An artificial intelligence-based data management system for audio and video acquisition equipment

Abstract

The present invention relates to the field of data processing technology, and in particular to an artificial intelligence-based audio and video acquisition equipment data management system, comprising: a receiving module, used to receive audio and video data collected by the audio and video acquisition equipment during operation, and store the received audio and video data; a preprocessing module, used to obtain the audio and video data stored in the receiving module, and extract audio data from the obtained audio and video data; the present invention extracts audio data from the audio and video data as a reference for tracking targets in the audio and video data, captures characteristic frequency bands in the audio data, and constructs a database, thereby tracking the query target based on selecting the characteristic frequency band corresponding to the query target and combining the calculation of the similarity of the characteristic frequency bands. This is different from the existing machine vision technology for tracking targets, so that the means of tracking specified targets in the audio and video data have more possibilities, and the accuracy and comprehensiveness of tracking targets in the audio and video data are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of data processing, and in particular to an artificial intelligence-based data management system for audio and video acquisition equipment. Background Art

[0002] Audio and video capture devices are tools used to record sound and images. Common examples include cameras, which shoot high-quality video, capture light through an optical lens, and store the images on built-in memory or a memory card. Other devices include voice recorders, which use highly sensitive microphones to precisely capture sound. These devices play a key role in many fields, including film and television production, interviews, and meeting recording.

[0003] The invention patent with application number 202011241905.X discloses an artificial intelligence-based video behavior analysis and management system, including several mobile terminals, data acquisition modules, positioning modules, scene switching modules, alarm modules, display modules, behavior analysis modules and task processing modules; several mobile terminals are used for users to register and log in to the video behavior analysis and management system, and send the personal information of successful registration to the online platform. Several mobile terminals are also used for users to shoot video images of multiple scenes and upload the video images to the online platform: the data acquisition module is used to collect video images of multiple scenes, and the data acquisition module includes high-definition cameras installed in multiple scenes and several mobile terminals; the positioning module is used to locate the video images and mobile terminals, and send the positioning information to the online platform: the scene switching module is used to switch video images between multiple scenes; the display module is used to display video images of multiple scenes; the alarm module is used to alarm for abnormal video behavior; the behavior analysis module is used to analyze the behavior in the video images of multiple scenes. The specific analysis process is as follows.

[0004] This application aims to solve the problem that: "In current technology, most video surveillance systems are still in the traditional mode, that is, "only recording but not judging", so abnormal situations and evidence can only be investigated through subsequent video playback. There is a disadvantage that abnormal behavior and alarm cannot be judged in real time, and staff are required to continuously monitor activities in the scene. The workload is heavy and it is easy to be affected by human sensory fatigue, resulting in missed detections and false detections, which loses the meaning of the monitoring system to conduct real-time on-site monitoring; the current video surveillance system cannot avoid the situation from worsening. The camera cannot automatically "discover the alarm" and actively "analyze" the monitoring targets in the field of view in real time, nor can it determine whether the behavior of these monitored targets poses a security threat. It cannot promptly issue alarms to security personnel for security threats that have occurred or are about to occur. The security prevention capabilities of the monitoring area need to be improved: as the scale of the monitoring system expands, the number of videos increases massively, and it is becoming increasingly difficult to obtain useful information or intelligence from them. The search efficiency is low, and it is difficult to meet the needs of the monitoring system."

[0005] Based on the above and existing technologies, the current technology for continuously tracking a specified target in a video image still remains at the visual technology level, and does not consider auditory technology to track the target.

[0006] To this end, an artificial intelligence-based data management system for audio and video acquisition equipment is proposed. Summary of the Invention

[0007] In view of the above-mentioned shortcomings of the prior art, the present invention provides an audio and video acquisition equipment data management system based on artificial intelligence, which solves the technical problems raised in the above-mentioned background technology.

