CN110674761B - A regional behavior early warning method and system - Google Patents
A regional behavior early warning method and systemDownload PDFInfo
- Publication number
- CN110674761B CN110674761BCN201910921181.4ACN201910921181ACN110674761BCN 110674761 BCN110674761 BCN 110674761BCN 201910921181 ACN201910921181 ACN 201910921181ACN 110674761 BCN110674761 BCN 110674761B
- Authority
- CN
- China
- Prior art keywords
- information
- character
- future
- behavior
- scene
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Images
Classifications
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
- G06V40/10—Human or animal bodies, e.g. vehicle occupants or pedestrians; Body parts, e.g. hands
- G06V40/16—Human faces, e.g. facial parts, sketches or expressions
- G06V40/161—Detection; Localisation; Normalisation
- G06V40/166—Detection; Localisation; Normalisation using acquisition arrangements
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/049—Temporal neural networks, e.g. delay elements, oscillating neurons or pulsed inputs
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
- G06Q50/10—Services
- G06Q50/26—Government or public services
- G06Q50/265—Personal security, identity or safety
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/20—Scenes; Scene-specific elements in augmented reality scenes
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
- G06V40/10—Human or animal bodies, e.g. vehicle occupants or pedestrians; Body parts, e.g. hands
- G06V40/16—Human faces, e.g. facial parts, sketches or expressions
- G06V40/168—Feature extraction; Face representation
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Health & Medical Sciences (AREA)
- General Physics & Mathematics (AREA)
- General Health & Medical Sciences (AREA)
- Business, Economics & Management (AREA)
- Multimedia (AREA)
- Oral & Maxillofacial Surgery (AREA)
- Software Systems (AREA)
- Computational Linguistics (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Mathematical Physics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Biophysics (AREA)
- Tourism & Hospitality (AREA)
- Biomedical Technology (AREA)
- Human Computer Interaction (AREA)
- Artificial Intelligence (AREA)
- Evolutionary Computation (AREA)
- Computer Security & Cryptography (AREA)
- Development Economics (AREA)
- Educational Administration (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Economics (AREA)
- Human Resources & Organizations (AREA)
- Marketing (AREA)
- Primary Health Care (AREA)
- Strategic Management (AREA)
- General Business, Economics & Management (AREA)
- Image Analysis (AREA)
- Alarm Systems (AREA)
Abstract
Description
Translated fromChinese技术领域technical field
本发明涉及计算机技术领域，特别涉及一种区域行为预警方法及系统。The invention relates to the field of computer technology, in particular to a regional behavior early warning method and system.
背景技术Background technique
随着深度神经网络近年来算法优化和预测准确性上的长足进步，将深度学习应用到生活的各个领域几乎都能取得很好的效果，节省了大量人力成本。近年来随着城市居民的不断增加，区域的安防预警面临前所未有的挑战，依靠老式的监控和人为查验很难取得很好的安防效果。With the great progress in algorithm optimization and prediction accuracy of deep neural networks in recent years, applying deep learning to almost all areas of life can achieve good results and save a lot of labor costs. In recent years, with the continuous increase of urban residents, regional security early warning is facing unprecedented challenges. It is difficult to achieve good security effects by relying on old-fashioned monitoring and human inspection.
为了提高区域行为预警的效果，可以引入深度神经网络实现，具体地说，有以下几种方式实现。In order to improve the effect of regional behavior early warning, a deep neural network can be introduced. Specifically, there are several ways to achieve it.
