CN113365018A

CN113365018A - Period adjusting method, device and equipment

Info

Publication number: CN113365018A
Application number: CN202010152896.0A
Authority: CN
Inventors: 乔勇
Original assignee: Hangzhou Hikvision Digital Technology Co Ltd
Current assignee: Hangzhou Hikvision Digital Technology Co Ltd
Priority date: 2020-03-06
Filing date: 2020-03-06
Publication date: 2021-09-07
Anticipated expiration: 2040-03-06
Also published as: CN113365018B

Abstract

The application provides a period adjustment method, a period adjustment device and a period adjustment device, wherein the period adjustment method comprises the following steps: acquiring data to be analyzed based on a target data extraction period; carrying out intelligent analysis on data to be analyzed to obtain an intelligent analysis result return value; and if the target data extraction period is determined to need to be adjusted according to the intelligent analysis result return value, adjusting the target data extraction period, updating the adjusted data extraction period to the target data extraction period, and returning to execute the acquisition of the data to be analyzed based on the target data extraction period. Through the technical scheme, hardware resources can be saved, and the intelligent analysis result is guaranteed to meet the requirement on analysis response time in practical application.

Description

Period adjusting method, device and equipment

Technical Field

The present application relates to the field of monitoring technologies, and in particular, to a method, an apparatus, and a device for adjusting a period.

Background

Deep learning is a new research direction in the field of machine learning, the internal rules and the expression levels of sample data can be learned, and information obtained in the learning process is greatly helpful for explaining data such as characters, images and sounds. The final goal of deep learning is to make a machine capable of human-like analytical learning, and to recognize data such as characters, images, and sounds. Deep learning has achieved many achievements in search technology, data mining, machine learning, machine translation, natural language processing, multimedia learning, speech, recommendation and personalization technologies, and other related fields. The deep learning enables the machine to imitate human activities such as audio-visual and thinking, solves a plurality of complex pattern recognition problems, and makes great progress on the artificial intelligence related technology.

With the popularization of video monitoring technology, a deep learning algorithm can be deployed in the field of video monitoring, for example, intelligent analysis is performed on video images through the deep learning algorithm, so that the workload of manual inspection is effectively reduced, the analysis and judgment of artificial subjective reasons are reduced, and the accuracy of analysis results is improved.

In the related art, intelligent analysis needs to be performed on each frame of video image in a video stream, although the analysis process is more comprehensive, and details of the analysis are not easily lost. However, the above method can perform intelligent analysis on a large number of invalid video images, which wastes hardware resources and cannot make good use of the hardware resources of the intelligent analysis.

Disclosure of Invention

In view of the above, the present application provides a period adjustment method, including:

acquiring data to be analyzed based on a target data extraction period;

carrying out intelligent analysis on the data to be analyzed to obtain an intelligent analysis result return value;

and if the target data extraction period is determined to need to be adjusted according to the intelligent analysis result return value, adjusting the target data extraction period, updating the adjusted data extraction period to the target data extraction period, and returning to execute the acquisition of the data to be analyzed based on the target data extraction period.

The present application provides a period adjustment apparatus, the apparatus comprising:

the acquisition module is used for acquiring data to be analyzed based on the target data extraction period;

the analysis module is used for intelligently analyzing the data to be analyzed to obtain an intelligent analysis result return value;

and the processing module is used for adjusting the target data extraction period if the target data extraction period needs to be adjusted according to the intelligent analysis result return value, and updating the adjusted data extraction period to the target data extraction period.

The present application provides a storage device comprising: a processor and a machine-readable storage medium storing machine-executable instructions executable by the processor;

the processor is configured to execute machine executable instructions to perform the steps of:

acquiring data to be analyzed based on a target data extraction period;

According to the technical scheme, the data to be analyzed are acquired based on the target data extraction period and are subjected to intelligent analysis, each frame of video image in the video stream is not subjected to intelligent analysis, so that intelligent analysis of a large amount of invalid data is reduced, hardware resources are saved, the hardware resources of intelligent analysis are better utilized, and the flexibility of intelligent analysis application is improved. The target data extraction period can be dynamically adjusted according to the return value of the intelligent analysis result, so that the appropriate target data extraction period is set, and the intelligent analysis result can meet the requirement on analysis response time in practical application while hardware resources are saved.

Drawings

In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings needed to be used in the description of the embodiments of the present application or the prior art will be briefly described below, it is obvious that the drawings in the following description are only some embodiments described in the present application, and other drawings can be obtained by those skilled in the art according to the drawings of the embodiments of the present application.

FIG. 1 is a schematic diagram of an application scenario in an embodiment of the present application;

FIG. 2 is a flow chart of a period adjustment method in one embodiment of the present application;

FIGS. 3A and 3B are flow diagrams of a period adjustment method in one embodiment of the present application;

FIG. 4 is a block diagram of a cycle adjustment apparatus according to an embodiment of the present application;

fig. 5 is a block diagram of a storage device according to an embodiment of the present application.

Detailed Description

The terminology used in the embodiments of the present application is for the purpose of describing particular embodiments only and is not intended to be limiting of the application. As used in this application and the claims, the singular forms "a", "an", and "the" are intended to include the plural forms as well, unless the context clearly indicates otherwise. It should also be understood that the term "and/or" as used herein is meant to encompass any and all possible combinations of one or more of the associated listed items.

It should be understood that although the terms first, second, third, etc. may be used in the embodiments of the present application to describe various information, the information should not be limited to these terms. These terms are only used to distinguish one type of information from another. For example, first information may also be referred to as second information, and similarly, second information may also be referred to as first information, without departing from the scope of the present application. Depending on the context, moreover, the word "if" as used may be interpreted as "at … …" or "when … …" or "in response to a determination".

With the popularization of video monitoring technology, a deep learning algorithm can be deployed in the field of video monitoring, for example, intelligent analysis is performed on monitored video images through the deep learning algorithm, so that the workload of manual inspection is effectively reduced, the analysis and judgment of artificial subjective reasons are reduced, and the accuracy of analysis results is improved.

For example, in an application scene of road illegal parking, a video image of the area A can be acquired through a camera, and the video image is intelligently analyzed through a deep learning algorithm, so that whether illegal parking exists in the area A is judged, and a user does not need to go to the area A to manually patrol whether illegal parking exists in the area A. For another example, in an application scenario of garbage bin full detection, a video image of a garbage bin can be acquired through a camera, and the video image is intelligently analyzed through a deep learning algorithm, so that whether the garbage bin is full is judged, and a user does not need to go to the site of the garbage bin to manually patrol whether the garbage bin is full. Of course, the above are just two examples of deploying the deep learning algorithm to the video monitoring field, and the method is not limited thereto.

In one possible implementation, intelligent analysis can be performed on each frame of video image in the video stream, although the analysis process is more comprehensive and the details of the analysis are not easily lost. However, the above method can perform intelligent analysis on a large number of invalid video images, which wastes hardware resources and cannot make good use of the hardware resources of the intelligent analysis. For example, in application scenarios such as garbage bin full detection, road parking against regulations, straw burning detection and the like, it is not necessary to obtain an analysis result quickly in real time, and it is not necessary to perform intelligent analysis on each frame of video image in a video stream, so that if each frame of video image in the video stream is subjected to intelligent analysis, a large number of invalid video images are subjected to intelligent analysis.

In another possible implementation, a data extraction period may be set, and in each data extraction period, a frame of video image is selected from all video images of a video stream, and after the selected frame of video image is converted into a picture, the picture is intelligently analyzed instead of each frame of video image of the video stream, so as to save hardware resources and better utilize the hardware resources of intelligent analysis. For example, assuming that the data extraction period is 30 minutes, one frame of video image is selected at intervals of 30 minutes, and after the selected frame of video image is converted into a picture, the picture is intelligently analyzed.

However, the above method requires a fixed data extraction period, and once the data extraction period is set, the data extraction period cannot be dynamically adjusted, and thus the method lacks flexibility and cannot fully utilize hardware resources.

