Video monitoring system based on AI

Abstract

The invention belongs to the technical field of video monitoring, and particularly relates to an AI-based video monitoring system, which comprises a monitoring module used for shooting an environment to be monitored; the image analysis module is used for carrying out dangerous action analysis on the content shot by the monitoring module, and if the dangerous action is contained, the person forming the dangerous action is locked; the sound analysis module is used for performing language and voice analysis on the content shot by the monitoring module, and if the dangerous voice is contained, the person forming the dangerous voice is locked; and the control module is used for controlling the monitoring module according to the figure locked by the action analysis module and the figure locked by the sound analysis module to realize tracking monitoring. The sound analysis module used for performing language and voice analysis on the content shot by the monitoring module is arranged, so that when voice recognition is performed, the voice recognition efficiency is improved, and the voice recognition efficiency is improved. The identification accuracy of the AI-based video monitoring system can be improved.

Description

Technical Field

[0001] The present invention relates to the field of video surveillance technology, and in particular to an AI-based video surveillance system. Background Art

[0002] Video surveillance is an important part of the security system. The traditional monitoring system includes front-end cameras, transmission cables and video monitoring platforms. Video surveillance is an efficient monitoring technology with the characteristics of real-time, reliability, intuitiveness, etc. It is easy to use and has attracted attention from all walks of life. Using AI for video surveillance is a video analysis technology that integrates artificial intelligence technology. It aims to automatically identify, analyze and understand the content in the surveillance video. By using machine learning and deep learning algorithms, it can identify people, vehicles, objects and specific behaviors or events, thereby greatly improving monitoring efficiency and accuracy.

[0003] Although the existing AI-based video surveillance system can recognize the monitored sound, the currently recognized sound is usually Mandarin. If the monitored sound is in other languages ​​(including but not limited to English and dialects), it may not be recognized. As a result, when recognizing the sound, the recognition accuracy is low, which may affect the AI-based video surveillance effect. Therefore, an AI-based video surveillance system is invented. Summary of the invention

[0004] According to one aspect of the present invention, the present invention provides the following technical solution:

[0005] An AI-based video surveillance system, comprising:

[0006] A monitoring module, used to shoot the monitored environment;

[0007] The image analysis module is used to analyze the dangerous actions of the content captured by the monitoring module. If there are dangerous actions, the person who performs the dangerous actions will be locked;

[0008] The sound analysis module is used to analyze the language and voice of the content captured by the monitoring module. If there is any dangerous voice, the person who makes the dangerous voice will be locked;

[0009] The control module is used to control the monitoring module according to the person locked by the action analysis module and the person locked by the sound analysis module to achieve tracking monitoring.

[0010] As a preferred solution of the AI-based video surveillance system described in the present invention, the image analysis module includes:

[0011] An image acquisition module is used to acquire the image content taken by the monitoring module;

[0012] The motion analysis module is used to perform motion analysis on the images collected by the image acquisition module to analyze whether there are any dangerous actions in the images collected by the image acquisition module;

[0013] The feature extraction module is used to lock the person who performs the dangerous action according to the person's characteristics when the action analyzed by the action analysis module contains dangerous actions.

[0014] As a preferred solution of the AI-based video surveillance system described in the present invention, the action analysis module includes:

[0015] An action acquisition module is used to extract actions from the images collected by the image acquisition module;

[0016] The action storage module is used to store various dangerous actions.

[0017] As a preferred solution of the AI-based video surveillance system described in the present invention, the action analysis module further includes:

[0018] An action comparison module, used to compare the action extracted by the action acquisition module with the action stored in the action storage module;

[0019] The action judgment module is used to judge whether the action extracted by the action acquisition module is similar to that in the action storage module. If so, it means that the action extracted by the action acquisition module is a dangerous action.

[0020] As a preferred solution of the AI-based video surveillance system described in the present invention, the sound analysis module includes:

[0021] A sound collection module is used to collect the sound content monitored by the monitoring module;

[0022] A language analysis module is used to analyze the sound content collected by the sound collection module to analyze the language of the sound;

[0023] A speech analysis module, used to perform speech analysis corresponding to the language analyzed by the language analysis module, so as to analyze whether the sound content collected by the sound collection module contains dangerous speech;

[0024] The sound source analysis module is used to lock the person who makes the dangerous speech according to the position of the sound when the speech analyzed by the speech analysis module contains dangerous speech.