公开号为CN109064698A的专利申请公开了一种居民安防预警系统，该系统通过人脸识别进入小区的人员是否为非本小区居民；对非本小区居民，持续获取该非本小区居民在小区内的行径轨迹及行为模式，至少包括对非本小区居民进出单元的行为确定；累加非本小区居民在设定时间段内进出小区内的单元总数，并获取非本小区居民在小区内各单元的停留时间；当非本小区居民在设定时间段内进出小区内的单元总数不小于N时，根据非本小区居民在各单元的停留时间，判断非本小区居民是否有异常行为，并确定异常行为等级，N为整数且≥3；根据不同的异常行为等级发出警报。The patent application with publication number CN109064698A discloses a resident security warning system, which uses face recognition to identify whether the person entering the community is a non-resident of the community; Behavior trajectories and behavior patterns, including at least the determination of the behavior of non-residents entering and leaving the unit; accumulating the total number of units that non-residents enter and leaving the community within the set time period, and obtain the stay of non-residents in each unit in the community Time; when the total number of units in and out of the community by non-residents within the set time period is not less than N, judge whether non-residents in the community have abnormal behavior according to the stay time of non-residents in each unit, and determine the abnormal behavior Level, N is an integer and ≥ 3; alerts are issued according to different levels of abnormal behavior.
公开号为CN109189078A的专利申请公开了一种基于深度增强学习的家用安全防护机器人及方法，通过深度学习分享目标人物的行为状态信息，目标人物的位置及所处环境的障碍物位置，输出目标人物的运行轨迹，从而对目标人物实现锁定追踪。The patent application with publication number CN109189078A discloses a home security protection robot and method based on deep reinforcement learning. Through deep learning, it shares the behavior status information of the target person, the position of the target person and the location of obstacles in the environment, and outputs the target person. The running trajectory of the target person can be locked and tracked.
公开号为CN105975633A的专利申请公开了一种关于运动轨迹的获取方法及装置。根据当前地理位置及预设的人物信息对应的预设人物的运动状态预测预设人物的运动轨迹；开启运动轨迹上的目标拍摄设备。The patent application with the publication number CN105975633A discloses a method and device for acquiring a motion trajectory. The movement trajectory of the preset character is predicted according to the current geographic location and the movement state of the preset character corresponding to the preset character information; and the target shooting device on the movement trajectory is turned on.
公开号为CN108877121A的专利申请公开了一种基于云平台的人工智能预警系统，主要通过设于小区入口处的人脸采集模块，用于对出入小区的人脸进行采集，并形成人脸识别信息发送云服务器，通过与犯罪人员的照片进行比对，形成警示信号。The patent application with publication number CN108877121A discloses an artificial intelligence early warning system based on a cloud platform, which is mainly used to collect the faces entering and exiting the community through a face collection module located at the entrance of the community, and form face recognition information. Send the cloud server to form a warning signal by comparing it with the photos of the criminals.
可以看出，上述方案在实现区域行为预警时，此处区域以小区为例，多为在诸如小区门禁处等位置设置人脸识别单元，从而通过人脸识别确定进入小区人员的身份，或通过云平台与犯罪人员的照片进行比对，实现预警。但是相关人员一旦进入到小区内部，就只能通过设置的传统监控摄像记录其行踪，通过人为查验或简单的统计学方法判断异常行为，进行预警。人为查验视频监控要耗费大量人力成本，且在人员换岗，精神不集中时无法做到持续的监控，而采用统计学方法查验视频监控则并不准确且统计的标准数值难以确定。无论采用上述哪一种方式，都会使得预警准确率很低。It can be seen that when the above scheme realizes regional behavior early warning, the area here is taken as an example, and most of them are to set up face recognition units at locations such as the gate of the community, so as to determine the identities of people entering the community through face recognition, or through face recognition. The cloud platform compares the photos of criminals to realize early warning. However, once the relevant personnel enter the community, they can only record their whereabouts through the traditional surveillance cameras set up, and judge abnormal behaviors and give early warnings through human inspection or simple statistical methods. Manual inspection of video surveillance requires a lot of labor costs, and continuous monitoring cannot be achieved when personnel change posts and lack of concentration, while using statistical methods to inspect video surveillance is inaccurate and the statistical standard value is difficult to determine. No matter which of the above methods is adopted, the accuracy of early warning will be very low.
因此，如何在提高预警准确度及改善预警效果的基础上，实现区域行为预警，是一个亟待解决的问题。Therefore, how to realize regional behavior early warning on the basis of improving early warning accuracy and improving early warning effect is an urgent problem to be solved.