In another possible implementation, a data extraction period may be set, and in each data extraction period, a frame of video image is selected from all video images of the video stream, and after the selected frame of video image is converted into a picture, the picture is intelligently analyzed to obtain an intelligent analysis result return value. And then, adjusting the data extraction period according to the intelligent analysis result return value, selecting a frame of video image from all video images of the video stream based on the adjusted data extraction period, converting the selected frame of video image into a picture, and then carrying out intelligent analysis on the picture to obtain an intelligent analysis result return value, and so on.

For example, assuming that the data extraction period is 30 minutes, one frame of video image is selected at intervals of 30 minutes, and after the selected frame of video image is converted into a picture, the picture is intelligently analyzed to obtain an intelligent analysis result return value. And then, adjusting the data extraction period according to the intelligent analysis result return value, if the adjusted data extraction period is 60 minutes, selecting one frame of video image at an interval of 60 minutes, converting the selected frame of video image into a picture, and then carrying out intelligent analysis on the picture to obtain an intelligent analysis result return value, and so on. Obviously, in the above mode, intelligent analysis is not performed on each frame of video image of the video stream, so that hardware resources are saved, and the hardware resources of the intelligent analysis are better utilized. Moreover, based on the mode, the data extraction period can be dynamically adjusted, the flexibility of intelligent analysis application is improved, a proper data extraction period can be set, and the intelligent analysis result can meet the requirement of the analysis response time in practical application while hardware resources are saved.

The following describes in detail an implementation manner of setting the data extraction period and adjusting the data extraction period according to the return value of the intelligent analysis result, with reference to a specific embodiment.

Referring to fig. 1, which is a schematic view of an application scenario of the embodiment of the present application, a video monitoring system may include a storage device (also referred to as a backend device) and a front-end device, where the number of the front-end device may be at least one, and fig. 1 takes 3 front-end devices as an example, and in an actual application, the number of the front-end device may also be more. The processing procedure for each front-end device is the same, and the processing procedure of one front-end device is taken as an example in the following.

For example, the storage device may be an NVR (Network Video Recorder), a storage server, or the like, and the type of the storage device is not limited. The front-end device may be an IPC (IP Camera), a video capture device, or the like, and the type of the front-end device is not limited.

Based on the application scenario, an embodiment of the present application provides a period adjustment method, which may be applied to a storage device, and as shown in fig. 2, is a schematic flow diagram of the period adjustment method, where the method includes:

step 201, acquiring data to be analyzed based on the target data extraction period.

For example, an initial data extraction period (which may be set empirically, and is not limited to this initial data extraction period) may be set, the initial data extraction period is used as a target data extraction period, and the data to be analyzed is obtained based on the target data extraction period. In the subsequent process, the target data extraction period may be adjusted, the adjusted data extraction period is updated to the target data extraction period, the step 201 is executed again, and so on, the target data extraction period may be continuously adjusted, and after each adjustment, the adjusted data extraction period is updated to the target data extraction period, and the step 201 is executed again.

For example, the data to be analyzed may be a picture to be analyzed, that is, the data to be analyzed is picture data; or, the data to be analyzed may be a text to be analyzed, that is, the data to be analyzed is text data; alternatively, the data to be analyzed may be audio data to be analyzed, that is, the data to be analyzed is audio data; alternatively, the data to be analyzed may be a video to be analyzed (i.e., a video image in a video stream), i.e., the data to be analyzed is video data. Of course, the above are just a few examples of the data to be analyzed, and the type of the data to be analyzed is not limited.

Step 202, performing intelligent analysis on the data to be analyzed to obtain an intelligent analysis result return value.

For example, the deep learning algorithm may be deployed to a storage device in the video monitoring field, and based on this, after the storage device obtains the data to be analyzed, the data to be analyzed is intelligently analyzed by using the deep learning algorithm (i.e., intelligently analyzed by using a computer vision analysis technology), so as to obtain an intelligent analysis result return value.

For example, the storage device may be configured with an intelligent analysis chip, such as a GPU (Graphic Processing Unit), a CPU (Central Processing Unit), a TPU (temporal Processing Unit, Tensor processor), an FPGA (Field Programmable Gate Array), an embedded processor, and the like, where the intelligent analysis chip performs intelligent analysis on data to be analyzed by running a deep learning algorithm to obtain an intelligent analysis result return value, so that the intelligent analysis result return value can be used to replace or assist a human to perform judgment and Processing. The deep learning algorithm is a neural network intelligent analysis algorithm for simulating human brain to analyze and learn, can intelligently analyze data to be analyzed based on the deep learning algorithm, does not limit the intelligent analysis process of the data to be analyzed, and does not limit the type of the deep learning algorithm.

For example, in an application scene of road illegal parking, intelligent analysis is performed on data to be analyzed through a deep learning algorithm to obtain an intelligent analysis result return value, and the intelligent analysis result return value indicates that illegal parking exists or illegal parking does not exist. If the intelligent analysis result return value is the first identifier, indicating that illegal parking exists; or when the intelligent analysis result return value is the second identifier, the illegal parking does not exist.

For another example, in an application scenario of garbage bin full detection, intelligent analysis is performed on data to be analyzed through a deep learning algorithm, and an intelligent analysis result return value is obtained, wherein the intelligent analysis result return value indicates that a garbage bin is full or the garbage bin is not full. If the intelligent analysis result return value is the first identifier, the garbage can is full; or, when the intelligent analysis result return value is the second identifier, it may indicate that the trash can is not full.

In an application scene of crowd gathering detection, carrying out intelligent analysis on data to be analyzed through a deep learning algorithm to obtain an intelligent analysis result return value, wherein the intelligent analysis result return value represents the gathering state of a designated area. For example, the return value of the intelligent analysis result is 305, which indicates that 305 persons exist in the designated area, and the designated area is dense; or the return value of the intelligent analysis result is 155, which indicates that 155 persons exist in the designated area, and the designated area is normal; or, the return value of the intelligent analysis result is 70, which indicates that 70 persons exist in the designated area and the designated area is rare. For another example, the return value of the intelligent analysis result is a first identifier (e.g., 00) indicating that the designated area is dense (e.g., the number of people aggregated is greater than 300); or the intelligent analysis result return value is a second identifier (such as 01) which indicates that the designated area is normal (such as the number of the aggregated people is between 100 and 300); alternatively, the intelligent analysis result returns a third flag (e.g., 10) indicating that the designated area is rare (e.g., the number of people aggregated is less than 100).

Of course, the above are only a few examples, and the application scenarios are not limited, for example, in application scenarios such as human face analysis and comparison, vehicle analysis and recognition, people number analysis and statistics, object classification, and object detection, the above-mentioned manner can be adopted for processing, and no matter which application scenario, the data to be analyzed can be intelligently analyzed through a deep learning algorithm, so as to obtain an intelligent analysis result return value, and the analysis process is not limited.

Step 203, if it is determined that the target data extraction period needs to be adjusted according to the intelligent analysis result return value, adjusting the target data extraction period, updating the adjusted data extraction period to the target data extraction period, and returning to execute the operation of acquiring the data to be analyzed based on the target data extraction period.

For example, after obtaining the smart analysis result return value, it may be determined whether to adjust the target data extraction period according to the smart analysis result return value. If so, the target data extraction period may be adjusted, and the adjusted data extraction period is updated to a new target data extraction period, and then step 201 is performed based on the new target data extraction period. If not, the target data extraction period may not be adjusted, and step 201 is performed based on the original target data extraction period.

For example, the data b1 to be analyzed is acquired based on the data extraction cycle a1, and the data b1 to be analyzed is intelligently analyzed, so that the intelligent analysis result return value c1 is obtained. And if the data extraction period needs to be adjusted according to the intelligent analysis result return value c1, adjusting the data extraction period a1 to obtain a data extraction period a 2.

