Intelligent identification method and system for law enforcement recorder

Abstract

The invention discloses an intelligent recognition method and system for a law enforcement recorder, relates to the field of videos, and solves the problem that an existing intelligent recognition method for the law enforcement recorder is poor in recognition effect. The method comprises the following steps: S1, analyzing law enforcement videos in a real-time analysis period of a recorder to obtain law enforcement period division data and a characteristic law enforcement image quantity ratio of the recorder, and obtaining recorder image identification data, S2, carrying out real-time audio analysis on the law enforcement recorder according to the recorder image identification data, obtaining an audio period abnormal index, and carrying out video analysis according to the audio period abnormal index. S3, monitoring and early warning are conducted on the real-time analysis period of the recorder according to the recorder image recognition data and the audio period abnormal index, the effect of the law enforcement recorder can be brought into full play, and the law enforcement efficiency and the law enforcement safety of law enforcement officers are improved.

Description

Technical Field

[0001] The present invention belongs to the field of video and relates to image recognition technology, in particular to an intelligent recognition method and system for law enforcement recorders. Background Art

[0002] The intelligent identification method of law enforcement recorders has the following specific defects in the process of law enforcement:

[0003] 1. The existing intelligent recognition method of law enforcement recorders cannot process the image data recorded by law enforcement recorders in real time, and cannot identify potential risks or dangerous situations in real time, which easily puts law enforcement officers in a passive state when responding, increasing the risk of performing tasks;

[0004] 2. The existing intelligent recognition method of law enforcement recorders cannot simultaneously analyze and alarm dangerous situations through image data and audio data recorded by law enforcement recorders in real time, and cannot give full play to the role of law enforcement recorders;

[0005] To this end, we propose an intelligent identification method and system for law enforcement recorders. Summary of the invention

[0006] In view of the shortcomings of the prior art, the purpose of the present invention is to provide an intelligent identification method and system for law enforcement recorders. The present invention is based on obtaining a real-time analysis period of a recorder, respectively obtaining multiple law enforcement monitoring images within the real-time analysis period of the recorder, creating a law enforcement image recognition model to identify the multiple law enforcement monitoring images, and dividing the real-time analysis period of the recorder into a first type of law enforcement period and a second type of law enforcement period according to the recognition result to obtain recorder law enforcement period division data, and analyzing the second type of law enforcement period to obtain a characteristic law enforcement image quantity ratio, defining the characteristic law enforcement image quantity ratio and the recorder law enforcement period division data as recorder image recognition data, performing voice acquisition on the second type of law enforcement period according to the recorder image recognition data to obtain multiple law enforcement scene audios, creating a law enforcement audio recognition model to identify multiple law enforcement scene audios to obtain multiple target law enforcement recorder audios, obtaining an audio period anomaly index corresponding to the second type of law enforcement period by analyzing the multiple target law enforcement recorder audios, and monitoring and warning the recorder real-time analysis period according to the recorder image recognition data and the audio period anomaly index.

[0007] In order to achieve the above object, the present invention adopts the following technical solution: an intelligent identification method for law enforcement recorders, comprising the following specific steps:

[0008] Step S1: obtaining a real-time analysis cycle of a recorder, respectively obtaining a plurality of law enforcement monitoring images within the real-time analysis cycle of the recorder, creating a law enforcement image recognition model to recognize the plurality of law enforcement monitoring images, dividing the real-time analysis cycle of the recorder into a first type of law enforcement cycle and a second type of law enforcement cycle according to the recognition result, obtaining the law enforcement cycle division data of the recorder, and analyzing the second type of law enforcement cycle to obtain the number ratio of characteristic law enforcement images, and defining the number ratio of characteristic law enforcement images and the law enforcement cycle division data of the recorder as recorder image recognition data;

[0009] Step S2: Acquire the voice of the second type of law enforcement cycle according to the recorder image recognition data to obtain multiple law enforcement scene audios, create a law enforcement audio recognition model, and identify multiple law enforcement scene audios to obtain multiple target law enforcement recorder audios. By analyzing the multiple target law enforcement recorder audios, obtain the audio cycle anomaly index corresponding to the second type of law enforcement cycle;

[0010] Step S3: Monitor and warn the real-time analysis cycle of the recorder based on the recorder image recognition data and the audio cycle abnormality index.