发明内容SUMMARY OF THE INVENTION
有鉴于此，本发明实施例提供一种区域行为预警方法，该方法能够提高预警准确度，改善预警效果。In view of this, the embodiment of the present invention provides a regional behavior early warning method, which can improve the early warning accuracy and improve the early warning effect.
本发明实施例还提供一种区域行为预警系统，该系统能够提高预警准确度，改善预警效果。The embodiment of the present invention also provides a regional behavior early warning system, which can improve the early warning accuracy and improve the early warning effect.
本发明实施例是这样实现的：The embodiments of the present invention are implemented as follows:
一种区域行为预警方法，包括：A regional behavioral early warning method comprising:
将在区域内采集的人物图像信息输入到设置的第一神经网络中，输出得到在区域内的场景信息、人物肢体信息、人物与场景之间的关系信息、及人物与场景中的对象之间的关系信息；Input the character image information collected in the area into the set first neural network, and output the scene information in the area, the body information of the character, the relationship information between the character and the scene, and the relationship between the character and the objects in the scene. relationship information;
将在区域内的场景信息、人物肢体信息、人物与场景之间的关系信息、及人物与场景中的对象之间的关系信息抽象为一个信息张量；Abstract the scene information, character body information, relationship information between characters and scenes, and relationship information between characters and objects in the scene into an information tensor in the area;
将所述抽象的信息张量输入到设置的记忆网络模型和设置的第二神经网络中，以使所述记忆网络模型输出得到人物的未来运动轨迹信息，所述第二神经网络输出人物的未来活动标签和未来目的位置信息；The abstract information tensor is input into the set memory network model and the set second neural network, so that the memory network model output obtains the character's future motion trajectory information, and the second neural network outputs the character's future trajectory information. activity tags and future destination location information;
根据输出的人物的未来运动轨迹信息、人物的未来活动标签和未来目的信息，判定该人物行为是否可疑。According to the outputted future motion track information of the character, the character's future activity label and future purpose information, it is determined whether the character's behavior is suspicious.
所述在区域内采集人物图像信息之前，该方法还包括：Before collecting the person image information in the area, the method further includes:
采用人脸及身份识别方式对出现在所设置的监控摄像中的人物进行识别，判定该人物的身份信息，所述身份信息包括内部人员或外部人员。The person appearing in the set surveillance camera is identified by means of face and identity recognition, and the identity information of the person is determined, and the identity information includes internal personnel or external personnel.
所述判定该人物行为是否可疑还包括：The judging whether the behavior of the character is suspicious also includes:
根据输出的人物的未来运动轨迹信息、人物的未来活动标签、未来目的信息及该人物的身份信息判定该人物行为是否可疑。Whether the behavior of the character is suspicious is determined according to the outputted future motion track information of the character, the character's future activity label, future purpose information and the character's identity information.
所述将抽象的信息张量输入到设置的记忆网络模型中，输出得到人物的未来运动轨迹信息为：The abstract information tensor is input into the set memory network model, and the output of the character's future motion trajectory information is:
将所述抽象的信息张量进行记忆网络模型的压缩编码，得到视觉特征张量后，采用记忆网络模型进行解码，在实际坐标平面上输出人物的未来运动轨迹特征坐标信息。The abstract information tensor is compressed and encoded by the memory network model, and after the visual feature tensor is obtained, the memory network model is used for decoding, and the feature coordinate information of the character's future motion trajectory is output on the actual coordinate plane.
所述第一神经网络实时更新；所述记忆网络模型实时更新；所述第二神经网络实时更新。The first neural network is updated in real time; the memory network model is updated in real time; the second neural network is updated in real time.
所述方法还包括：The method also includes:
如果判定该人物行为可疑，进行预警。If it is determined that the person's behavior is suspicious, an early warning is issued.