[0008] To achieve the above objectives, the present invention is implemented through the following technical solutions:

[0009] An artificial intelligence-based audio and video acquisition equipment data management system, comprising:

[0010] The receiving module is used to receive the audio and video data collected by the audio and video acquisition device and store the received audio and video data; the preprocessing module is used to obtain the audio and video data stored in the receiving module and extract audio data from the obtained audio and video data; the marking module is used to receive the audio data extracted by the preprocessing module and distinguish the characteristic frequency bands in the audio data; the selection module is used to select the characteristic frequency bands that have been marked in the marking module and forward the characteristic frequency bands to the query module; the query module is used to receive the characteristic frequency bands selected in the selection module and query similar characteristic frequency bands in the remaining unselected characteristic frequency bands; the output module is used to receive the characteristic frequency bands queried in the query module and output the characteristic frequency bands.

[0011] Furthermore, the audio and video acquisition device is a monitoring device deployed in any public scene. When storing the audio and video data, the receiving module simultaneously marks the location coordinates of the audio and video acquisition device in the public scene where the audio and video data is located. After extracting the audio data from the audio and video data, the pre-processing module synchronously iterates the extracted audio data to the receiving module, and the iteration target is the audio and video data corresponding to the audio data.

[0012] The pre-processing module is provided with sub-modules at the lower level, including:

[0013] a copy unit for receiving the audio data extracted by the pre-processing module from the audio and video data and backing up the audio data;

[0014] The audio data copied by the copying unit and the audio data extracted from the audio and video data by the pre-processing module are respectively used for iteration to the receiving module and forwarding to the marking module.

[0015] Furthermore, after receiving the audio data, the marking module synchronously evaluates the quality of the audio data and decides whether to perform noise reduction processing on the audio data based on the audio data quality evaluation result;

[0016] The audio data quality assessment logic is expressed as:

[0017] ;

[0018] Where: Q is the audio data quality performance value; P signal is the signal power of the audio data; P noise is the noise power of the audio data; n is the set of frequency points in the audio data; G i is the gain at the i-th frequency point; G ideal is the ideal gain; P q is the qth harmonic power in the audio data; P1 is the fundamental wave power;

[0019] Wherein, the ideal gain G ideal The value of is customized by the system user. The larger the audio data quality performance value Q is, the higher the audio data quality is. When the audio data quality performance value is less than 0.95, noise reduction processing is performed on the corresponding audio data.

[0020] Furthermore, the noise reduction processing logic of the audio data is expressed as:

[0021] ;

[0022] Where: y(n) is the audio data after noise reduction; N is the number of discrete Fourier transform points; P s (k) is the original audio data; M is the number of noise samples; N m(k) is the discrete value of the mth noise sample in the frequency domain; S(k) is the result of discrete Fourier transform of the original audio data; j is the imaginary unit; k is the index of the frequency discrete point; t is the index of the time domain;

[0023] The audio data after the noise reduction process is processed synchronously with the corresponding audio spectrum outputted by Adobe Audition, and the audio spectrum corresponding to the audio data after the noise reduction process is provided to the lower-level sub-modules of the marking module for running applications.

[0024] Furthermore, the marking module is provided with submodules at the lower level, including:

[0025] The segmentation unit is used to obtain the audio spectrum corresponding to the audio data received by the marking module and perform segmentation processing on the audio spectrum;

[0026] A cloud database is used to receive the characteristic frequency bands marked by the marking module and store the marked characteristic frequency bands;

[0027] Among them, during the operation stage of the segmentation unit, the system user customizes the segmentation time domain, and the segmentation unit segments the audio spectrum based on the segmentation time domain to obtain several groups of audio segments. The marking module uses the audio segment as the characteristic frequency band identification target and performs the characteristic frequency band identification operation. The marking module distinguishes the characteristic frequency band with the following contents: the location coordinates of the audio and video acquisition device to which the audio data from the characteristic frequency band source belongs and the timestamp corresponding to the characteristic frequency band. When the cloud database stores the characteristic frequency band, it distinguishes and stores it based on the audio data from the characteristic frequency band source.