[0011] The specific working process of each module of the law enforcement recorder intelligent identification system is as follows:

[0012] Image recognition module: used to obtain the real-time analysis cycle of the recorder, respectively obtain multiple law enforcement monitoring images within the real-time analysis cycle of the recorder, create a law enforcement image recognition model to recognize the multiple law enforcement monitoring images, divide the real-time analysis cycle of the recorder into a first type of law enforcement cycle and a second type of law enforcement cycle according to the recognition result, obtain the recorder law enforcement cycle division data, and analyze the second type of law enforcement cycle to obtain the number ratio of characteristic law enforcement images, and define the number ratio of characteristic law enforcement images and the recorder law enforcement cycle division data as recorder image recognition data;

[0013] Speech recognition module: used to obtain speech for the second type of law enforcement cycle according to the recorder image recognition data, obtain multiple law enforcement scene audios, create a law enforcement audio recognition model to recognize the multiple law enforcement scene audios to obtain multiple target law enforcement recorder audios, and obtain the audio cycle anomaly index corresponding to the second type of law enforcement cycle by analyzing the multiple target law enforcement recorder audios;

[0014] Intelligent early warning module: used to monitor and warn the real-time analysis cycle of the recorder based on the recorder image recognition data and audio cycle abnormality index.

[0015] In summary, due to the adoption of the above technical solution, the beneficial effects of the present invention are:

[0016] 1. The present invention processes the image data recorded by the law enforcement recorder in real time by creating a law enforcement image recognition model and a law enforcement audio recognition model, which can effectively identify potential risks or dangerous situations and reduce the risks of law enforcement personnel in performing their tasks;

[0017] 2. The present invention obtains the audio cycle anomaly index and the number of characteristic law enforcement images, compares the image data and audio data recorded by the law enforcement recorder, and simultaneously analyzes dangerous situations and issues an alarm, thereby fully utilizing the law enforcement recorder. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] In order to facilitate understanding by those skilled in the art, the present invention is further described below with reference to the accompanying drawings.

[0019] Figure 1 It is a diagram of the implementation steps of the present invention;

[0020] Figure 2 is the overall system block diagram of the present invention;

[0021] Figure 3 It is the law enforcement image of the recorder in the present invention. DETAILED DESCRIPTION

[0022] 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, 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.

[0023] Embodiment 1

[0024] See also Figure 1 The present invention provides a technical solution: a method for intelligently identifying law enforcement recorders, comprising the following specific steps:

[0025] Step S1: obtaining a real-time analysis cycle of a recorder, respectively obtaining a plurality of law enforcement monitoring images within the real-time analysis cycle of the recorder, creating a law enforcement image recognition model to recognize the plurality of law enforcement monitoring images, dividing the real-time analysis cycle of the recorder into a first type of law enforcement cycle and a second type of law enforcement cycle according to the recognition result, obtaining the law enforcement cycle division data of the recorder, and analyzing the second type of law enforcement cycle to obtain the number ratio of characteristic law enforcement images, and defining the number ratio of characteristic law enforcement images and the law enforcement cycle division data of the recorder as recorder image recognition data;

[0026] The step S1 further includes the following specific steps:

[0027] Step S11: When the law enforcement recorder is working, the time value corresponding to the current moment is marked as the first image feature time point, the time value corresponding to the feature monitoring period before the first image feature time point is marked as the second image feature time point, and the period between the first image feature time point and the second image feature time point is marked as the recorder real-time analysis period;

[0028] Step S12: intercepting the law enforcement video images of the recorder within the real-time analysis period of the recorder frame by frame to obtain a plurality of law enforcement images of the recorder and obtain intercepted data of the law enforcement images of the recorder;

[0029] Step S13: In the real-time analysis cycle of the recorder, a number of time points are randomly selected to obtain a plurality of video monitoring time points, and the interval time between each two consecutive video monitoring points is equal in length, and the recorder law enforcement image corresponding to each video monitoring point is obtained according to the recorder law enforcement image interception data to obtain a plurality of law enforcement monitoring images;

[0030] Step S14: acquiring the number of law enforcement monitoring images within the real-time analysis period of the recorder to obtain the law enforcement image period number value;

[0031] Step S15: Creating a law enforcement image recognition model;

[0032] The step S15 further includes the following specific steps:

[0033] Step S151: using big data crawler technology to obtain multiple law enforcement recorder images using law enforcement recorder images as keywords;

[0034] Step S152: manually marking each law enforcement recorder image as a first type of law enforcement recorder image, a second type of law enforcement recorder image, and a third type of law enforcement recorder image to obtain law enforcement recorder image marking data;

[0035] Step S153: dividing the law enforcement recorder image labeling data into a law enforcement image training set and a law enforcement image test set according to the image training test ratio;

[0036] Step S154: creating an image recognition model through an existing artificial intelligence platform, and training the image recognition model using the law enforcement image training set, until each law enforcement camera image in the law enforcement image training set has trained the image recognition model once;

[0037] Step S155: Use the law enforcement image test set to test the image recognition model and obtain the recognition accuracy. When the recognition accuracy is greater than or equal to the target recognition accuracy, the image recognition model training is completed and the law enforcement image recognition model is obtained. When the recognition accuracy is less than the target recognition accuracy, continue to use the law enforcement image training set to train the image recognition model until the recognition accuracy is greater than or equal to the target recognition accuracy.

