failure reason, a real-time recognition record set is obtained, the number of real-time recognition records within the target time period is collected, and the real-time failure reason of a certain real-time recognition failure record in the real-time recognition record set is extracted. If the real-time failure reason is the failure reason, the real-time recognition failure record is marked, the number of marked real-time recognition failure records within the target time period is collected, and the real-time failure frequency of the failure reason within the optimization period of the image recognition system is calculated. If the real-time failure frequency is lower than the failure frequency threshold, the staff is reminded to optimize the image recognition system. If the real-time failure frequency is higher than the failure frequency threshold, step S502 is executed; Step S502: Summarize the real-time failure frequencies of all failure reasons in a certain failure level, and calculate the adjusted optimization period according to the following formula: ; Among them, T' represents the adjusted optimization period, R ae It is expressed as the real-time failure frequency of the e-th failure cause in the a-th failure level. The real-time failure frequency of the failure cause is calculated according to the calculated adjusted optimization period.

[0022] In order to better implement the above method, an image content recognition system based on artificial intelligence is also proposed, which includes an image recognition system module, a target data module, a target time period module, an optimization cycle module of the image recognition system, and a real-time adjustment module; Image recognition system module: collects personnel identity information, collects personnel image information, identifies personnel, forms an image recognition system, collects failure reasons in historical recognition failure records, calculates the frequency of occurrence of a certain failure reason, and determines the failure reason set of a certain failure level; The image recognition system module includes an image recognition system unit and a failure level determination unit: Establishing an image recognition system unit: obtaining the identity information of the personnel, collecting the feature data of the identity information in the information database, collecting the image information of the personnel by arranging image acquisition equipment on site, and preprocessing and extracting features of the collected image data, and generating an identification record at the same time, matching the extracted feature data with the feature data of the identity information, and if the match fails, setting the feature data that failed to match as abnormal data, and marking the reason for the failure, to form an image recognition system; Determine the failure level unit: obtain a set of historical recognition failure records, collect the failure reasons in a historical recognition failure record, summarize the historical recognition failure records with the same failure reason, collect the number of historical recognition failure records with a certain failure reason, calculate the occurrence frequency of the failure reason, summarize the occurrence frequency of all failure reasons, divide the occurrence frequency of the failure reason into several failure levels, collect the occurrence frequency range corresponding to a certain failure level, summarize the failure reasons that meet a certain failure level, and obtain a failure cause set for a certain failure level.

[0023] Target data module: obtains a set of historical recognition failure records of a certain failure cause, collects abnormal data in the historical recognition failure records, calculates the frequency of abnormal data occurrence, and determines the target data of a certain failure cause; Target time period module: Set the training cycle, divide a day into several characteristic time periods, obtain the set of recognition failure records in a certain characteristic time period, calculate the frequency of failure reasons in the characteristic time period, and determine the target time period for a certain failure reason; The target time period module includes a target recording unit and a target time period determination unit: Determine the target record unit: select several consecutive days as the training cycle, divide the time of one day into several characteristic time periods, obtain the time range corresponding to the characteristic time period, obtain the set of recognition failure records within the training cycle, collect the recognition time point in a certain recognition failure record, obtain the characteristic time period corresponding to the recognition time point, and then obtain the characteristic time period of the recognition failure record, summarize the recognition failure records belonging to the same characteristic time period, obtain the recognition failure record set of a certain characteristic time period, obtain the failure cause of a certain recognition failure record in the recognition failure record set of a certain characteristic time period, summarize the recognition failure records with the same failure cause, and extract abnormal data from a certain recognition failure record in the recognition failure record set of a certain failure cause, if the abnormal data is the target data of the failure cause, mark the recognition failure record as the target record; Determine the target time period unit: collect the target number of a certain failure cause marked as a target record in a certain feature time period, summarize the target number of a certain failure cause in a certain feature time period within the training cycle, calculate the frequency of the failure cause in the feature time period, set the frequency threshold of the failure cause, mark the feature time period in which the frequency of the failure cause exceeds the frequency threshold as the target time period, and obtain the target time period for a certain failure cause.