In the processing procedure of the next data extraction cycle, the data b2 to be analyzed can be acquired based on the data extraction cycle a2, and the data b2 to be analyzed is intelligently analyzed to obtain an intelligent analysis result return value c 2. If it is determined that the data extraction period is not adjusted according to the smart analysis result return value c2, the data extraction period a2 may be kept unchanged. In the processing process of the next data extraction cycle, the data b3 to be analyzed can be continuously acquired based on the data extraction cycle a2, and the data b3 to be analyzed is intelligently analyzed to obtain an intelligent analysis result return value c 3. And if the data extraction period needs to be adjusted according to the intelligent analysis result return value c3, adjusting the data extraction period a2 to obtain a data extraction period a3, and so on.

In step 203, if the return values of the intelligent analysis results of N consecutive (empirically configurable) data to be analyzed are the return values of the intelligent analysis results concerned by the user, it indicates that the user is concerned about the current situation, and the target data extraction period may be adjusted to be reduced. Or, if the return values of the intelligent analysis results of the M continuous (which can be configured according to experience) pieces of data to be analyzed are return values of the intelligent analysis results which are not concerned by the user, it indicates that the user is not concerned by the current situation, and the target data extraction period may be adjusted in an increasing manner.

According to the technical scheme, the data to be analyzed can be obtained based on the target data extraction period, intelligent analysis is conducted on the data to be analyzed instead of all data, and intelligent analysis of a large amount of invalid data is reduced, so that hardware resources (such as resources of an intelligent analysis chip) can be saved, the hardware resources of intelligent analysis are well utilized, and the flexibility of intelligent analysis application is improved. The target data extraction period can be dynamically adjusted according to the return value of the intelligent analysis result, so that the proper target data extraction period is set, and the intelligent analysis result can meet the requirement on analysis response time in practical application while hardware resources are saved.

In a possible implementation manner, the process of determining that the target data extraction period needs to be adjusted according to the return value of the intelligent analysis result, and adjusting the target data extraction period includes, but is not limited to:

in the first situation, if the return values of the intelligent analysis results of the continuous N data to be analyzed are in a preset first interval, it is determined that the target data extraction period needs to be adjusted. If the target data extraction period is greater than the minimum value of the data extraction period, reducing and adjusting the target data extraction period, wherein the adjusted data extraction period is not less than the minimum value of the data extraction period; or if the target data extraction period is equal to the minimum value of the data extraction period, the target data extraction period is not subjected to reduction adjustment, namely the target data extraction period is kept to be the minimum value of the data extraction period. Illustratively, N is a positive integer, and the preset first interval is set according to the intelligent analysis result, and an interval range or an interval value (i.e. a numerical value) of the data extraction period needs to be reduced.

In case one, the data extraction period minimum, N, and the preset first interval may be configured in advance.

The minimum value of the data extraction period may be configured empirically, and is not limited, for example, 10 minutes, 30 minutes, and the like, and the minimum value of the data extraction period is subsequently recorded as T1.

N may be configured empirically, and the value of N is not limited, such as 1, 2, 3, etc.

The preset first interval is set according to the intelligent analysis result, the interval range or the interval value of the data extraction period needs to be reduced, and the preset first interval can be configured according to experience without limitation.

For example, in an application scenario of road parking violations, if there is a parking violations, the data extraction period needs to be reduced, and therefore, the preset first interval may be an interval range or an interval value of the intelligent analysis result return value when "there is a parking violations". Referring to the above embodiment, when the intelligent analysis result return value is the first identifier, it indicates that there is illegal parking, and thus a preset first interval is set as the first identifier, that is, an interval value.

For another example, in an application scenario of garbage bin full detection, if the garbage bin is full, the data extraction period needs to be reduced, so the preset first interval may be an interval range or an interval value of the intelligent analysis result return value when the garbage bin is full. Referring to the above embodiment, when the intelligent analysis result return value is the first identifier, it indicates that the trash can is full, and therefore, the preset first interval may be set as the first identifier, that is, an interval value.

For another example, in an application scenario of crowd sourcing detection, if the aggregation state is dense, it is necessary to reduce the data extraction period, and therefore, the first interval may be an interval range or an interval value of the return value of the intelligent analysis result when the aggregation state is dense. Referring to the above embodiment, when the return value of the intelligent analysis result is 300-positive infinity, it indicates that the aggregation state is dense, and a preset first interval is set to be 300-positive infinity, that is, an interval range; or, when the intelligent analysis result return value is the first identifier, the aggregation state is dense, and a preset first interval is set as the first identifier, namely an interval value.

Of course, the above are only examples of presetting the first interval, and in different application scenarios, the preset first interval corresponding to the application scenario may be set, which is not limited to this, as long as the preset first interval is set according to the intelligent analysis result, and the interval range or the interval value of the data extraction period needs to be reduced.

And judging whether the intelligent analysis result return value is in the preset first interval or not after the intelligent analysis result return value is obtained based on the preset data extraction period minimum value T1, N and the preset first interval. For example, in an application scene of road illegal parking, if the intelligent analysis result return value is a first identifier, the intelligent analysis result return value is determined to be in a preset first interval; in an application scenario of crowd gathering detection, if an intelligent analysis result return value is greater than 300, it is determined that the intelligent analysis result return value is in a preset first interval.

And if the intelligent analysis result return values are in the preset first interval, judging whether the intelligent analysis result return values of the N continuous data to be analyzed are in the preset first interval. For example, if N is 1, it is determined that the return values of the intelligent analysis results of N consecutive data to be analyzed are all in the preset first interval. If N is 2 and the return value of the intelligent analysis result of the last data to be analyzed is in the preset first interval, determining that the return values of the intelligent analysis results of the N continuous data to be analyzed are in the preset first interval; if N is 2 and the return value of the intelligent analysis result of the last data to be analyzed is not in the preset first interval, determining that the return values of the intelligent analysis results of the N continuous data to be analyzed are not in the preset first interval, and so on.

And if the return values of the intelligent analysis results of the N continuous data to be analyzed are all in a preset first interval, determining that the target data extraction period needs to be adjusted. When the target data extraction period needs to be adjusted, it is also necessary to determine whether the target data extraction period is greater than the minimum value T1 of the data extraction period.

And if the target data extraction period is greater than the minimum value T1 of the data extraction period, performing reduction adjustment on the target data extraction period, wherein the adjusted data extraction period is not less than the minimum value T1 of the data extraction period. For example, when the target data extraction period is subjected to the reduction adjustment, the value of the reduction adjustment is not limited as long as the adjusted data extraction period is not less than the minimum value T1 of the data extraction period.

If the target data extraction period is equal to the data extraction period minimum value T1, then no reduction adjustment is made to the target data extraction period, i.e., the target data extraction period is maintained at the data extraction period minimum value T1.

And in the second situation, if the return values of the intelligent analysis results of the N continuous data to be analyzed are in the preset first interval, determining that the target data extraction period needs to be adjusted, and performing reduction adjustment on the target data extraction period. Illustratively, N is a positive integer, the preset first interval is set according to the intelligent analysis result, and an interval range or an interval value (which may be a numerical value) of the data extraction period needs to be reduced.

In case two, compared to case one, in case two, the limitation of the data extraction period minimum value T1 is not required, that is, the target data extraction period is not required to be restricted to be greater than or equal to the data extraction period minimum value T1. Based on this, when it is determined that the target data extraction period needs to be adjusted, the target data extraction period may be directly subjected to reduction adjustment without determining whether the target data extraction period is greater than the minimum value T1 of the data extraction period, and when the target data extraction period is subjected to reduction adjustment, the value of the reduction adjustment is not limited.

And thirdly, if the return values of the intelligent analysis results of the continuous M data to be analyzed are in a preset second interval, determining that the target data extraction period needs to be adjusted. If the target data extraction period is smaller than the maximum value of the data extraction period, increasing and adjusting the target data extraction period, wherein the adjusted data extraction period is not larger than the maximum value of the data extraction period; or, if the target data extraction period is equal to the maximum value of the data extraction period, the target data extraction period is not increased and adjusted, that is, the target data extraction period is kept to be the maximum value of the data extraction period. Illustratively, M is a positive integer, and the preset second interval is set according to the intelligent analysis result, and an interval range or an interval value (i.e. a numerical value) of the data extraction period needs to be increased.