[0038] Step S16: using the law enforcement image recognition model to identify each law enforcement monitoring image within the real-time analysis period of the recorder, and dividing the real-time analysis period of the recorder into a first type of law enforcement period and a second type of law enforcement period according to the recognition result, to obtain the law enforcement period division data of the recorder;

[0039] The step S16 further includes the following specific steps:

[0040] Step S161: When the law enforcement image recognition model recognizes that any law enforcement monitoring image within the recorder real-time analysis period is a first-type law enforcement recorder image, the recorder real-time analysis period is classified as a first-type law enforcement period;

[0041] Step S162: when the law enforcement image recognition model recognizes that any law enforcement monitoring image in the recorder real-time analysis period is not a first-type law enforcement recorder image, the recorder real-time analysis period is divided into a second-type law enforcement period;

[0042] Step S17: when the real-time analysis period of the recorder is a first type of law enforcement period, acquiring a second type of law enforcement recorder image from a plurality of law enforcement monitoring images in the real-time analysis period of the recorder to obtain a number value of the second type of images;

[0043] Step S18: Calculate the ratio of the second type image quantity value to the law enforcement image period quantity value to obtain the characteristic law enforcement image quantity ratio;

[0044] Step S19: defining the characteristic law enforcement image quantity ratio and the recorder law enforcement cycle division data as recorder image recognition data;

[0045] Step S2: Acquire the voice of the second type of law enforcement cycle according to the recorder image recognition data to obtain multiple law enforcement scene audios, create a law enforcement audio recognition model, and identify multiple law enforcement scene audios to obtain multiple target law enforcement recorder audios. By analyzing the multiple target law enforcement recorder audios, obtain the audio cycle anomaly index corresponding to the second type of law enforcement cycle;

[0046] The step S2 further includes the following specific steps:

[0047] Step S21: obtaining the recorder image recognition data, and obtaining the characteristic law enforcement image quantity ratio and the recorder law enforcement cycle division data according to the recorder image recognition data;

[0048] Step S22: obtaining the recorder audio data corresponding to the second type of law enforcement period, and using audio software to divide the recorder audio data into several different types of law enforcement scene audios to obtain multiple law enforcement scene audios;

[0049] Step S23: Creating a law enforcement audio recognition model;

[0050] The step S23 further includes the following specific steps:

[0051] Step S231: using big data crawler technology to obtain multiple law enforcement recorder audios using law enforcement recorder audios as keywords;

[0052] Step S232: manually marking each law enforcement recorder audio as a target law enforcement recorder audio and a non-target law enforcement recorder audio, to obtain law enforcement recorder audio marking data;

[0053] Step S233: dividing the law enforcement recorder audio tag data into a law enforcement audio training set and a law enforcement audio test set according to the audio training test ratio;

[0054] Step S234: Create an audio recognition model through an existing artificial intelligence platform, and use the law enforcement audio training set to train the audio recognition model until each law enforcement recorder audio in the law enforcement audio training set is used to train the audio recognition model once;

[0055] Step S235: Use the law enforcement audio test set to test the audio recognition model and obtain the audio recognition accuracy. When the audio recognition accuracy is greater than or equal to the target audio recognition accuracy, the audio recognition model training is completed and the law enforcement audio recognition model is obtained. When the audio recognition accuracy is less than the target audio recognition accuracy, continue to use the law enforcement audio training set to train the audio recognition model until the audio recognition accuracy is greater than or equal to the target audio recognition accuracy.

[0056] Step S24: using the law enforcement audio recognition model to recognize multiple law enforcement scene audios, and obtaining the recognition results as target law enforcement recorder audios to obtain multiple target law enforcement recorder audios;

[0057] Step S25: naming the obtained multiple target law enforcement recorder audios as the first target law enforcement audio to the ath target law enforcement audio;

[0058] Step S26: respectively obtain the periodic audio loudness corresponding to the first target law enforcement audio to the ath target law enforcement audio in the second type of law enforcement cycle, and obtain the average loudness of the first audio cycle to the average loudness of the ath audio cycle;

[0059] Step S27: respectively obtain the duration of the cycles corresponding to the first target law enforcement audio to the ath target law enforcement audio in the second type of law enforcement cycle, and obtain the duration of the first audio cycle to the ath audio cycle;

[0060] Step S28: Calculate the audio cycle abnormality index corresponding to the second type of enforcement cycle by combining the average loudness of the first audio cycle to the average loudness of the a-th audio cycle and the duration of the first audio cycle to the duration of the a-th audio cycle;