[0025] As a preferred solution of the AI-based video surveillance system described in the present invention, there are several speech analysis modules, so that each language corresponds to a speech analysis module.

[0026] As a preferred solution of the AI-based video surveillance system described in the present invention, the language analysis module includes:

[0027] Language storage module, used to store various languages;

[0028] A language comparison module, used to compare the sound collected by the sound collection module with the sound stored in the language storage module;

[0029] The language determination module is used to determine whether the sound collected by the sound collection module is similar to that in the language storage module. If so, the corresponding language will be found.

[0030] As a preferred solution of the AI-based video surveillance system described in the present invention, the speech analysis module includes:

[0031] A voice acquisition module is used to extract the voice from the sound collected by the sound collection module;

[0032] The voice storage module is used to store various dangerous voices.

[0033] As a preferred solution of the AI-based video surveillance system described in the present invention, the speech analysis module further includes:

[0034] A voice comparison module, used to compare the voice extracted by the voice acquisition module with the voice stored in the voice storage module;

[0035] The voice judgment module is used to judge whether the voice extracted by the voice acquisition module is similar to that in the voice storage module. If so, it means that the voice extracted by the voice acquisition module is a dangerous voice.

[0036] Compared with existing technologies:

[0037] By providing a sound analysis module for performing language and voice analysis on the content captured by the monitoring module, it is possible to improve the recognition accuracy of the AI-based video monitoring system when performing sound recognition; in addition, by providing an image analysis module for performing dangerous action analysis on the content captured by the monitoring module and a control module for controlling the monitoring module according to the person locked by the action analysis module and the person locked by the sound analysis module, it is possible to track and monitor dangerous persons so that personnel can understand their movements. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] Figure 1 It is a schematic diagram of the process of the present invention. DETAILED DESCRIPTION

[0039] To make the objectives, technical solutions and advantages of the present invention more clear, the embodiments of the present invention will be further described in detail below with reference to the accompanying drawings.

[0040] The present invention provides a video surveillance system based on AI, please refer to Figure 1 ;

[0041] It includes: a monitoring module, which is used to shoot the monitored environment; an image analysis module, which is used to analyze the content shot by the monitoring module for dangerous actions. If there are dangerous actions, the person who makes the dangerous actions will be locked; a sound analysis module, which is used to analyze the language and voice of the content shot by the monitoring module. If there are dangerous voices, the person who makes the dangerous voices will be locked; and a control module, which is used to control the monitoring module according to the person locked by the action analysis module and the person locked by the sound analysis module to achieve tracking monitoring.

[0042] The image analysis module includes: an image acquisition module, which is used to acquire the content of the image taken by the monitoring module; a motion analysis module, which is used to perform motion analysis on the image acquired by the image acquisition module to analyze whether there are dangerous actions in the image acquired by the image acquisition module; and a feature extraction module, which is used to lock the person who performs the dangerous action according to the person's characteristics when the action analyzed by the action analysis module contains dangerous actions.

[0043] The action analysis module includes: an action acquisition module, which is used to extract actions from the images collected by the image acquisition module; an action storage module, which is used to store various dangerous actions; an action comparison module, which is used to compare the actions extracted by the action acquisition module with the actions stored in the action storage module; and an action judgment module, which is used to judge whether the actions extracted by the action acquisition module are similar to those in the action storage module. If so, it means that the actions extracted by the action acquisition module are dangerous actions.

[0044] The sound analysis module includes: a sound collection module, which is used to collect the sound content monitored by the monitoring module; a language analysis module, which is used to analyze the sound content collected by the sound collection module to analyze the language of the sound; a speech analysis module, which is used to perform corresponding speech analysis according to the language analyzed by the language analysis module to analyze whether the sound content collected by the sound collection module contains dangerous speech; a sound source analysis module, which is used to lock the person who makes the dangerous speech according to the position of the sound when the speech analyzed by the speech analysis module contains dangerous speech; a plurality of speech analysis modules are provided, so that each language corresponds to a speech analysis module.