In case three, the maximum value of the data extraction period, M, and the preset second interval may be configured in advance.

The maximum value of the data extraction period may be configured empirically, and is not limited, such as 200 minutes, 300 minutes, and the like, and the maximum value of the data extraction period is subsequently denoted as T2. For example, the data extraction period maximum value T2 may be greater than the data extraction period minimum value T1.

M can be configured empirically, and the value of M is not limited, such as 1, 2, 3, etc.

The preset second interval is set according to the intelligent analysis result, an interval range or an interval value of a data extraction period needs to be increased, and the preset second interval can be configured according to experience without limitation.

For example, in an application scenario of road parking violations, if there is no parking violations, the data extraction period needs to be increased, and therefore, the preset second interval may be an interval range or an interval value of the return value of the intelligent analysis result when "there is no parking violations". Referring to the above embodiment, when the intelligent analysis result return value is the second identifier, it indicates that there is no illegal parking, and therefore, a preset second interval may be set as the second identifier, that is, an interval value.

For another example, in an application scenario of crowd sourcing detection, if the aggregation state is sparse, it is necessary to increase the data extraction period, and therefore, the preset second interval may be an interval range or an interval value of the return value of the intelligent analysis result when the aggregation state is sparse. Referring to the above embodiment, when the return value of the intelligent analysis result is 0 to 100, it indicates that the aggregation state is rare, and therefore, the preset second interval may be set to 0 to 100, that is, an interval range.

Of course, the above are only a few examples of the preset second interval, and the present invention is not limited thereto.

And judging whether the intelligent analysis result return value is in the preset second interval or not after the intelligent analysis result return value is obtained based on the preset data extraction period maximum value T2, M and the preset second interval. For example, in an application scene of road illegal parking, if the intelligent analysis result return value is a second identifier, determining that the intelligent analysis result return value is in a preset second interval; in the application scene of crowd gathering detection, if the return value of the intelligent analysis result is between 0 and 100, the return value of the intelligent analysis result is determined to be in a preset second interval.

And if the intelligent analysis result return values are in the preset second interval, judging whether the intelligent analysis result return values of the continuous M data to be analyzed are in the preset second interval. For example, if M is 1, it is determined that the return values of the intelligent analysis results of the M consecutive data to be analyzed are all in the preset second interval. If M is 2 and the return value of the intelligent analysis result of the last data to be analyzed is in the preset second interval, determining that the return values of the intelligent analysis results of the M continuous data to be analyzed are all in the preset second interval, if M is 2 and the return value of the intelligent analysis result of the last data to be analyzed is not in the preset second interval, determining that the return values of the intelligent analysis results of the M continuous data to be analyzed are not all in the preset second interval, and so on.

And if the return values of the intelligent analysis results of the continuous M data to be analyzed are all in the preset second interval, determining that the target data extraction period needs to be adjusted. When the target data extraction period needs to be adjusted, it is also necessary to determine whether the target data extraction period is smaller than the maximum value T2 of the data extraction period.

If the target data extraction period is less than the maximum data extraction period T2, the target data extraction period is incrementally adjusted, and the adjusted data extraction period is not greater than the maximum data extraction period T2. For example, when performing the increase adjustment on the target data extraction period, the value of the increase adjustment is not limited as long as the adjusted data extraction period is not greater than the maximum value T2 of the data extraction period.

If the target data extraction period is equal to the data extraction period maximum value T2, no incremental adjustment is made to the target data extraction period, i.e., the target data extraction period is maintained at the data extraction period maximum value T2.

And fourthly, if the return values of the intelligent analysis results of the continuous M data to be analyzed are in a preset second interval, determining that the target data extraction period needs to be adjusted, and increasing and adjusting the target data extraction period. Illustratively, M is a positive integer, and the preset second interval is set according to the intelligent analysis result, and an interval range or an interval value (i.e. a numerical value) of the data extraction period needs to be increased.

In case four, the limitation of the data extraction period maximum value T2 is not required, that is, the target data extraction period is not required to be restricted to be less than or equal to the data extraction period maximum value T2, as compared with case three. Based on this, when it is determined that the target data extraction period needs to be adjusted, the target data extraction period may be directly increased and adjusted without determining whether the target data extraction period is smaller than the maximum value T2 of the data extraction period.

And fifthly, after the data to be analyzed is intelligently analyzed to obtain an intelligent analysis result return value, if the intelligent analysis result return value is in a preset third interval, determining that the target data extraction period does not need to be adjusted. For example, the preset third interval may be an interval range or an interval value (i.e., a numerical value) set according to the intelligent analysis result, which does not need to adjust the data extraction period.

In case five, a preset third interval may be configured in advance, where the preset third interval is set according to the intelligent analysis result, and an interval range or an interval value that does not need to be adjusted for the data extraction period is set, and the preset third interval may be configured according to experience, and is not limited. For example, in an application scenario of crowd sourcing detection, if the aggregation state is normal, the data extraction period does not need to be adjusted, and therefore, the preset third interval may be an interval range or an interval value of the return value of the intelligent analysis result when the aggregation state is normal.

Referring to the above embodiment, when the return value of the intelligent analysis result is 100-; or, when the intelligent analysis result return value is the second identifier, it indicates that the aggregation state is normal, and therefore, a preset third interval may be set as the second identifier, that is, an interval value.

Of course, the above is only an example of the preset third interval, and in different application scenarios, the preset third interval corresponding to the application scenario may be set, or the preset third interval may not be set, which is not limited to this.

And judging whether the intelligent analysis result return value is in the preset third interval or not after the intelligent analysis result return value is obtained based on the preset third interval. For example, in the application scenario of crowd detection, if the return value of the intelligent analysis result is located between 100 and 300, it is determined that the return value of the intelligent analysis result is in the preset third interval. Then, if the intelligent analysis result return value is in a preset third interval, determining that the target data extraction period does not need to be adjusted, namely keeping the target data extraction period unchanged.

And sixthly, after the data to be analyzed is intelligently analyzed to obtain the return value of the intelligent analysis result, if the return value of the intelligent analysis result is in the preset first interval, but the return values of the intelligent analysis results of the N continuous data to be analyzed are not in the preset first interval (if N is 3, currently, the return values of the 2 continuous intelligent analysis results are in the preset first interval), determining that the target data extraction period does not need to be adjusted.

And seventhly, after the data to be analyzed is intelligently analyzed to obtain the return values of the intelligent analysis results, if the return values of the intelligent analysis results are in the preset second interval, but the return values of the intelligent analysis results of the continuous M data to be analyzed are not in the preset second interval (if M is 3, currently, the return values of the continuous 2 intelligent analysis results are in the preset second interval), determining that the target data extraction period does not need to be adjusted.

In summary, based on the above-mentioned cases one to seven, it can be determined whether to adjust the target data extraction period, and when the target data extraction period needs to be adjusted, the target data extraction period can be adjusted to be decreased, or the target data extraction period can be adjusted to be increased.

In one possible implementation, for the first case and the second case, the target data extraction period may be subjected to reduction adjustment, for example, the target data extraction period is subjected to reduction adjustment in the following manner:

the first method is to adopt an equal difference adjustment method, for example, determine a first adjustment tolerance corresponding to a preset first interval, and perform reduction adjustment on the target data extraction period according to the first adjustment tolerance. For example, the adjusted target data extraction period is a difference between the target data extraction period before adjustment and the first adjustment tolerance.

For example, a first adjustment tolerance d1 may be set for a preset first interval. When there are a plurality of preset first intervals, the first adjustment tolerance d1 for the same preset first interval may be a fixed value, i.e. the first adjustment tolerance d1 is not changed, and therefore, the analysis result is the same for each time the first adjustment tolerance d1 is within the preset first interval. The first adjustment tolerance d1 for different preset first intervals may be the same or different, and therefore, the analysis result may be the same or different for each first adjustment tolerance in different preset first intervals.