[0061] The audio cycle anomaly index corresponding to the second type of enforcement cycle is calculated using the following formula:

[0062]

[0063] Among them, Yyz is the audio cycle abnormality index corresponding to the second type of enforcement cycle, Scx1 to Scxa are respectively the duration of the first audio cycle to the duration of the ath audio cycle, and Xdc1 to Xdca are respectively the average loudness of the first audio cycle to the average loudness of the ath audio cycle;

[0064] Step S3: monitoring and warning the real-time analysis cycle of the recorder according to the recorder image recognition data and the audio cycle abnormality index;

[0065] The step S3 further includes the following specific steps:

[0066] Step S31: obtaining the recorder image recognition data, and obtaining the recorder law enforcement cycle division data according to the recorder image recognition data;

[0067] Step S32: When the real-time analysis period of the recorder is the first type of law enforcement period, the law enforcement recorder automatically issues a video warning;

[0068] Step S33: when the real-time analysis period of the recorder is the second type of law enforcement period, further study and judge the real-time analysis period of the recorder;

[0069] The step S33 further includes the following specific steps:

[0070] Step S331: obtaining a characteristic law enforcement image quantity ratio according to the recorder image recognition data;

[0071] Step S332: Obtaining an audio cycle abnormality index;

[0072] Step S333: Calculate the video anomaly coefficient corresponding to the second type of law enforcement cycle by using the characteristic law enforcement image quantity ratio and the audio cycle anomaly index;

[0073] The video abnormality coefficient is calculated, and the specific formula is as follows:

[0074] Spy=Yyz 2 +Tsb;

[0075] Among them, Spy is the video anomaly coefficient, Yyz is the audio cycle anomaly index, and Tsb is the number ratio of characteristic law enforcement images;

[0076] Step S334: obtaining a video abnormality coefficient threshold, comparing the video abnormality coefficient with the video abnormality coefficient threshold, and issuing an early warning for the real-time analysis period of the recorder in the second type of law enforcement period according to the numerical comparison result;

[0077] The step S334 further includes the following specific steps:

[0078] Step S3341: Obtaining a characteristic law enforcement image quantity ratio threshold and an audio cycle abnormality index threshold;

[0079] Step S3342: Calculate the video anomaly coefficient threshold corresponding to the second type of law enforcement cycle by using the characteristic law enforcement image quantity ratio threshold and the audio cycle anomaly index threshold;

[0080] The video anomaly coefficient threshold is calculated, and the specific formula is as follows:

[0081] Spyy=Yyzy 2 +Tsby;

[0082] Among them, Spy is the video anomaly coefficient, Yyz is the audio cycle anomaly index threshold, and Tsb is the characteristic law enforcement image number ratio threshold;

[0083] Step S3343: When the video abnormality coefficient is greater than or equal to the video abnormality coefficient threshold, the law enforcement recorder automatically issues a video warning;

[0084] Step S3344: When the video anomaly coefficient is less than the video anomaly coefficient threshold, the law enforcement recorder does not issue a video warning.

[0085] In this application, if corresponding calculation formulas appear, the above calculation formulas are all dimensionless and take their numerical calculations. The weight coefficients, proportional coefficients and other coefficients in the formulas are set to a result value obtained by quantifying each parameter. The size of the weight coefficient and the proportional coefficient can be determined as long as it does not affect the proportional relationship between the parameter and the result value.

[0086] Embodiment 2

[0087] See also Figure 2 , based on another concept of the same invention, an intelligent recognition system for law enforcement recorders is now proposed, including an image recognition module, a voice recognition module, an intelligent warning module and a server, wherein the image recognition module, the voice recognition module and the intelligent warning module are respectively connected to the server, and the server controls the image recognition module, the voice recognition module and the intelligent warning module respectively;

[0088] The image recognition module obtains the real-time analysis cycle of the recorder, obtains multiple law enforcement monitoring images within the real-time analysis cycle of the recorder respectively, creates a law enforcement image recognition model to recognize the multiple law enforcement monitoring images, divides the real-time analysis cycle of the recorder into a first type of law enforcement cycle and a second type of law enforcement cycle according to the recognition result, obtains the recorder law enforcement cycle division data, and analyzes the second type of law enforcement cycle to obtain the characteristic law enforcement image quantity ratio, and defines the characteristic law enforcement image quantity ratio and the recorder law enforcement cycle division data as recorder image recognition data;

[0089] When the law enforcement recorder is working, the time value corresponding to the current moment is marked as the first image feature time point, the time value corresponding to the feature monitoring period before the first image feature time point is marked as the second image feature time point, and the period between the first image feature time point and the second image feature time point is marked as the recorder real-time analysis period;

[0090] It should be noted here that:

[0091] In this application, the time length corresponding to the characteristic monitoring period involved here is specifically 30 seconds;