一种区域行为预警系统，包括：人物行为模块、人物交互模块、轨迹生成模块及活动预测模块，其中，A regional behavior early warning system, comprising: a character behavior module, a character interaction module, a trajectory generation module and an activity prediction module, wherein,
人物行为模块，用于设置第一神经网络、设置记忆网络模型及第二神经网络；The character behavior module is used for setting the first neural network, setting the memory network model and the second neural network;
人物交互模块，用于将在区域内采集的人物图像信息输入到设置的第一神经网络中，输出得到在区域内的场景信息、人物肢体信息、人物与场景之间的关系信息、及人物与场景中的对象之间的关系信息，将在区域内的场景信息、人物肢体信息、人物与场景之间的关系信息、及人物与场景中的对象之间的关系信息抽象为一个信息张量；The character interaction module is used to input the image information of the characters collected in the area into the set first neural network, and output the scene information in the area, the body information of the characters, the relationship information between the characters and the scene, and the relationship between the characters and the characters. The relationship information between the objects in the scene, the scene information in the area, the body information of the characters, the relationship information between the characters and the scene, and the relationship information between the characters and the objects in the scene are abstracted into an information tensor;
轨迹生成模块，用于将所述抽象的信息张量输入到设置的记忆网络模型，以使所述记忆网络模型输出得到人物的未来运动轨迹信息；A trajectory generation module, used for inputting the abstract information tensor into the set memory network model, so that the memory network model output obtains the future motion trajectory information of the character;
活动预测模块，用于将所述抽象的信息张量输入到设置的第二神经网络中，所述第二神经网络输出人物的未来活动标签和未来目的位置信息；an activity prediction module, which is used to input the abstract information tensor into the set second neural network, and the second neural network outputs the character's future activity label and future destination location information;
异常报警模块，用于根据输出的人物的未来运动轨迹信息、人物的未来活动标签和未来目的信息，判定该人物行为是否可疑。The abnormal alarm module is used to determine whether the behavior of the character is suspicious according to the output information of the character's future movement track, the character's future activity label and future purpose information.
所述系统还包括：The system also includes:
人脸及身份识别模块，用于对出现在所设置的监控摄像中的人物进行识别，判定该人物的身份信息，所述身份信息包括内部人员或外部人员。The face and identity recognition module is used to identify the person appearing in the set surveillance camera, and determine the identity information of the person, and the identity information includes internal personnel or external personnel.
所述轨迹生成模块，还用于将所述抽象的信息张量进行记忆网络模型的压缩编码，得到视觉特征张量后，采用记忆网络模型进行解码，在实际坐标平面上输出人物的未来运动轨迹特征坐标信息。The trajectory generation module is also used to compress and encode the abstract information tensor with the memory network model, and after obtaining the visual feature tensor, use the memory network model for decoding, and output the future motion trajectory of the character on the actual coordinate plane. Feature coordinate information.
所述异常报警模块，用于判定该人物行为可疑时，进行预警。The abnormal alarm module is used to give an early warning when it is determined that the character's behavior is suspicious.
如上所见，本发明实施例将在区域内采集的人物图像信息输入到设置的第一神经网络中，输出得到在区域内的场景信息、人物肢体信息、人物与场景之间的关系信息、及人物与场景中的对象之间的关系信息；将得到的上述信息抽象为一个信息张量，分别输入到设置的记忆网络模型和设置的第二神经网络中，所述记忆网络模型输出得到人物的未来运动轨迹信息，所述第二神经网络输出人物的未来活动标签和未来目的位置信息；根据输出的人物的未来运动轨迹信息、人物的未来活动标签和未来目的信息，判定该人物行为是否可疑。由于本发明实施例采用了多个神经网络结合准确预测人物未来活动相关信息，以此来判定人物行为是否可疑，从而提高预警准确度，并改善预警效果。As can be seen above, in this embodiment of the present invention, the image information of the person collected in the area is input into the set first neural network, and the output obtains the scene information in the area, the person's body information, the relationship information between the person and the scene, and The relationship information between the character and the objects in the scene; the obtained above information is abstracted into an information tensor, which is respectively input into the set memory network model and the set second neural network, and the output of the memory network model is obtained. Future movement track information, the second neural network outputs the character's future activity label and future destination position information; according to the output character's future movement track information, character's future activity label and future purpose information, determine whether the character's behavior is suspicious. Since the embodiments of the present invention use multiple neural networks to accurately predict the future activity-related information of the person, it is used to determine whether the person's behavior is suspicious, thereby improving the early warning accuracy and improving the early warning effect.