[0028] Furthermore, the recognition logic of the characteristic frequency band is expressed as:

[0029] Set the characteristic frequency band screening threshold, calculate the average sound intensity of each characteristic frequency band, and screen the audio segments that meet the characteristic frequency band screening threshold;

[0030] Arrange the filtered audio segments in descending order based on their respective average sound intensities, and obtain the audio segments in the first half of the arrangement queue as feature frequency bands;

[0031] Arrange the filtered audio segments in descending order based on their average sound intensity, obtain the audio segments in the second half of the arrangement queue, use the second half of the audio segments as capture targets, identify whether there are audio segments with a similarity greater than 95% for each capture target in the first half of the audio segments, and use the corresponding capture targets with audio segments with a similarity greater than 95% as feature frequency bands.

[0032] Furthermore, when selecting characteristic frequency bands, the selection module uses characteristic frequency bands stored in the cloud database as selection targets, and when the query module queries similar characteristic frequency bands, the remaining unselected characteristic frequency bands are derived from the cloud database.

[0033] Furthermore, during the operation phase of the query module, the timestamp of the selected characteristic frequency band is synchronously read, and based on the read timestamp in the cloud database, the characteristic frequency band after the timestamp is selected as the query target;

[0034] The similarity of the characteristic frequency bands in the query module is calculated by the following formula:

[0035] ;

[0036] Where: SIMM (α, β) is the similarity between the characteristic frequency band α and the characteristic frequency band β; G is the sampling point in the characteristic frequency band; α g is the energy of the g-th sampling point in the characteristic frequency band α; β g is the energy of the g-th sampling point in the characteristic frequency band β; ε is the correction factor;

[0037] in, Express The similarity calculation logic of the audio segments is the same as the above formula. The correction factor ε takes a value of 1 or -1. When the numerator of the fraction where the correction factor ε is located is less than or equal to the denominator, the correction factor ε takes a value of 1. Otherwise, the correction factor ε takes a value of -1. When SIMM(α, β) ≥ 95%, it indicates that the two feature frequency bands are feature frequency bands of similarity.

[0038] Furthermore, when the characteristic frequency band outputted by the output module is outputted, characteristic evaluation mark information is simultaneously acquired, and the characteristic frequency band is synchronously outputted together with its mark information;

[0039] The output module is connected to a mobile computer device held by a system end user via a wireless network, and the system end user reads the output content of the output module on the mobile computer device.

[0040] Furthermore, the receiving module is interactively connected to the preprocessing module via a wireless network, the preprocessing module is interactively connected to the copy unit via a wireless network, the preprocessing module is interactively connected to the marking module via a wireless network, the marking module is interactively connected to the segmentation unit and the cloud database via a wireless network, the marking module is interactively connected to the selection module, the query module and the output module via a wireless network, and the selection module and the query module are interactively connected to the cloud database via a wireless network.

[0041] Compared with the known public technology, the technical solution provided by the present invention has the following beneficial effects:

[0042] The present invention provides an artificial intelligence-based data management system for audio and video acquisition equipment. During operation, the system extracts audio data from audio and video data as a reference for tracking targets in the audio and video data, captures characteristic frequency bands in the audio data, and constructs a database. The system then tracks the query target based on selecting the characteristic frequency band corresponding to the query target and calculating the similarity of the characteristic frequency bands. This method is different from existing machine vision technology for tracking targets, and provides more possibilities for tracking specified targets in audio and video data, thereby improving the accuracy and comprehensiveness of tracking targets in audio and video data. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] To more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are only some embodiments of the present invention. Those skilled in the art can also derive other drawings based on these drawings without inventive effort.

[0044] Figure 1 The diagram is a structural diagram of an artificial intelligence-based data management system for audio and video acquisition equipment. DETAILED DESCRIPTION

[0045] To make the purpose, technical solutions, and advantages of the embodiments of the present invention more clear, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings 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 making any creative efforts shall fall within the scope of protection of the present invention.

[0046] The present invention will be further described below with reference to the embodiments.