[0092] In the present application, as the time value corresponding to the current moment changes, the first characteristic monitoring time point and the second characteristic monitoring time point also change accordingly, thereby realizing the dynamic update of the real-time analysis cycle of the recorder;

[0093] Intercepting the law enforcement video images of the recorder frame by frame within the real-time analysis period of the recorder to obtain a plurality of law enforcement images of the recorder and obtain interception data of the law enforcement images of the recorder;

[0094] In the real-time analysis cycle of the recorder, several time points are randomly selected to obtain multiple video monitoring time points, and the interval time between each two consecutive video monitoring points is equal. The recorder law enforcement image corresponding to each video monitoring point is obtained according to the recorder law enforcement image interception data to obtain multiple law enforcement monitoring images;

[0095] The number of law enforcement monitoring images within the real-time analysis period of the recorder is obtained to obtain the number value of the law enforcement image period;

[0096] Creating image recognition models for law enforcement;

[0097] The details are as follows:

[0098] Using big data crawler technology to obtain multiple law enforcement camera images using law enforcement camera images as keywords;

[0099] By manually marking each law enforcement recorder image as a first type of law enforcement recorder image, a second type of law enforcement recorder image, and a third type of law enforcement recorder image, respectively, to obtain law enforcement recorder image marking data;

[0100] It should be noted here that:

[0101] In this application, the first type of law enforcement recorder images involved here are law enforcement images with great social harm, including but not limited to violent law enforcement, punitive law enforcement and violent incidents. The second type of law enforcement recorder images involved here are law enforcement images with general social harm, including but not limited to routine arrest law enforcement, small-scale conflicts and non-physical resistance to law enforcement. The third type of law enforcement recorder images involved here are law enforcement images that are not socially harmful, including but not limited to law enforcement patrols, routine traffic law enforcement and legal inspections. Please refer to Figure 3 ;

[0102] The law enforcement recorder image labeling data is divided into a law enforcement image training set and a law enforcement image test set according to the image training and testing ratio;

[0103] It should be noted here that:

[0104] In this application, the image training-test ratio is specifically set to 7:3, that is, the ratio of the number of law enforcement camera images in the law enforcement image training set and the law enforcement image test set is 7:3;

[0105] Create an image recognition model through an existing artificial intelligence platform, and train the image recognition model using the law enforcement image training set until the image recognition model is trained once for each body camera image in the law enforcement image training set;

[0106] The image recognition model is tested using the law enforcement image test set, and the recognition accuracy is obtained. When the recognition accuracy is greater than or equal to the target recognition accuracy, the image recognition model training is completed and the law enforcement image recognition model is obtained. When the recognition accuracy is less than the target recognition accuracy, the image recognition model is continuously trained using the law enforcement image training set until the recognition accuracy is greater than or equal to the target recognition accuracy.

[0107] It should be noted here that:

[0108] The target recognition accuracy involved here is specifically set to 95% in this application;

[0109] Use the law enforcement image recognition model to identify each law enforcement monitoring image within the real-time analysis period of the recorder, and divide the real-time analysis period of the recorder into a first type of law enforcement period and a second type of law enforcement period according to the recognition result, so as to obtain the law enforcement period division data of the recorder;

[0110] The details are as follows:

[0111] When the law enforcement image recognition model recognizes that any one of the multiple law enforcement monitoring images in the recorder real-time analysis period is a first-type law enforcement recorder image, the recorder real-time analysis period is classified as a first-type law enforcement period;

[0112] When the law enforcement image recognition model recognizes that any law enforcement monitoring image in the recorder real-time analysis period is not a first-type law enforcement recorder image, the recorder real-time analysis period is divided into a second-type law enforcement period;

[0113] When the real-time analysis period of the recorder is a first type of law enforcement period, a second type of law enforcement recorder image is acquired from a plurality of law enforcement monitoring images in the real-time analysis period of the recorder to obtain a quantity value of the second type of images;

[0114] Calculate the ratio of the second type image quantity value to the law enforcement image period quantity value to obtain the characteristic law enforcement image quantity ratio;

[0115] The characteristic law enforcement image quantity ratio and the data of the division of the law enforcement cycle of the recorder are defined as the recorder image recognition data;

[0116] The image recognition module acquires the image recognition data of the recorder and transmits it to the voice recognition module and the intelligent warning module;

[0117] The speech recognition module acquires speech of the second type of law enforcement cycle according to the recorder image recognition data, obtains multiple law enforcement scene audios, creates a law enforcement audio recognition model to recognize the multiple law enforcement scene audios to acquire multiple target law enforcement recorder audios, and obtains the audio cycle anomaly index corresponding to the second type of law enforcement cycle by analyzing the multiple target law enforcement recorder audios;