For example, for the preset first interval 1 and the preset first interval 2, the first adjustment tolerance d1 is set to 2 minutes for the preset first interval 1, and the first adjustment tolerance d1 is kept unchanged for 2 minutes. Of course, in practical applications, the first adjustment tolerance d1 may be adjusted according to actual needs, and is not limited thereto. The first adjustment tolerance d1 for the first preset interval 2 may be set to be 2 minutes (same as the first adjustment tolerance d1 for the first preset interval 1) or 4 minutes (different from the first adjustment tolerance d1 for the first preset interval 1).

For example, in the reduction adjustment of the target data extraction period according to the first adjustment tolerance, if the adjusted target data extraction period is smaller than the above-described data extraction period minimum value T1, the adjusted target data extraction period may be further set to the data extraction period minimum value T1.

For example, if the target data extraction period is 8 minutes, the first adjustment tolerance d1 is 2 minutes, and the minimum value T1 of the data extraction period is 3 minutes, when the target data extraction period is adjusted to be decreased for the first time, the adjusted target data extraction period is 6 minutes, when the target data extraction period is adjusted to be decreased for the second time, the adjusted target data extraction period is 4 minutes, when the target data extraction period is adjusted to be decreased for the third time, the adjusted target data extraction period is 3 minutes, and thereafter, the target data extraction period is not adjusted to be decreased, that is, the target data extraction period is maintained at 3 minutes.

And secondly, adopting an equal ratio adjustment method, for example, determining a first adjustment common ratio corresponding to a preset first interval, and reducing and adjusting the target data extraction period according to the first adjustment common ratio. For example, the adjusted target data extraction period is a product of the target data extraction period before adjustment and the first adjustment common ratio.

For example, the first adjustment common ratio q1 may be set for the preset first interval, and the first adjustment common ratio q1 is a value between 0 and 1, such as 4/5, 3/4, 2/3, 1/2, and the like, which is not limited thereto. When there are a plurality of preset first intervals, the first adjustment common ratio q1 for the same preset first interval may be a fixed value, i.e. the first adjustment common ratio q1 is not changed, and therefore, the analysis result is the same for each time the first adjustment common ratio q1 is within the preset first interval. Of course, in practical applications, the first adjustment common ratio q1 may be adjusted according to actual needs. The first adjustment common ratios q1 for different preset first intervals may be the same or different, and therefore, the analysis results may be the same or different for each first adjustment common ratio q1 in different preset first intervals.

For example, in the process of performing the reduction adjustment on the target data extraction period according to the first adjustment common ratio, if the adjusted target data extraction period is smaller than the above-mentioned data extraction period minimum value T1, the adjusted target data extraction period may be further set to the data extraction period minimum value T1.

For example, if the target data extraction period is 8 minutes, the first adjustment common ratio q1 is 1/2, and the data extraction period minimum value T1 is 3 minutes, when the target data extraction period is adjusted to decrease for the first time, the adjusted target data extraction period is 4 minutes (i.e., 8 × 1/2), and when the target data extraction period is adjusted to decrease for the second time, the adjusted target data extraction period is 3 minutes, and thereafter, the target data extraction period is not adjusted to decrease, that is, the target data extraction period is kept constant for 3 minutes.

In the third mode, a fixed value adjustment method is adopted, for example, the target data extraction period is directly adjusted to the minimum value T1 of the data extraction period, instead of the gradual adjustment process in the first mode or the second mode.

For example, a minimum value T1 of the data extraction period may be set for a preset first interval, and when there are multiple preset first intervals, the minimum value T1 of the data extraction period for the same preset first interval may be a fixed value, that is, the minimum value T1 of the data extraction period for the same preset first interval is unchanged. The minimum values T1 of the data extraction periods for different preset first intervals may be the same or different, that is, the minimum values T1 of the data extraction periods of the analysis results in different preset first intervals may be the same or different.

For example, if the target data extraction period is 8 minutes and the minimum value T1 of the data extraction period is 3 minutes, when the target data extraction period is adjusted to be decreased for the first time, the adjusted target data extraction period is 3 minutes, and thereafter, the target data extraction period is not adjusted to be decreased and is kept unchanged.

Based on the above manner, the reduction adjustment may be performed on the target data extraction period, but the above manner is merely an example, and is not limited thereto, as long as the reduction adjustment can be performed on the target data extraction period.

In one possible implementation, for the case three and the case four, the target data extraction period may be incrementally adjusted, for example, in the following manner:

in the method 1, an equal difference adjustment method is adopted, for example, a second adjustment tolerance corresponding to a preset second interval is determined, and the target data extraction period is increased and adjusted according to the second adjustment tolerance. For example, the adjusted target data extraction period is the sum of the target data extraction period before adjustment and the second adjustment tolerance.

For example, the second adjustment tolerance d2 can be set for the preset second interval, the second adjustment tolerance d2 can be the same as or different from the first adjustment tolerance d1, and the second adjustment tolerance d2 is the same as the first adjustment tolerance d1 for the following description. When there are a plurality of preset second intervals, the second adjustment tolerance d2 for the same preset second interval may be a fixed value, i.e. the second adjustment tolerance d2 is not changed, and therefore, the analysis result is the same for the second adjustment tolerance d2 every time within the preset second interval. Of course, in practical applications, the second adjustment tolerance d2 may also be adjusted according to practical needs, and is not limited thereto. The second adjustment tolerance d2 for different preset second intervals may be the same or different, and therefore, the analysis result may be the same or different for each second adjustment tolerance d2 in different preset second intervals.

For example, in the process of performing the incremental adjustment of the target data extraction period according to the second adjustment tolerance, if the adjusted target data extraction period is greater than the above-mentioned data extraction period maximum value T2, the adjusted target data extraction period may also be set to the data extraction period maximum value T2.

For example, if the target data extraction period is 8 minutes, the second adjustment tolerance d2 is 2 minutes, and the maximum value T2 of the data extraction period is 13 minutes, when the target data extraction period is increased and adjusted for the first time, the adjusted target data extraction period is 10 minutes, when the target data extraction period is increased and adjusted for the second time, the adjusted target data extraction period is 12 minutes, when the target data extraction period is increased and adjusted for the third time, the adjusted target data extraction period is 13 minutes, and thereafter, the target data extraction period is not increased and adjusted, that is, the target data extraction period is maintained at 13 minutes.

And 2, adopting an equal ratio adjustment method, for example, determining a second adjustment common ratio corresponding to the preset second interval, and performing increasing adjustment on the target data extraction period according to the second adjustment common ratio. For example, the adjusted target data extraction period is a product of the target data extraction period before adjustment and the second adjustment common ratio.

For example, a second adjustment common ratio q2 may be set for the preset second interval, and the second adjustment common ratio q2 may be a value greater than 1, such as 5/4, 4/3, 3/2, 2/1, and the like. For example, the second adjusted common ratio q2 may be the inverse of the first adjusted common ratio q 1. When there are a plurality of preset second intervals, the second adjustment common ratio q2 for the same preset second interval may be a fixed value, and the analysis result is the same for each second adjustment common ratio q2 in the preset second interval. Of course, in practical applications, the second adjustment common ratio q2 may also be adjusted according to actual needs. The second adjustment common ratios q2 for different preset second intervals may be the same or different, that is, the analysis results may be the same or different for each second adjustment common ratio q2 in different preset second intervals.

For example, in the process of performing the incremental adjustment on the target data extraction period according to the second adjustment common ratio, if the adjusted target data extraction period is greater than the above-mentioned maximum value T2 of the data extraction period, the adjusted target data extraction period may be further set to the maximum value T2 of the data extraction period.