[0047] Example:

[0048] This embodiment is an artificial intelligence-based audio and video acquisition equipment data management system, such as Figure 1 As shown, including:

[0049] The receiving module is used to receive the audio and video data collected by the audio and video acquisition device and store the received audio and video data;

[0050] A pre-processing module, configured to obtain the audio and video data stored in the receiving module and extract audio data from the obtained audio and video data;

[0051] The audio and video acquisition device is a monitoring device deployed in any public scene. When the receiving module stores the audio and video data, it is synchronously marked with the location coordinates of the audio and video acquisition device in the public scene where the audio and video data is located. After the pre-processing module extracts the audio data from the audio and video data, it synchronously iterates the extracted audio data to the receiving module, and the iteration target is the audio and video data corresponding to the audio data;

[0052] The pre-processing module is equipped with sub-modules, including:

[0053] a copy unit for receiving the audio data extracted by the pre-processing module from the audio and video data and backing up the audio data;

[0054] The audio data copied by the copying unit and the audio data extracted from the audio and video data by the pre-processing module are respectively used to iterate to the receiving module and forwarded to the marking module;

[0055] The marking module is used to receive the audio data extracted by the pre-processing module and distinguish and mark the characteristic frequency bands in the audio data;

[0056] After receiving the audio data, the marking module synchronously evaluates the audio data quality and decides whether to perform noise reduction processing on the audio data based on the audio data quality evaluation result;

[0057] The audio data quality assessment logic is expressed as:

[0058] ;

[0059] Where: Q is the audio data quality performance value; P signal is the signal power of the audio data; P noise is the noise power of the audio data; n is the set of frequency points in the audio data; G i is the gain at the i-th frequency point; G ideal is the ideal gain; P q is the qth harmonic power in the audio data; P1 is the fundamental wave power;

[0060] Wherein, the ideal gain G ideal The value of is customized by the system user. The larger the audio data quality performance value Q is, the higher the audio data quality is. When the audio data quality performance value is less than 0.95, noise reduction processing is performed on the corresponding audio data.

[0061] The quality of the audio data is evaluated by the above audio data quality evaluation logic formula, thereby providing a decision for the noise reduction processing of the audio data.

[0062] The noise reduction processing logic of audio data is expressed as:

[0063] ;

[0064] Where: y(n) is the audio data after noise reduction; N is the number of discrete Fourier transform points; P s (k) is the original audio data; M is the number of noise samples; N m (k) is the discrete value of the mth noise sample in the frequency domain; S(k) is the result of discrete Fourier transform of the original audio data; j is the imaginary unit; k is the index of the frequency discrete point; t is the index of the time domain;

[0065] The audio data after the noise reduction process is synchronously output with the corresponding audio spectrum based on Adobe Audition, and the audio spectrum corresponding to the audio data after the noise reduction process is provided to the lower-level sub-modules of the marking module for running applications;

[0066] Through the above logic formula, the specified noise reduction processing logic is provided for the noise reduction processing of the audio data.

[0067] The marking module is equipped with submodules, including:

[0068] The segmentation unit is used to obtain the audio spectrum corresponding to the audio data received by the marking module and perform segmentation processing on the audio spectrum;

[0069] A cloud database is used to receive the characteristic frequency bands marked by the marking module and store the marked characteristic frequency bands;

[0070] In the running stage of the segmentation unit, the user of the system defines the segmentation time domain. The segmentation unit segments the audio spectrum based on the segmentation time domain to obtain several groups of audio segments. The marking module uses the audio segment as the target of characteristic frequency band identification and performs the characteristic frequency band identification operation. The marking module distinguishes the characteristic frequency bands with the following contents: the location coordinates of the audio and video acquisition device to which the audio data of the characteristic frequency band source belongs and the timestamp corresponding to the characteristic frequency band. When the cloud database stores the characteristic frequency bands, it distinguishes and stores them based on the audio data of the characteristic frequency band source.