[0118] Acquire the recorder image recognition data, and acquire the characteristic law enforcement image quantity ratio and the recorder law enforcement cycle division data respectively according to the recorder image recognition data;

[0119] It should be noted here that:

[0120] The data on the division of the enforcement cycles of the recorders involved here include the first type of enforcement cycle and the second type of enforcement cycle;

[0121] Acquire the recorder audio data corresponding to the second type of law enforcement cycle, and use audio software to divide the recorder audio data into several different types of law enforcement scene audios to obtain multiple law enforcement scene audios;

[0122] Creating law enforcement audio recognition models;

[0123] The details are as follows:

[0124] Use big data crawler technology to obtain multiple law enforcement recorder audios using law enforcement recorder audio as a keyword;

[0125] Each law enforcement recorder audio is respectively marked as a target law enforcement recorder audio and a non-target law enforcement recorder audio by manual marking, to obtain law enforcement recorder audio marking data;

[0126] It should be noted here that:

[0127] In this application, the target law enforcement recorder audio involved here is the law enforcement audio with great social harm, including but not limited to insults, fighting background sound and group riot background sound; the non-target law enforcement recorder audio involved here is the law enforcement audio without social harm, including but not limited to law enforcement patrol audio, routine traffic law enforcement audio and legal inspection audio;

[0128] The law enforcement recorder audio labelled data is divided into a law enforcement audio training set and a law enforcement audio test set according to the audio training and testing ratio;

[0129] It should be noted here that:

[0130] In this application, the audio training test ratio is specifically set to 7:3, that is, the ratio of the number of law enforcement recorder audios in the law enforcement audio training set and the law enforcement audio test set is 7:3;

[0131] Create an audio recognition model through an existing artificial intelligence platform, and train the audio recognition model using the law enforcement audio training set until each law enforcement camera audio in the law enforcement audio training set is used to train the audio recognition model once;

[0132] The audio recognition model is tested using the law enforcement audio test set, and the audio recognition accuracy is obtained. When the audio recognition accuracy is greater than or equal to the target audio recognition accuracy, the audio recognition model training is completed and the law enforcement audio recognition model is obtained. When the audio recognition accuracy is less than the target audio recognition accuracy, the audio recognition model is continuously trained using the law enforcement audio training set until the audio recognition accuracy is greater than or equal to the target audio recognition accuracy.

[0133] It should be noted here that:

[0134] The target audio recognition accuracy involved here is specifically set to 95% in this application;

[0135] Use the law enforcement audio recognition model to recognize multiple law enforcement scene audios, and obtain the recognition results as target law enforcement recorder audios to obtain multiple target law enforcement recorder audios;

[0136] The obtained multiple target law enforcement recorder audios are named as the first target law enforcement audio to the ath target law enforcement audio respectively;

[0137] It should be noted here that:

[0138] In this application, a referred to herein is the quantity value corresponding to the audio of the target law enforcement recorder, and a is an integer greater than 0;

[0139] Respectively obtain the periodic audio loudness corresponding to the first target law enforcement audio to the ath target law enforcement audio in the second type of law enforcement cycle, and obtain the average loudness of the first audio cycle to the ath audio cycle;

[0140] Respectively obtain the duration of the cycles corresponding to the first target law enforcement audio to the ath target law enforcement audio in the second type of law enforcement cycle, and obtain the duration of the first audio cycle to the ath audio cycle;

[0141] The audio cycle anomaly index corresponding to the second type of enforcement cycle is obtained by calculating the average loudness of the first audio cycle to the average loudness of the ath audio cycle and the duration of the first audio cycle to the duration of the ath audio cycle;

[0142] The audio cycle anomaly index corresponding to the second type of enforcement cycle is calculated using the following formula:

[0143]

[0144] Among them, Yyz is the audio cycle abnormality index corresponding to the second type of enforcement cycle, Scx1 to Scxa are respectively the duration of the first audio cycle to the duration of the ath audio cycle, and Xdc1 to Xdca are respectively the average loudness of the first audio cycle to the average loudness of the ath audio cycle;

[0145] The speech recognition module obtains the audio cycle abnormality index and transmits it to the intelligent warning module;

[0146] The intelligent early warning module monitors and warns the real-time analysis cycle of the recorder based on the recorder image recognition data and audio cycle abnormality index;

[0147] Obtaining the recorder image recognition data, and obtaining the recorder enforcement cycle division data according to the recorder image recognition data;

[0148] When the real-time analysis cycle of the recorder is the first type of law enforcement cycle, the law enforcement recorder automatically issues a video warning;

[0149] When the real-time analysis cycle of the recorder is the second type of law enforcement cycle, further analysis and judgment of the real-time analysis cycle of the recorder is conducted;