[0045] The language analysis module includes: a language storage module for storing various languages; a language comparison module for comparing the sound collected by the sound collection module with the sound stored in the language storage module; and a language judgment module for judging whether the sound collected by the sound collection module is similar to the sound in the language storage module. If so, the corresponding language will be found.

[0046] The speech analysis module includes: a speech acquisition module, which is used to extract speech from the sound collected by the sound collection module; a speech storage module, which is used to store various dangerous speech; a speech comparison module, which is used to compare the speech extracted by the speech acquisition module with the speech stored in the speech storage module; and a speech judgment module, which is used to judge whether the speech extracted by the speech acquisition module is similar to the speech in the speech storage module. If so, it means that the speech extracted by the speech acquisition module is a dangerous speech.

[0047] When used, the specific steps are as follows:

[0048] Step 1: Use the monitoring module to shoot the monitored environment;

[0049] Step 2: The image content captured by the monitoring module is collected by the image acquisition module. After the collection, the action in the image collected by the image acquisition module is extracted by the action acquisition module. After the extraction, the action extracted by the action acquisition module is compared with the action stored in the action storage module by the action comparison module. After the comparison, the action judgment module is used to judge whether the action extracted by the action acquisition module is similar to the action storage module. If there is a similarity, it means that the action extracted by the action acquisition module is a dangerous action. At this time, when the action analyzed by the action analysis module contains a dangerous action, the feature extraction module can lock the person who forms the dangerous action according to the person's characteristics;

[0050] Step 3: The sound content monitored by the monitoring module is collected by the sound collection module. After the collection, the sound collected by the sound collection module is compared with the sound stored in the language storage module through the language comparison module. After the comparison, the language judgment module is used to determine whether the sound collected by the sound collection module is similar to the sound stored in the language storage module. If there is a similarity, the corresponding language will be found. Then, the voice in the sound collected by the sound collection module is extracted by the voice acquisition module. After the extraction, the voice extracted by the voice acquisition module is compared with the voice stored in the voice storage module through the voice comparison module. After the comparison, the voice judgment module is used to determine whether the voice extracted by the voice acquisition module is similar to the voice storage module. If there is a similarity, it means that the voice extracted by the voice acquisition module is a dangerous voice. At this time, when the voice analyzed by the voice analysis module contains dangerous voice, the sound source analysis module can lock the person who forms the dangerous voice according to the position of the voice;

[0051] Step 4: The control module controls the monitoring module according to the person locked by the action analysis module and the person locked by the sound analysis module to achieve tracking monitoring.

[0052] Based on the above, the present invention monitors a certain community, and when a stabbing incident occurs in the community, the monitoring module will shoot a video of the stabbing. After shooting, the image of the stabbing shot by the monitoring module will be collected by the image acquisition module. After collection, the stabbing action in the stabbing image collected by the image acquisition module will be extracted by the action acquisition module. After extraction, the stabbing action extracted by the action acquisition module will be compared with the action stored in the action storage module through the action comparison module. After comparison, the action judgment module will be used to judge whether the stabbing action extracted by the action acquisition module is similar to that in the action storage module. If there is a similarity, it means that the stabbing action extracted by the action acquisition module is a dangerous action. At this time, the wearing features and appearance features of the dangerous person will be extracted by the feature extraction module to lock the dangerous person. At the same time, the sound content in the stabbing video will be collected by the sound collection module. Suppose the dangerous person said "stabbing you to death" in Cantonese when stabbing someone, at this time, the "stabbing you to death" sound collected by the sound collection module will be compared with the language comparison module through the language comparison module. After the comparison, the language judgment module will be used to judge whether the sound of "Cut you to death" collected by the sound collection module is similar to that in the language storage module. If so, the corresponding language will be found, that is, the collected sound is Cantonese. Then, the voice in the "Cut you to death" sound will be extracted by the voice acquisition module. After the extraction, the "Cut you to death" voice extracted by the voice acquisition module will be compared with the voice stored in the voice storage module through the voice comparison module. After the comparison, the voice judgment module will be used to judge whether the "Cut you to death" voice extracted by the voice acquisition module is similar to that in the voice storage module. If so, it means that the "Cut you to death" voice extracted by the voice acquisition module is a dangerous voice. At this time, the sound source analysis module will be used to analyze the position of the dangerous person in the community, so that the person who forms the dangerous voice can be locked according to the position of the sound; so far, the control module can control the monitoring module according to the basic characteristics of the dangerous person (including but not limited to wearing characteristics, appearance characteristics and voice characteristics) to achieve tracking monitoring.