[0071] The recognition logic of the characteristic frequency band is expressed as:

[0072] Set the characteristic frequency band screening threshold, calculate the average sound intensity of each characteristic frequency band, and screen the audio segments that meet the characteristic frequency band screening threshold;

[0073] Arrange the filtered audio segments in descending order based on their respective average sound intensities, and obtain the audio segments in the first half of the arrangement queue as feature frequency bands;

[0074] Arrange the filtered audio segments in descending order based on their average sound intensity, obtain the audio segments in the second half of the arrangement queue, use the second half of the audio segments as capture targets, identify whether there are audio segments with a similarity greater than 95% for each capture target in the first half of the audio segments, and use the capture targets corresponding to the identified audio segments with a similarity greater than 95% as feature frequency bands;

[0075] A selection module is used to select the characteristic frequency band marked by the marking module and forward the characteristic frequency band to the query module;

[0076] A query module, configured to receive the characteristic frequency band selected by the selection module and query similar characteristic frequency bands in the remaining unselected characteristic frequency bands;

[0077] During the query module operation phase, the timestamp of the selected characteristic frequency band is read synchronously, and the characteristic frequency band after the timestamp is selected as the query target based on the read timestamp in the cloud database;

[0078] The similarity of the characteristic frequency bands in the query module is calculated using the following formula:

[0079] ;

[0080] Where: SIMM (α, β) is the similarity between the characteristic frequency band α and the characteristic frequency band β; G is the sampling point in the characteristic frequency band; α g is the energy of the g-th sampling point in the characteristic frequency band α; β g is the energy of the g-th sampling point in the characteristic frequency band β; ε is the correction factor;

[0081] in, Express The logic for calculating the similarity of the audio segments is the same as the above formula. The correction factor ε takes a value of 1 or -1. When the numerator of the fraction where the correction factor ε is located is less than or equal to the denominator, the correction factor ε takes a value of 1. Otherwise, the correction factor ε takes a value of -1. When SIMM(α, β) ≥ 95%, it indicates that the two feature frequency bands are feature frequency bands of similarity.

[0082] Through the above logic formula, the similarity calculation logic of the characteristic frequency band is further limited, ensuring that the necessary operation data support is provided for the operation of the output module in the above system.

[0083] An output module is used to receive the characteristic frequency band queried by the query module and output the characteristic frequency band;

[0084] The receiving module is interactively connected to the preprocessing module through a wireless network, the preprocessing module is interactively connected to the copy unit through a wireless network, the preprocessing module is interactively connected to the marking module through a wireless network, the marking module is interactively connected to the segmentation unit and the cloud database through a wireless network, the marking module is interactively connected to the selection module, the query module and the output module through a wireless network, and the selection module and the query module are interactively connected to the cloud database through a wireless network.

[0085] In this embodiment, the receiving module operates to receive audio and video data collected by the audio and video capture device and stores the received audio and video data. The preprocessing module synchronously obtains the audio and video data stored in the receiving module and extracts audio data from the obtained audio and video data. The copying unit receives the audio data extracted by the preprocessing module in the audio and video data in real time and backs up the audio data. The marking module is post-operated to receive the audio data extracted by the preprocessing module and distinguish the characteristic frequency bands in the audio data. The segmentation unit synchronously obtains the audio spectrum corresponding to the audio data received by the marking module and performs segmentation processing on the audio spectrum. The cloud database receives the characteristic frequency bands marked by the marking module in real time and stores the marked characteristic frequency bands. The selection module then selects the characteristic frequency bands marked by the marking module and forwards the characteristic frequency bands to the query module. The query module further receives the characteristic frequency bands selected by the selection module and queries for similar characteristic frequency bands in the remaining unselected characteristic frequency bands. Finally, the output module receives the characteristic frequency bands queried in the query module and outputs the characteristic frequency bands.

[0086] Through the operation of the system in the above embodiment, another new tracking method is provided for tracking specified targets in audio and video data in the existing technology. It is completely different from the existing machine vision tracking technology. In actual application scenarios, it can be combined with machine vision technology to implement more accurate and effective tracking of specified tracking targets.