For example, if the target data extraction period is 8 minutes, the second adjustment common ratio q2 is 2, and the maximum value T2 of the data extraction period is 20 minutes, the adjusted target data extraction period is 16 minutes (i.e., 8 × 2) when the target data extraction period is increased and adjusted for the first time, and the adjusted target data extraction period is 20 minutes when the target data extraction period is increased and adjusted for the second time, and thereafter, the target data extraction period is not increased and adjusted, that is, the target data extraction period is maintained at 20 minutes.

In the mode 3, a fixed value adjustment method is adopted, for example, the target data extraction period is directly adjusted to the maximum value T2 of the data extraction period, instead of the gradual adjustment process in the mode 1 or the mode 2.

For example, the maximum value T2 of the data extraction period may be set for a preset second interval, and when there are a plurality of preset second intervals, the maximum value T2 of the data extraction period for the same preset second interval may be a fixed value. The maximum values T2 of the data extraction periods for different preset second intervals may be the same or different.

For example, if the target data extraction period is 8 minutes and the maximum value of the data extraction period T2 is 20 minutes, when the target data extraction period is increased and adjusted for the first time, the adjusted target data extraction period is 20 minutes, and thereafter, the target data extraction period is not increased and adjusted, and the target data extraction period is kept unchanged.

Based on the above manner, the target data extraction period may be increased and adjusted, but the above manner is only an example, and is not limited thereto, as long as the target data extraction period can be increased and adjusted.

In one possible implementation, the data to be analyzed may include a picture to be analyzed, and in the process of obtaining the picture to be analyzed based on the target data extraction period, the storage device may configure the target data extraction period to the front-end device, so that the front-end device obtains the picture to be analyzed from the continuous video images based on the target data extraction period. Based on this, the storage device can receive the picture to be analyzed sent by the front-end device.

For example, if the front-end device supports a function of acquiring a picture to be analyzed (which may also be referred to as a picture timing extraction function), that is, supports extracting a frame of video image from consecutive video images and converting the extracted frame of video image into a picture to be analyzed, the storage device may configure a target data extraction period to the front-end device after obtaining the target data extraction period.

The front-end device acquires a picture to be analyzed from a continuous video image (i.e., a video stream) based on the target data extraction period. For example, one frame of video image is extracted from consecutive video images in a target data extraction period, and the extracted one frame of video image is converted into a picture to be analyzed.

After the picture to be analyzed is obtained, the front-end device sends the picture to be analyzed to the storage device, obviously, as one frame of video image is selected from the video stream of the target data extraction period and is converted into the picture to be analyzed, the number of the picture to be analyzed sent to the storage device is small, and thus the processing resource of the storage device is saved.

The storage device receives the picture to be analyzed sent by the front-end device, intelligently analyzes the picture to be analyzed to obtain an intelligent analysis result return value, and decides whether to adjust the target data extraction period according to the intelligent analysis result return value. If the target data extraction period is adjusted, the storage device may further configure the adjusted target data extraction period to the front-end device, so that the front-end device extracts one frame of video image from the continuous video images based on the adjusted target data extraction period, converts the extracted one frame of video image into a picture to be analyzed, and so on.

In another possible implementation, the data to be analyzed may include a picture to be analyzed, and in the process of obtaining the picture to be analyzed based on the target data extraction period, the storage device may decode a video coding bitstream sent by the front-end device to obtain a continuous video image. Based on this, the storage device acquires one frame of video image from the continuous video image based on the target data extraction period, and converts the extracted one frame of video image into a picture to be analyzed.

For example, when the front-end device does not support the function of acquiring the picture to be analyzed, the storage device does not configure the target data extraction period to the front-end device. Based on the method, the front-end equipment collects continuous video images, encodes the video images to obtain a video coding bit stream, and sends the video coding bit stream to the storage equipment. After receiving the video coding bit stream, the storage device decodes the video coding bit stream to obtain continuous video images. Then, the storage device extracts one frame of video image from the continuous video images based on the target data extraction period, and converts the extracted one frame of video image into a picture to be analyzed.

After the picture to be analyzed is obtained, the picture to be analyzed can be intelligently analyzed to obtain an intelligent analysis result return value, and whether the target data extraction period is adjusted or not is decided according to the intelligent analysis result return value.

In the above embodiment, if the data to be analyzed is the picture to be analyzed, the process of intelligently analyzing the picture to be analyzed to obtain the return value of the intelligent analysis result may include, but is not limited to, the following cases:

in the first situation, the picture to be analyzed is intelligently analyzed based on a deep learning algorithm, and an intelligent analysis result return value is obtained. For example, a deep learning algorithm may be deployed on an intelligent analysis chip of a storage device, and the type of the deep learning algorithm is not limited. The picture to be analyzed can be provided for the intelligent analysis chip, so that the intelligent analysis chip can carry out intelligent analysis on the picture to be analyzed based on the deep learning algorithm to obtain an intelligent analysis result return value, the intelligent analysis process is not limited, and the intelligent analysis process is related to the type of the deep learning algorithm.

And in the second situation, selecting a target sub-picture from the pictures to be analyzed, and intelligently analyzing the target sub-picture based on a deep learning algorithm to obtain an intelligent analysis result return value. For example, a deep learning algorithm is deployed on an intelligent analysis chip, a target sub-picture is provided for the intelligent analysis chip, and the intelligent analysis chip performs intelligent analysis on the target sub-picture based on the deep learning algorithm to obtain an intelligent analysis result return value; or, the picture to be analyzed is provided for the intelligent analysis chip, the intelligent analysis chip selects the target sub-picture from the picture to be analyzed, and the target sub-picture is intelligently analyzed based on the deep learning algorithm to obtain an intelligent analysis result return value.

Illustratively, target extraction may be performed on the picture to be analyzed according to a predetermined target type, so as to obtain a target sub-picture. For example, a preset face in the picture to be analyzed is subjected to target extraction, so as to obtain a target sub-picture including the preset face. For another example, a target extraction is performed on a preset human body in the picture to be analyzed, so as to obtain a target sub-picture including the preset human body. For another example, the target extraction is performed on the preset vehicle in the picture to be analyzed, so as to obtain a target sub-picture including the preset vehicle. For another example, a target extraction is performed on a preset trash can in the picture to be analyzed, so as to obtain a target sub-picture including the preset trash can. Of course, the above is merely an example, and no limitation is made thereto. Obviously, by analyzing the target sub-picture, the amount of data analyzed can be reduced.

For example, a designated target on the picture to be analyzed may be filtered to obtain a target sub-picture. For example, assuming that the picture to be analyzed includes a human body and a vehicle, a sub-picture 1 including the human body may be filtered from the picture to be analyzed, and the sub-picture 1 is taken as a target sub-picture; or filtering the sub-picture 2 including the vehicle from the picture to be analyzed, and taking the remaining sub-pictures except the sub-picture 2 in the picture to be analyzed as target sub-pictures.

For example, a sub-picture of a designated coordinate region may be selected from the picture to be analyzed, and the sub-picture may be used as a target sub-picture. For example, a plurality of coordinate regions may be set in advance, and the sub-pictures of the coordinate regions need to be analyzed, and the sub-pictures outside the coordinate regions need not be analyzed. Based on this, sub-pictures of these coordinate regions can be selected from the picture to be analyzed, these sub-pictures being the target sub-pictures.

Of course, the above is only an example of selecting the target sub-picture from the picture to be analyzed, and the method is not limited thereto.

And in the third case, a time period can be set, intelligent analysis is carried out in the time period, for example, face analysis is carried out at 8-10 am, vehicle analysis is carried out at 13-15 pm, and no analysis is carried out in the rest time. Based on the above, at 8-10 am, the intelligent analysis based on the face detection can be performed on the picture to be analyzed based on the deep learning algorithm, so as to obtain an intelligent analysis result return value. At 13-15 pm, intelligent analysis based on vehicle detection can be performed on the picture to be analyzed based on a deep learning algorithm, and an intelligent analysis result return value is obtained.