[0024] Optimization cycle module of image recognition system: collects the number of all recognition records in the target time period of a certain failure reason on a certain day within the training cycle, calculates the failure frequency of a certain failure reason on a certain day, determines the optimization cycle of a certain failure reason, and calculates the optimization cycle of the image recognition system; The optimization cycle module of the image recognition system includes an abnormal time determination unit and an optimization cycle calculation unit of the image recognition system: Determine the abnormal time unit: on a certain day in the training cycle, collect the number of all identification records in the target time period of a certain failure reason, calculate the failure frequency of a certain failure reason on a certain day, sort the failure frequency of a certain failure reason in the training cycle in chronological order, set a failure frequency threshold, and mark the time when the failure frequency is lower than the failure frequency threshold as abnormal time; Calculate the optimization cycle unit of the image recognition system: sort the abnormal time set of a certain failure cause in chronological order, obtain the bit sequence corresponding to each abnormal time, set the abnormal time number threshold, and when the bit sequence corresponding to the abnormal time reaches the abnormal time number threshold, set the abnormal time as the optimization cycle of the failure cause, summarize the optimization cycles of all failure causes in a certain failure level, calculate the optimization cycle of the image recognition system, and obtain the optimization cycle of the image recognition system.

[0025] Real-time adjustment module: calculates the real-time failure frequency of the failure reasons within the optimization period of the image recognition system. If the real-time failure frequency is lower than the failure frequency threshold, the staff is reminded. If the real-time failure frequency is higher than the failure frequency threshold, the adjusted optimization period is calculated.

[0026] It will be apparent to those skilled in the art that the invention is not limited to the details of the exemplary embodiments described above and that the invention can be implemented in other specific forms without departing from the spirit or essential features of the invention. Therefore, the embodiments should be considered exemplary and non-limiting in all respects, and the scope of the invention is defined by the appended claims rather than the foregoing description, and it is intended that all variations falling within the meaning and scope of the equivalent elements of the claims be included in the invention. Any reference numeral in a claim should not be considered as limiting the claim to which it relates.

Claims

1. An image content recognition method based on artificial intelligence, characterized in that: Methods include: Step S100: by collecting the identity information of the personnel, collecting the image information of the personnel, identifying the personnel, forming an image recognition system, collecting the failure reasons in the historical recognition failure records, calculating the occurrence frequency of a certain failure reason, and determining a failure reason set of a certain failure level; Step S200: obtaining a set of historical recognition failure records of a certain failure cause, collecting abnormal data in the historical recognition failure records, calculating the frequency of occurrence of the abnormal data, and determining target data of a certain failure cause; Step S300: setting a training cycle, dividing a day into several characteristic time periods, obtaining a set of recognition failure records in a certain characteristic time period, calculating the frequency of failure reasons in the characteristic time period, and determining a target time period for a certain failure reason; Step S400: collecting the number of all recognition records in a target time period of a certain failure reason on a certain day within the training cycle, calculating the failure frequency of a certain failure reason on a certain day, determining the optimization period of a certain failure reason, and calculating the optimization period of the image recognition system; Step S500: Calculate the real-time failure frequency of the failure reasons within the optimization period of the image recognition system. If the real-time failure frequency is lower than the failure frequency threshold, remind the staff. If the real-time failure frequency is higher than the failure frequency threshold, calculate the adjusted optimization period.

2. The method for image content recognition based on artificial intelligence according to claim 1, characterized in that: The step S100 includes the following steps: Step S101: Acquire the identity information of the person, collect the feature data of the identity information in the information database, arrange image acquisition equipment on site, collect the image information of the person, perform preprocessing and feature extraction on the collected image data, and generate a recognition record at the same time, match the extracted feature data with the feature data of the identity information, if the match fails, set the feature data that failed to match as abnormal data, and mark the reason for the failure, so as to form an image recognition system; Step S102: Obtain a set of historical recognition failure records, collect the failure cause in a certain historical recognition failure record, summarize the historical recognition failure records with the same failure cause, collect the number of historical recognition failure records with a certain failure cause, and calculate the occurrence frequency of the failure cause according to the following formula: ； Where A represents the occurrence frequency of the failure cause, S represents the number of historical recognition failure records of the failure cause, and S' represents the total number of historical recognition failure records of all failure causes; Step S103: Summarize the occurrence frequencies of all failure causes, divide the occurrence frequencies of failure causes into several failure levels, collect the occurrence frequency range corresponding to a certain failure level, summarize the failure causes that meet a certain failure level, and obtain a failure cause set of a certain failure level.

3. The method for image content recognition based on artificial intelligence according to claim 2, characterized in that: The step S200 includes the following steps: Step S201: In a set of failure reasons at a certain failure level, a set of historical recognition failure records for a certain failure reason is obtained, abnormal data in a certain historical recognition failure record is collected, historical recognition failure records with the same abnormal data are summarized, and the number of historical recognition failure records is collected, and the frequency of abnormal data occurrence is calculated according to the following formula: ； Wherein, B represents the frequency of abnormal data occurrence, C represents the number of historical recognition failure records of abnormal data, and C' represents the number of historical recognition failure records of failure causes; Step S202: setting a frequency threshold of abnormal data occurrence, setting abnormal data whose frequency exceeds the frequency threshold as target data, and obtaining target data of a certain failure cause.