[0150] The details are as follows:

[0151] The ratio of the number of characteristic law enforcement images obtained based on the image recognition data of the recorder;

[0152] Get the audio cycle anomaly index;

[0153] The video anomaly coefficient corresponding to the second type of enforcement cycle is obtained by calculating the number ratio of characteristic enforcement images and the audio cycle anomaly index;

[0154] The video abnormality coefficient is calculated, and the specific formula is as follows:

[0155] Spy=Yyz 2 +Tsb;

[0156] Among them, Spy is the video anomaly coefficient, Yyz is the audio cycle anomaly index, and Tsb is the number ratio of characteristic law enforcement images;

[0157] Obtain a video anomaly coefficient threshold, compare the video anomaly coefficient with the video anomaly coefficient threshold, and issue an early warning for the real-time analysis period of the recorder in the second type of law enforcement period according to the numerical comparison result;

[0158] The details are as follows:

[0159] Obtaining a threshold value of the number ratio of characteristic law enforcement images and a threshold value of an abnormal index of an audio cycle;

[0160] The video anomaly coefficient threshold corresponding to the second type of law enforcement cycle is obtained by calculating the characteristic law enforcement image quantity ratio threshold and the audio cycle anomaly index threshold;

[0161] The video anomaly coefficient threshold is calculated, and the specific formula is as follows:

[0162] Spyy=Yyzy 2 +Tsby;

[0163] Among them, Spy is the video anomaly coefficient, Yyz is the audio cycle anomaly index threshold, and Tsb is the characteristic law enforcement image number ratio threshold;

[0164] When the video anomaly coefficient is greater than or equal to the video anomaly coefficient threshold, the law enforcement recorder automatically issues a video warning;

[0165] When the video anomaly coefficient is less than the video anomaly coefficient threshold, the law enforcement recorder will not issue a video warning.

[0166] The preferred embodiments of the present invention disclosed above are only used to help explain the present invention. The preferred embodiments do not describe all the details in detail, nor do they limit the invention to only specific implementation methods. Obviously, many modifications and changes can be made according to the content of this specification. This specification selects and specifically describes these embodiments in order to better explain the principles and practical applications of the present invention, so that those skilled in the art can understand and use the present invention well. The present invention is limited only by the claims and their full scope and equivalents.

Claims

1. An intelligent identification method for law enforcement recorders, characterized in that: include: Step S1: obtaining a real-time analysis cycle of a recorder, respectively obtaining a plurality of law enforcement monitoring images within the real-time analysis cycle of the recorder, creating a law enforcement image recognition model to recognize the plurality of law enforcement monitoring images, dividing the real-time analysis cycle of the recorder into a first type of law enforcement cycle and a second type of law enforcement cycle according to the recognition result, obtaining the law enforcement cycle division data of the recorder, and analyzing the second type of law enforcement cycle to obtain the number ratio of characteristic law enforcement images, and defining the number ratio of characteristic law enforcement images and the law enforcement cycle division data of the recorder as recorder image recognition data; Step S2: Acquire voice for the second type of law enforcement cycle according to the recorder image recognition data to obtain multiple law enforcement scene audios, create a law enforcement audio recognition model to recognize the multiple law enforcement scene audios, and obtain multiple target law enforcement recorder audios, and obtain the audio cycle anomaly index corresponding to the second type of law enforcement cycle by analyzing the multiple target law enforcement recorder audios; Step S3: Monitor and warn the real-time analysis cycle of the recorder based on the recorder image recognition data and the audio cycle abnormality index.

2. The intelligent identification method of law enforcement recorder according to claim 1 is characterized in that: The step S1 further includes the following specific steps: Step S11: When the law enforcement recorder is in operation, the time value corresponding to the current moment is marked as the first image feature time point, the time value corresponding to the feature monitoring period before the first image feature time point is marked as the second image feature time point, and the period between the first image feature time point and the second image feature time point is marked as the real-time analysis period of the recorder.

3. The intelligent identification method of law enforcement recorder according to claim 2 is characterized in that: The step S1 further includes the following specific steps: Step S12: intercepting the law enforcement video image of the recorder within the real-time analysis period of the recorder frame by frame to obtain a plurality of law enforcement images of the recorder, and obtaining intercepted data of the law enforcement image of the recorder; Step S13: In the real-time analysis cycle of the recorder, several time points are randomly selected to obtain multiple video monitoring time points, and the interval time between each two consecutive video monitoring points is equal. According to the recorder law enforcement image interception data, the recorder law enforcement image corresponding to each video monitoring point is obtained to obtain multiple law enforcement monitoring images.