[0053] Although the present invention has been described above with reference to the embodiments, various modifications may be made thereto and parts thereof may be replaced by equivalents without departing from the scope of the present invention. In particular, as long as there is no structural conflict, the various features in the embodiments disclosed in the present invention may be used in combination with each other in any manner, and the fact that these combinations are not exhaustively described in this specification is only for the sake of omitting space and saving resources. Therefore, the present invention is not limited to the specific embodiments disclosed herein, but includes all technical solutions falling within the scope of the claims.

Claims

1. An AI-based video surveillance system, characterized in that: include: A monitoring module, used to shoot the monitored environment; The image analysis module is used to analyze the dangerous actions of the content captured by the monitoring module. If there are dangerous actions, the person who performs the dangerous actions will be locked; The sound analysis module is used to analyze the language and voice of the content captured by the monitoring module. If there is any dangerous voice, the person who makes the dangerous voice will be locked; A control module, used to control the monitoring module according to the person locked by the action analysis module and the person locked by the sound analysis module, so as to realize tracking monitoring; The sound analysis module comprises: A sound collection module is used to collect the sound content monitored by the monitoring module; A language analysis module is used to analyze the sound content collected by the sound collection module to analyze the language of the sound; A speech analysis module, used to perform speech analysis corresponding to the language analyzed by the language analysis module, so as to analyze whether the sound content collected by the sound collection module contains dangerous speech; A sound source analysis module, which is used to identify the person who makes the dangerous speech according to the location of the sound when the speech analyzed by the speech analysis module contains dangerous speech; The speech analysis module comprises: A voice acquisition module is used to extract the voice from the sound collected by the sound collection module; Voice storage module, used to store various dangerous voices; The speech analysis module also includes: A voice comparison module, used to compare the voice extracted by the voice acquisition module with the voice stored in the voice storage module; The voice judgment module is used to judge whether the voice extracted by the voice acquisition module is similar to that in the voice storage module. If so, it means that the voice extracted by the voice acquisition module is a dangerous word.

2. The AI-based video surveillance system according to claim 1, characterized in that: The image analysis module comprises: An image acquisition module is used to acquire the image content taken by the monitoring module; The motion analysis module is used to perform motion analysis on the images collected by the image acquisition module to analyze whether there are any dangerous actions in the images collected by the image acquisition module; The feature extraction module is used to lock the person who performs the dangerous action according to the person's characteristics when the action analyzed by the action analysis module contains dangerous actions.

3. The AI-based video surveillance system according to claim 2, characterized in that: The action analysis module comprises: An action acquisition module is used to extract actions from the images collected by the image acquisition module; The action storage module is used to store various dangerous actions.

4. The AI-based video surveillance system according to claim 3, characterized in that: The action analysis module also includes: An action comparison module, used to compare the action extracted by the action acquisition module with the action stored in the action storage module; The action judgment module is used to judge whether the action extracted by the action acquisition module is similar to that in the action storage module. If so, it means that the action extracted by the action acquisition module is a dangerous action.

5. The AI-based video surveillance system according to claim 1, characterized in that: There are several speech analysis modules, so that each language corresponds to a speech analysis module.

6. The AI-based video surveillance system according to claim 1, characterized in that: The language analysis module includes: Language storage module, used to store various languages; A language comparison module, used to compare the sound collected by the sound collection module with the sound stored in the language storage module; The language determination module is used to determine whether the sound collected by the sound collection module is similar to that in the language storage module. If so, the corresponding language will be found.

Images