4. The method for image content recognition based on artificial intelligence according to claim 3, characterized in that: The step S300 includes the following steps: Step S301: select a number of consecutive days as a training cycle, divide a day into a number of characteristic time periods, and obtain the time range corresponding to the characteristic time period; Step S302: Obtain a set of recognition failure records within a training cycle, collect the recognition time point in a certain recognition failure record, obtain the characteristic time period corresponding to the recognition time point, and then obtain the characteristic time period of the recognition failure record, summarize the recognition failure records belonging to the same characteristic time period, and obtain a set of recognition failure records of a certain characteristic time period; Step S303: in the set of recognition failure records in a certain characteristic time period, the failure cause of a certain recognition failure record is obtained, the recognition failure records with the same failure cause are summarized, and abnormal data in a certain recognition failure record in the set of recognition failure records with a certain failure cause is extracted. If the abnormal data is the target data of the failure cause, the recognition failure record is marked as the target record; Step S304: Collect the target number of a certain failure reason marked as a target record in a certain characteristic time period, summarize the target number of a certain failure reason in a certain characteristic time period in the training cycle, and calculate the frequency of the failure reason in the characteristic time period according to the following formula: ； Where D represents the frequency of the failure cause in the characteristic time period, E represents the target number of failure causes in the characteristic time period, and E' represents the target total number of failure causes; Step S305: setting a frequency threshold of the failure cause, marking a characteristic time period in which the frequency of the failure cause exceeds the frequency threshold as a target time period, and obtaining a target time period for a certain failure cause.

5. The method for image content recognition based on artificial intelligence according to claim 4, characterized in that: The step S400 includes the following steps: Step S401: On a certain day in the training cycle, the number of all recognition records in the target time period of a certain failure reason is collected, and the failure frequency of a certain failure reason on a certain day is calculated according to the following formula: ； Among them, F ir It is expressed as the failure frequency of the target time period for the rth failure reason on the i-th day, H ir It is represented as the number of recognition failure records of the target time period of the rth failure reason in the i-th day, H ir ' represents the number of all identified records in the target time period for the rth failure reason on the i-th day; Step S402: sorting the failure frequencies of a certain failure cause within the training cycle in chronological order, setting a failure frequency threshold, marking the time when the failure frequency is lower than the failure frequency threshold as abnormal time, and obtaining an abnormal time set of a certain failure cause; Step S403: sorting the abnormal time set of a certain failure cause in chronological order, obtaining the bit sequence corresponding to each abnormal time, setting the abnormal time number threshold, and when the bit sequence corresponding to the abnormal time reaches the abnormal time number threshold, setting the abnormal time as the optimization period of the failure cause; Step S404: Summarize the optimization cycles of all failure reasons in a certain failure level, and calculate the optimization cycle of the image recognition system according to the following formula: ； Where T represents the optimization period of the image recognition system, t ae It is expressed as the optimization period of the e-th failure reason in the a-th failure level, f a Expressed as the total number of failure reasons in the ath failure level, α a It is represented as the weight of the ath failure level, b is represented as the total number of failure levels, and the optimization cycle of the image recognition system is obtained.

6. The method for image content recognition based on artificial intelligence according to claim 5, characterized in that: The step S500 includes the following steps: Step S501: within a target time period of a certain failure reason, a real-time recognition record set is obtained, the number of real-time recognition records within the target time period is collected, and the real-time failure reason of a certain real-time recognition failure record in the real-time recognition record set is extracted. If the real-time failure reason is the failure reason, the real-time recognition failure record is marked, the number of marked real-time recognition failure records within the target time period is collected, and the real-time failure frequency of the failure reason within the optimization period of the image recognition system is calculated. If the real-time failure frequency is lower than the failure frequency threshold, the staff is reminded to optimize the image recognition system. If the real-time failure frequency is higher than the failure frequency threshold, step S502 is executed; Step S502: Summarize the real-time failure frequencies of all failure reasons in a certain failure level, and calculate the adjusted optimization period according to the following formula: ； Among them, T' represents the adjusted optimization period, R ae It is expressed as the real-time failure frequency of the e-th failure cause in the a-th failure level. The real-time failure frequency of the failure cause is calculated according to the calculated adjusted optimization period.