4. The intelligent identification method of law enforcement recorder according to claim 3 is characterized in that: The step S1 also includes the following specific steps: Step S14: acquiring the number of law enforcement monitoring images within the real-time analysis period of the recorder to obtain the law enforcement image period number value; Step S15: Creating a law enforcement image recognition model; Step S16: Use the law enforcement image recognition model to identify each law enforcement monitoring image within the recorder's real-time analysis period, and divide the recorder's real-time analysis period into a first type of law enforcement period and a second type of law enforcement period according to the recognition result to obtain the recorder's law enforcement period division data.

5. The intelligent identification method of law enforcement recorder according to claim 4 is characterized in that: The step S1 further includes the following specific steps: Step S17: when the real-time analysis period of the recorder is a first-type law enforcement period, acquiring a second-type law enforcement recorder image from a plurality of law enforcement monitoring images in the real-time analysis period of the recorder to obtain a second-type image quantity value; Step S18: Calculate the ratio of the second type image quantity value to the law enforcement image period quantity value to obtain the characteristic law enforcement image quantity ratio; Step S19: Define the characteristic law enforcement image quantity ratio and the recorder law enforcement period division data as recorder image recognition data.

6. The intelligent identification method of law enforcement recorder according to claim 5 is characterized in that: The step S15 further includes the following specific steps: Step S151: using big data crawler technology to obtain multiple law enforcement recorder images using law enforcement recorder images as keywords; Step S152: manually marking each law enforcement recorder image as a first type of law enforcement recorder image, a second type of law enforcement recorder image, and a third type of law enforcement recorder image to obtain law enforcement recorder image marking data; Step S153: Divide the law enforcement recorder image marking data into a law enforcement image training set and a law enforcement image test set according to the image training test ratio.

7. The intelligent identification method of law enforcement recorder according to claim 6 is characterized in that: Step S154: creating an image recognition model through an existing artificial intelligence platform, and training the image recognition model using the law enforcement image training set, until each law enforcement recorder image in the law enforcement image training set has trained the image recognition model once; Step S155: Use the law enforcement image test set to test the image recognition model and obtain the recognition accuracy. When the recognition accuracy is greater than or equal to the target recognition accuracy, the image recognition model training is completed and the law enforcement image recognition model is obtained. When the recognition accuracy is less than the target recognition accuracy, continue to use the law enforcement image training set to train the image recognition model until the recognition accuracy is greater than or equal to the target recognition accuracy.

8. The intelligent identification method of law enforcement recorder according to claim 7 is characterized in that: The step S16 further includes the following specific steps: Step S161: When the law enforcement image recognition model recognizes that any law enforcement monitoring image within the recorder real-time analysis period is a first-type law enforcement recorder image, the recorder real-time analysis period is classified as a first-type law enforcement period; Step S162: When the law enforcement image recognition model recognizes that any law enforcement monitoring image in the recorder real-time analysis period is not a first-type law enforcement recorder image, the recorder real-time analysis period is divided into a second-type law enforcement period.

9. The intelligent identification method of law enforcement recorder according to claim 1, characterized in that: The step S3 further includes the following specific steps: Step S31: obtaining the recorder image recognition data, and obtaining the recorder law enforcement cycle division data according to the recorder image recognition data; Step S32: When the real-time analysis period of the recorder is the first type of law enforcement period, the law enforcement recorder automatically issues a video warning; Step S33: When the real-time analysis cycle of the recorder is the second type of law enforcement cycle, further analysis and judgment are performed on the real-time analysis cycle of the recorder.

10. An intelligent identification system for law enforcement recorders, applicable to the intelligent identification method for law enforcement recorders described in any one of claims 1 to 9, characterized in that: The specific working process of each module of the intelligent recognition system is as follows: Image recognition module: used to obtain the real-time analysis cycle of the recorder, respectively obtain multiple law enforcement monitoring images within the real-time analysis cycle of the recorder, create a law enforcement image recognition model to recognize the multiple law enforcement monitoring images, divide the real-time analysis cycle of the recorder into a first type of law enforcement cycle and a second type of law enforcement cycle according to the recognition result, obtain the recorder law enforcement cycle division data, and analyze the second type of law enforcement cycle to obtain the number ratio of characteristic law enforcement images, and define the number ratio of characteristic law enforcement images and the recorder law enforcement cycle division data as recorder image recognition data; Speech recognition module: used to obtain speech for the second type of law enforcement cycle according to the recorder image recognition data, obtain multiple law enforcement scene audios, create a law enforcement audio recognition model to recognize the multiple law enforcement scene audios to obtain multiple target law enforcement recorder audios, and obtain the audio cycle anomaly index corresponding to the second type of law enforcement cycle by analyzing the multiple target law enforcement recorder audios; Intelligent early warning module: used to monitor and warn the real-time analysis cycle of the recorder based on the recorder image recognition data and audio cycle abnormality index.

Images