In the above embodiments, the intelligent analysis result return value may include, but is not limited to: the number of detected objects, such as the number of people and the number of vehicles. The value of the detected object, such as score, confidence, position information, etc., for example, the score value of a human face. Alternatively, the type of the detected object, such as various results of the detected object's attributes, for example, the age attribute may be child, young, middle-aged, old, etc. Or the state of the detected object, such as the state that the illegal parking exists or the state that the illegal parking does not exist, the state that the garbage bin is full or the state that the garbage bin is not full, and the like. Of course, the above are just a few examples.

The above-mentioned period adjustment method is described below with reference to specific application scenarios.

When the front-end device is added to the storage device, it needs to be determined whether the front-end device supports a function of extracting a picture to be analyzed (which may also be referred to as a picture timing extraction function), if so, the period adjustment procedure shown in fig. 3A may be adopted, and if not, the period adjustment procedure shown in fig. 3B may be adopted.

Referring to fig. 3A, a schematic diagram of a period adjustment process is shown, where the period adjustment process may include:

in step 311, the storage device configures a deep learning algorithm, an initial data extraction period (taking the initial data extraction period as a target data extraction period), a minimum value T1 of the data extraction period, a maximum value T2 of the data extraction period, a value N, a value M, a preset first interval, a preset second interval, and a preset third interval.

In step 312, the storage device configures the target data extraction period to the front-end device.

Step 313, the front-end device extracts a frame of video image from the continuous video images collected by the front-end device based on the target data extraction period, converts the extracted frame of video image into a picture to be analyzed, and sends the picture to be analyzed to the storage device.

In step 314, the storage device receives the picture to be analyzed.

And 315, the storage device intelligently analyzes the picture to be analyzed to obtain an intelligent analysis result return value.

And step 316, the storage device judges whether the target data extraction period needs to be adjusted according to the return value of the intelligent analysis result. If yes, go to step 317; if not, waiting to receive the next picture to be analyzed.

Step 317, adjusting the target data extraction period, updating the adjusted data extraction period to the target data extraction period, configuring the target data extraction period to the front-end device, and returning to step 313.

For example, the steps 315 to 317 may be implemented based on parameters such as a deep learning algorithm, a minimum value T1 of a data extraction period, a maximum value T2 of the data extraction period, a value N, a value M, a preset first interval, a preset second interval, and a preset third interval.

Referring to fig. 3B, a schematic diagram of a period adjustment process is shown, where the period adjustment process may include:

in step 321, the storage device is configured with a deep learning algorithm, an initial data extraction period (taking the initial data extraction period as a target data extraction period), a minimum value T1 of the data extraction period, a maximum value T2 of the data extraction period, a value N, a value M, a preset first interval, a preset second interval and a preset third interval.

And 322, the front-end equipment collects the video image and codes the video image to obtain a video coding bit stream.

Step 323, the front-end device sends the video coded bit stream to the storage device, and the storage device decodes the video coded bit stream after receiving the video coded bit stream to obtain a continuous video image.

In step 324, the storage device extracts a frame of video image from the continuous video image based on the target data extraction period, and converts the extracted frame of video image into a picture to be analyzed.

Step 325, the storage device intelligently analyzes the picture to be analyzed to obtain a return value of the intelligent analysis result.

Step 326, the storage device determines whether the target data extraction period needs to be adjusted according to the return value of the intelligent analysis result. If so, go to step 327; if not, return to step 324.

Step 327, adjusting the target data extraction period, updating the adjusted data extraction period to the target data extraction period, and returning to step 324 based on the adjusted target data extraction period.

For example, the steps 325 to 327 may be implemented based on parameters such as a deep learning algorithm, a minimum value T1 of a data extraction period, a maximum value T2 of the data extraction period, a value N, a value M, a preset first interval, a preset second interval, and a preset third interval.

In the application scenario of crowd accumulation detection, the above process may be adopted to monitor the number of people in a specified area, and alarm when the number of people exceeds a set threshold (e.g. 300). When the number of people is less than 100, no alarm is needed, and the target data extraction period can be increased until the target data extraction period is the maximum value T2 of the data extraction period. When the number of people is between 100 and 300, attention needs to be started, but an alarm is not needed, and the target data extraction period can be kept unchanged. When the number of people is more than 300, the alarm needs to be given with emphasis, and the target data extraction period can be reduced until the target data extraction period is the minimum value T1 of the data extraction period.

In the application scene that whether detection fire control passageway is occupied, can adopt above-mentioned flow, whether detection fire control passageway is occupied, if occupied then need report to the police. When the fire fighting access is not occupied, no alarm is needed, and the target data extraction period can be increased until the target data extraction period is the maximum value of the data extraction period T2. When the fire fighting access is occupied, the alarm needs to be given out in a key mode, and the target data extraction period can be reduced until the target data extraction period is the minimum value T1 of the data extraction period.

Based on the same application concept as the method, the embodiment of the present application further provides a period adjustment apparatus, as shown in fig. 4, which is a structural diagram of the period adjustment apparatus, and the apparatus includes:

an obtaining module 41, configured to obtain data to be analyzed based on a target data extraction cycle;

the analysis module 42 is configured to perform intelligent analysis on the data to be analyzed to obtain an intelligent analysis result return value;

and the processing module 43 is configured to adjust the target data extraction period if it is determined that the target data extraction period needs to be adjusted according to the intelligent analysis result return value, and update the adjusted data extraction period to the target data extraction period.

The processing module 43 is specifically configured to: if the return values of the intelligent analysis results of the N continuous data to be analyzed are the return values of the intelligent analysis results concerned by the user, reducing and adjusting the target data extraction period; or if the intelligent analysis result return values of the M continuous data to be analyzed are the intelligent analysis result return values which are not concerned by the user, increasing and adjusting the target data extraction period.

The processing module 43 is specifically configured to: if the return values of the intelligent analysis results of the N continuous data to be analyzed are in a preset first interval, determining that the target data extraction period needs to be adjusted; n is a positive integer, a preset first interval is set according to an intelligent analysis result, and an interval range or an interval value of a data extraction period needs to be reduced; and if the target data extraction period is greater than the minimum value of the data extraction period, reducing and adjusting the target data extraction period, wherein the adjusted data extraction period is not less than the minimum value of the data extraction period.

The processing module 43 is specifically configured to, when performing reduction adjustment on the target data extraction period: determining a first adjustment tolerance corresponding to the preset first interval, and performing reduction adjustment on the target data extraction period according to the first adjustment tolerance; or determining a first adjustment common ratio corresponding to the preset first interval, and performing reduction adjustment on the target data extraction period according to the first adjustment common ratio; or adjusting the target data extraction period to be the minimum value of the data extraction period.

The processing module 43 is specifically configured to: if the return values of the intelligent analysis results of the continuous M data to be analyzed are in a preset second interval, determining that the target data extraction period needs to be adjusted; m is a positive integer, a preset second interval is set according to an intelligent analysis result, and an interval range or an interval value of a data extraction period needs to be increased; and if the target data extraction period is smaller than the maximum value of the data extraction period, increasing and adjusting the target data extraction period, wherein the adjusted data extraction period is not larger than the maximum value of the data extraction period.

The processing module 43 is specifically configured to, when performing increase adjustment on the target data extraction period: determining a second adjustment tolerance corresponding to the preset second interval, and performing increasing adjustment on the target data extraction period according to the second adjustment tolerance; or determining a second adjustment common ratio corresponding to the preset second interval, and performing addition adjustment on the target data extraction period according to the second adjustment common ratio; or adjusting the target data extraction period to the maximum value of the data extraction period.

The processing module 43 is further configured to: if the intelligent analysis result return value is in a preset third interval, determining that the target data extraction period does not need to be adjusted; the preset third interval is set according to an intelligent analysis result, and an interval range or an interval value which does not need to be adjusted for the data extraction period is set.