7. An artificial intelligence-based image content recognition system, used to implement an artificial intelligence-based image content recognition method as described in any one of claims 1 to 6, characterized in that: The system includes an image recognition system module, a target data module, a target time period module, an optimization cycle module of the image recognition system, and a real-time adjustment module; The image recognition system module: collects the identity information of the personnel, collects the image information of the personnel, identifies the personnel, forms an image recognition system, collects the failure reasons in the historical recognition failure records, calculates the occurrence frequency of a certain failure reason, and determines the failure reason set of a certain failure level; The target data module: obtains a set of historical recognition failure records of a certain failure cause, collects abnormal data in the historical recognition failure records, calculates the frequency of occurrence of abnormal data, and determines the target data of a certain failure cause; The target time period module: sets a training cycle, divides a day into several characteristic time periods, obtains a set of recognition failure records for a certain characteristic time period, calculates the frequency of failure reasons in the characteristic time period, and determines a target time period for a certain failure reason; The optimization cycle module of the image recognition system: collects the number of all recognition records in a target time period of a certain failure reason on a certain day within the training cycle, calculates the failure frequency of a certain failure reason on a certain day, determines the optimization cycle of a certain failure reason, and calculates the optimization cycle of the image recognition system; The real-time adjustment module calculates the real-time failure frequency of the failure reasons within the optimization period of the image recognition system, and if the real-time failure frequency is lower than the failure frequency threshold, reminds the staff; if the real-time failure frequency is higher than the failure frequency threshold, calculates the adjusted optimization period.

8. The image content recognition system based on artificial intelligence according to claim 7, characterized in that: The image recognition system module includes an image recognition system unit and a failure level determination unit: The image recognition system unit is constructed as follows: obtaining the identity information of the personnel, collecting the feature data of the identity information in the information database, collecting the image information of the personnel by arranging image acquisition equipment on site, and preprocessing and extracting features of the collected image data, and generating an identification record at the same time, matching the extracted feature data with the feature data of the identity information, and if the match fails, setting the feature data that fails to match as abnormal data, and marking the reason for the failure, so as to form an image recognition system; The failure level determination unit: obtains a set of historical recognition failure records, collects the failure cause in a historical recognition failure record, summarizes the historical recognition failure records with the same failure cause, collects the number of historical recognition failure records with a certain failure cause, calculates the occurrence frequency of the failure cause, summarizes the occurrence frequency of all failure causes, divides the occurrence frequency of the failure cause into several failure levels, collects the occurrence frequency range corresponding to a certain failure level, summarizes the failure causes that meet a certain failure level, and obtains a failure cause set for a certain failure level.

9. The image content recognition system based on artificial intelligence according to claim 7, characterized in that: The target time period module includes a target recording unit and a target time period determination unit: The target record determination unit: selects several consecutive days as a training cycle, divides a day into several characteristic time periods, obtains the time range corresponding to the characteristic time period, obtains a set of recognition failure records within the training cycle, collects the recognition time point in a certain recognition failure record, obtains the characteristic time period corresponding to the recognition time point, and then obtains the characteristic time period of the recognition failure record, summarizes the recognition failure records belonging to the same characteristic time period, obtains the recognition failure record set of a certain characteristic time period, obtains the failure cause of a certain recognition failure record in the recognition failure record set of a certain characteristic time period, summarizes the recognition failure records of the same failure cause, and extracts abnormal data from a certain recognition failure record in the recognition failure record set of a certain failure cause, and if the abnormal data is the target data of the failure cause, marks the recognition failure record as a target record; The target time period determination unit collects the target number of a certain failure cause marked as a target record in a certain feature time period, summarizes the target number of a certain failure cause in a certain feature time period within a training cycle, calculates the frequency of the failure cause in the feature time period, sets a frequency threshold of the failure cause, marks the feature time period whose frequency of the failure cause exceeds the frequency threshold as a target time period, and obtains a target time period for a certain failure cause.

10. The image content recognition system based on artificial intelligence according to claim 7, characterized in that: The optimization cycle module of the image recognition system includes an abnormal time determination unit and an optimization cycle calculation unit of the image recognition system: The abnormal time determination unit: on a certain day in the training cycle, collect the number of all identification records in the target time period of a certain failure reason, calculate the failure frequency of a certain failure reason on a certain day, sort the failure frequency of a certain failure reason in the training cycle in chronological order, set a failure frequency threshold, and mark the time when the failure frequency is lower than the failure frequency threshold as abnormal time; The optimization cycle unit of the image recognition system is configured to sort the abnormal time set of a certain failure cause in chronological order, obtain the bit sequence corresponding to each abnormal time, set the abnormal time number threshold, and when the bit sequence corresponding to the abnormal time reaches the abnormal time number threshold, set the abnormal time as the optimization cycle of the failure cause, summarize the optimization cycles of all failure causes in a certain failure level, calculate the optimization cycle of the image recognition system, and obtain the optimization cycle of the image recognition system.

Images