[0087] like Figure 1 As shown, when the selection module selects the characteristic frequency band, the characteristic frequency band stored in the cloud database is used as the selection target, and when the query module queries similar characteristic frequency bands, the remaining unselected characteristic frequency bands are derived from the cloud database;

[0088] When the characteristic frequency band output by the output module is output, the characteristic evaluation mark information is obtained synchronously, and the characteristic frequency band is output together with its mark information synchronously;

[0089] The output module is connected to a mobile computer device held by a system end user via a wireless network, and the system end user reads the output content of the output module on the mobile computer device.

[0090] Through the above settings, further operation data support is provided for the operation of the system in the above embodiment, ensuring a more stable operation of the system in the above embodiment.

[0091] In summary, during operation, the system in the above embodiment extracts audio data from the audio and video data as a reference for tracking targets in the audio and video data, captures characteristic frequency bands in the audio data, and constructs a database. Thus, based on selecting the characteristic frequency band corresponding to the query target and combining the calculation of the similarity of the characteristic frequency bands, the query target is tracked. This is different from the existing machine vision technology for tracking targets, and provides more possibilities for tracking specified targets in audio and video data, thereby improving the accuracy and comprehensiveness of tracking targets in audio and video data.

[0092] The above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements will not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the various embodiments of the present invention.