The data to be analyzed includes a picture to be analyzed, and the obtaining module 41 is specifically configured to: configuring the target data extraction period to front-end equipment so that the front-end equipment extracts a frame of video image from continuous video images based on the target data extraction period and converts the extracted frame of video image into a picture to be analyzed; receiving the picture to be analyzed sent by the front-end equipment; or, decoding the video coding bit stream sent by the front-end equipment to obtain a continuous video image; and extracting a frame of video image from the continuous video image based on the target data extraction period, and converting the extracted frame of video image into a picture to be analyzed.

The analysis module 42 is specifically configured to: if the data to be analyzed comprises the picture to be analyzed, intelligently analyzing the picture to be analyzed based on a deep learning algorithm to obtain an intelligent analysis result return value; or selecting a target sub-picture from the pictures to be analyzed, and intelligently analyzing the target sub-picture based on a deep learning algorithm to obtain an intelligent analysis result return value.

Based on the same application concept as the method, an embodiment of the present application further provides a storage device, such as a Video storage device for security monitoring, such as an nvr (network Video recorder), a storage server, and an intelligent analysis server, and a hardware architecture diagram of the storage device provided in the embodiment of the present application can be seen in fig. 5 from a hardware level. The storage device may include: a processor 51 and a machine-readable storage medium 52, the machine-readable storage medium 52 storing machine-executable instructions executable by the processor 51; the processor 51 is configured to execute machine executable instructions to implement the methods disclosed in the above examples of the present application. For example, the processor 51 is for executing machine executable instructions to implement the steps of:

acquiring data to be analyzed based on a target data extraction period;

Based on the same application concept as the method, embodiments of the present application further provide a machine-readable storage medium, where the machine-readable storage medium stores thereon several computer instructions, and when the computer instructions are executed by a CPU, the method disclosed in the above example of the present application can be implemented.

For example, the computer instructions, when executed by a processor, enable the following steps:

acquiring data to be analyzed based on a target data extraction period;

The machine-readable storage medium may be, for example, any electronic, magnetic, optical, or other physical storage device that can contain or store information such as executable instructions, data, and the like. For example, the machine-readable storage medium may be: random Access Memory (RAM), volatile Memory, non-volatile Memory, flash Memory, a storage drive (e.g., a hard drive), a solid state drive, any type of storage disk (e.g., an optical disk, a dvd, etc.), or similar storage medium, or a combination thereof.

The systems, devices, modules or units illustrated in the above embodiments may be implemented by a computer chip or an entity, or by a product with certain functions. A typical implementation device is a computer, which may take the form of a personal computer, laptop computer, cellular telephone, camera phone, smart phone, personal digital assistant, media player, navigation device, email messaging device, game console, tablet computer, wearable device, or a combination of any of these devices.

For convenience of description, the above devices are described as being divided into various units by function, and are described separately. Of course, the functionality of the units may be implemented in one or more software and/or hardware when implementing the present application.

As will be appreciated by one skilled in the art, embodiments of the present application may be provided as a method, system, or computer program product. Accordingly, the present application may take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, embodiments of the present application may take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, and the like) having computer-usable program code embodied therein.

The present application is described with reference to flowchart illustrations and/or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the application. It will be understood that each flow and/or block of the flow diagrams and/or block diagrams, and combinations of flows and/or blocks in the flow diagrams and/or block diagrams, can be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general purpose computer, special purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.

Furthermore, these computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instruction means which implement the function specified in the flowchart flow or flows and/or block diagram block or blocks.

These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.

The above description is only an example of the present application and is not intended to limit the present application. Various modifications and changes may occur to those skilled in the art. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application should be included in the scope of the claims of the present application.

Claims

1. A method of cycle adjustment, the method comprising:

acquiring data to be analyzed based on a target data extraction period;

2. The method of claim 1,

if it is determined that the target data extraction period needs to be adjusted according to the intelligent analysis result return value, adjusting the target data extraction period includes:

if the return values of the intelligent analysis results of the N continuous data to be analyzed are the return values of the intelligent analysis results concerned by the user, reducing and adjusting the target data extraction period; alternatively, the first and second electrodes may be,

and if the intelligent analysis result return values of the continuous M data to be analyzed are the intelligent analysis result return values which are not concerned by the user, increasing and adjusting the target data extraction period.

3. The method of claim 1,

if the return values of the intelligent analysis results of the N continuous data to be analyzed are in a preset first interval, determining that the target data extraction period needs to be adjusted; n is a positive integer, a preset first interval is set according to an intelligent analysis result, and an interval range or an interval value of a data extraction period needs to be reduced;

and if the target data extraction period is greater than the minimum value of the data extraction period, reducing and adjusting the target data extraction period, wherein the adjusted data extraction period is not less than the minimum value of the data extraction period.

4. The method according to claim 2 or 3,

the performing reduction adjustment on the target data extraction period includes:

determining a first adjustment tolerance corresponding to the preset first interval, and performing reduction adjustment on the target data extraction period according to the first adjustment tolerance; alternatively, the first and second electrodes may be,

determining a first adjustment common ratio corresponding to the preset first interval, and performing reduction adjustment on the target data extraction period according to the first adjustment common ratio; alternatively, the first and second electrodes may be,

and adjusting the target data extraction period to be the minimum value of the data extraction period.

5. The method of claim 1,

if the return values of the intelligent analysis results of the continuous M data to be analyzed are in a preset second interval, determining that the target data extraction period needs to be adjusted; wherein M is a positive integer, the preset second interval is set according to the intelligent analysis result, and the interval range or the interval value of the data extraction period needs to be increased;

and if the target data extraction period is smaller than the maximum value of the data extraction period, increasing and adjusting the target data extraction period, wherein the adjusted data extraction period is not larger than the maximum value of the data extraction period.

6. The method according to claim 2 or 5,

the increasing and adjusting the target data extraction period comprises:

determining a second adjustment tolerance corresponding to the preset second interval, and performing increasing adjustment on the target data extraction period according to the second adjustment tolerance; alternatively, the first and second electrodes may be,

determining a second adjustment common ratio corresponding to the preset second interval, and performing additional adjustment on the target data extraction period according to the second adjustment common ratio; alternatively, the first and second electrodes may be,

and adjusting the target data extraction period to be the maximum value of the data extraction period.

7. The method of claim 1, wherein after performing the intelligent analysis on the data to be analyzed to obtain an intelligent analysis result return value, the method further comprises:

if the intelligent analysis result return value is in a preset third interval, determining that the target data extraction period does not need to be adjusted; the preset third interval is set according to an intelligent analysis result, and an interval range or an interval value which does not need to be adjusted for the data extraction period is set.

8. The method according to any one of claims 1-3, 5, and 7, wherein the data to be analyzed comprises a picture to be analyzed, and the obtaining the data to be analyzed based on the target data extraction period comprises:

configuring the target data extraction period to front-end equipment so that the front-end equipment extracts a frame of video image from continuous video images based on the target data extraction period and converts the extracted frame of video image into a picture to be analyzed;

receiving the picture to be analyzed sent by the front-end equipment; alternatively, the first and second electrodes may be,

decoding a video coding bit stream sent by front-end equipment to obtain a continuous video image;

and extracting a frame of video image from the continuous video image based on the target data extraction period, and converting the extracted frame of video image into a picture to be analyzed.

9. The method according to any one of claims 1 to 3, 5 and 7,

the intelligent analysis of the data to be analyzed to obtain an intelligent analysis result return value comprises the following steps:

if the data to be analyzed comprises the picture to be analyzed, intelligently analyzing the picture to be analyzed based on a deep learning algorithm to obtain an intelligent analysis result return value; alternatively, the first and second electrodes may be,

and selecting a target sub-picture from the pictures to be analyzed, and intelligently analyzing the target sub-picture based on a deep learning algorithm to obtain an intelligent analysis result return value.

10. A cycle adjusting apparatus, characterized in that the apparatus comprises:

11. A storage device, comprising: a processor and a machine-readable storage medium storing machine-executable instructions executable by the processor;

acquiring data to be analyzed based on a target data extraction period;