Claims

1. An artificial intelligence-based audio and video acquisition equipment data management system, characterized in that: include: The receiving module is used to receive the audio and video data collected by the audio and video acquisition device and store the received audio and video data; A pre-processing module, configured to obtain the audio and video data stored in the receiving module and extract audio data from the obtained audio and video data; The marking module is used to receive the audio data extracted by the pre-processing module and distinguish and mark the characteristic frequency bands in the audio data; A selection module is used to select the characteristic frequency band marked by the marking module and forward the characteristic frequency band to the query module; A query module, configured to receive the characteristic frequency band selected by the selection module and query similar characteristic frequency bands in the remaining unselected characteristic frequency bands; An output module is used to receive the characteristic frequency band queried by the query module and output the characteristic frequency band; After receiving the audio data, the marking module synchronously evaluates the quality of the audio data and decides whether to perform noise reduction processing on the audio data based on the audio data quality evaluation result; The audio data quality assessment logic is expressed as: Where: Q is the audio data quality performance value; P signal is the signal power of the audio data; P noise is the noise power of the audio data; n is the set of frequency points in the audio data; G i is the gain at the i-th frequency point; G ideal is the ideal gain; P q is the qth harmonic power in the audio data; P1 is the fundamental wave power; Wherein, the ideal gain G ideal The value of is customized by the system user. The larger the audio data quality performance value Q is, the higher the audio data quality is. When the audio data quality performance value is less than 0.95, noise reduction processing is performed on the corresponding audio data. The noise reduction processing logic of the audio data is expressed as follows: Where: y(n) is the audio data after noise reduction; N is the number of discrete Fourier transform points; P s (k) is the original audio data; M is the number of noise samples; N m (k) is the discrete value of the mth noise sample in the frequency domain; S(k) is the result of discrete Fourier transform of the original audio data; j is the imaginary unit; k is the index of the frequency discrete point; t is the index of the time domain; The audio data after the noise reduction process is synchronously output with the corresponding audio spectrum based on Adobe Audition, and the audio spectrum corresponding to the audio data after the noise reduction process is provided to the lower-level sub-modules of the marking module to run the application; The query module synchronously reads the timestamp of the selected characteristic frequency band mark during the running phase, and selects the characteristic frequency band after the timestamp as the query target based on the read timestamp in the cloud database; The similarity of the characteristic frequency bands in the query module is calculated by the following formula: Where: SIMM (α, β) is the similarity between the characteristic frequency band α and the characteristic frequency band β; G is the sampling point in the characteristic frequency band; α g is the energy of the g-th sampling point in the characteristic frequency band α; β g is the energy of the g-th sampling point in the characteristic frequency band β; ε is the correction factor; in, Express The logic for calculating the similarity of the audio segments is the same as the above formula. The correction factor ε takes a value of 1 or -1. When the numerator of the fraction where the correction factor ε is located is less than or equal to the denominator, the correction factor ε takes a value of 1. Otherwise, the correction factor ε takes a value of -1. When SIMM(α, β) ≥ 95%, it indicates that the two feature frequency bands are feature frequency bands of similarity. The audio and video acquisition device is a monitoring device deployed in any public scene. When storing the audio and video data, the receiving module simultaneously marks the location coordinates of the audio and video acquisition device in the public scene where the audio and video data is located. After extracting the audio data from the audio and video data, the pre-processing module synchronously iterates the extracted audio data to the receiving module, with the iteration target being the audio and video data corresponding to the audio data. The pre-processing module is provided with sub-modules at the lower level, including: a copy unit for receiving the audio data extracted by the pre-processing module from the audio and video data and backing up the audio data; The audio data copied by the copying unit and the audio data extracted from the audio and video data by the pre-processing module are respectively used to iterate to the receiving module and forwarded to the marking module; The marking module is provided with submodules at the lower level, including: The segmentation unit is used to obtain the audio spectrum corresponding to the audio data received by the marking module and perform segmentation processing on the audio spectrum; A cloud database is used to receive the characteristic frequency bands marked by the marking module and store the marked characteristic frequency bands; In the operation phase of the segmentation unit, the user of the system defines the segmentation time domain. The segmentation unit segments the audio spectrum based on the segmentation time domain to obtain a number of audio segments. The marking module uses the audio segment as the target of characteristic frequency band identification and performs the characteristic frequency band identification operation. The marking module distinguishes the characteristic frequency bands with the following contents: the location coordinates of the audio and video acquisition device to which the audio data of the characteristic frequency band source belongs and the timestamp corresponding to the characteristic frequency band. When the cloud database stores the characteristic frequency bands, it distinguishes and stores them based on the audio data of the characteristic frequency band source. The identification logic of the characteristic frequency band is expressed as: Set the characteristic frequency band screening threshold, calculate the average sound intensity of each characteristic frequency band, and screen the audio segments that meet the characteristic frequency band screening threshold; Arrange the filtered audio segments in descending order based on their respective average sound intensities, and obtain the audio segments in the first half of the arrangement queue as feature frequency bands; Arrange the filtered audio segments in descending order based on their average sound intensity, obtain the audio segments in the second half of the arrangement queue, use the audio segments in the second half as capture targets, identify whether there are audio segments with a similarity greater than 95% for each capture target in the first half of the audio segments, and use the corresponding capture targets with audio segments with a similarity greater than 95% as feature frequency bands.

2. The artificial intelligence-based audio and video acquisition equipment data management system according to claim 1, characterized in that: When selecting a characteristic frequency band, the selection module uses the characteristic frequency band stored in the cloud database as a selection target. When the query module queries similar characteristic frequency bands, the remaining unselected characteristic frequency bands are derived from the cloud database.

3. The artificial intelligence-based audio and video acquisition equipment data management system according to claim 1, characterized in that: When outputting the characteristic frequency band output by the output module, characteristic evaluation mark information is simultaneously acquired, and the characteristic frequency band and its mark information are synchronously output; The output module is connected to a mobile computer device held by a system end user via a wireless network, and the system end user reads the output content of the output module on the mobile computer device.

4. The artificial intelligence-based audio and video acquisition equipment data management system according to claim 1, characterized in that: The receiving module is interactively connected to the preprocessing module via a wireless network, the preprocessing module is interactively connected to the copy unit via a wireless network, the preprocessing module is interactively connected to the marking module via a wireless network, the marking module is interactively connected to the segmentation unit and the cloud database via a wireless network, the marking module is interactively connected to the selection module, the query module and the output module via a wireless network, and the selection module and the query module are interactively connected to the cloud database via a wireless network.

Images