附图说明Description of drawings
图1为本发明实施例提供的区域行为预警方法流程图；1 is a flowchart of a method for early warning of regional behavior provided by an embodiment of the present invention;
图2为本发明实施例提供的区域行为预警系统结构示意图；2 is a schematic structural diagram of a regional behavior early warning system provided by an embodiment of the present invention;
图3为本发明实施例提供的人脸及身份识别模块的执行过程示意图；3 is a schematic diagram of an execution process of a face and identity recognition module provided by an embodiment of the present invention;
图4为本发明实施例提供的人物行为模块的处理过程示意图；4 is a schematic diagram of a processing process of a character behavior module provided by an embodiment of the present invention;
图5为本发明实施例提供的人物交互模块的处理流程示意图；5 is a schematic diagram of a processing flow of a character interaction module provided by an embodiment of the present invention;
图6为本发明实施例提供的活动预测模块的结构示意图；6 is a schematic structural diagram of an activity prediction module provided by an embodiment of the present invention;
图7为本发明实施例提供的异常报警模块的处理流程示意图。FIG. 7 is a schematic diagram of a processing flow of an abnormal alarm module according to an embodiment of the present invention.
具体实施方式Detailed ways
为使本发明的目的、技术方案及优点更加清楚明白，以下参照附图并举实施例，对本发明进一步详细说明。In order to make the objectives, technical solutions and advantages of the present invention clearer, the present invention will be described in further detail below with reference to the accompanying drawings and embodiments.
从背景技术可以看出，在区域行为预警时造成预警准确度低及预警效果不好的主要原因是，只是采用人脸识别单元进行可疑人员的确定实现预警，或者采用监控摄像技术跟踪人物行踪，人为或采用统计学方法确定是否人物的行为是否可疑。更进一步地，背景技术中叙述的关于人物的行为预测多为轨迹预测，主要目的是预测出可疑人物的下一步的行进方向，从而进行实时追踪，没有涉及到人物未来的行为轨迹预测，预警准确度很低。It can be seen from the background technology that the main reasons for the low early warning accuracy and poor early warning effect in regional behavior early warning are that only the face recognition unit is used to determine suspicious persons to achieve early warning, or the surveillance camera technology is used to track the whereabouts of people. Determining whether a character's behavior is suspicious or not, either artificially or using statistical methods. Further, most of the behavior predictions about characters described in the background art are trajectory predictions, and the main purpose is to predict the next travel direction of suspicious characters, so as to perform real-time tracking, without involving the future behavior trajectory prediction of characters, and the early warning is accurate. degree is very low.
随着人工智能(AI)技术的发展，AI技术得到广泛应用。随着更先进的网络模型和算法不断被提出，AI应用的准确率也随之提高，未来必然能在诸多领域代替人工查验。因此，本发明实施例为了克服背景技术在区域行为预警时产生的问题，将当今最先进的视频人物行动轨迹和行为预测应用到区域行为预警中，取代背景技术中的人工监控和统计学方法，不仅能确定人物身份，同时还可以通过大量训练数据对设置的多个深度神经网络进行训练，由多个深度神经网络实时预测人物的异常行为，做到提前预警，防患于未然。具体地说，本发明实施例将在区域内采集的人物图像信息输入到设置的第一神经网络中，输出得到在区域内的场景信息、人物肢体信息、人物与场景之间的关系信息、及人物与场景中的对象之间的关系信息；将得到的上述信息抽象为一个信息张量，分别输入到设置的记忆网络模型和设置的第二神经网络中，所述记忆网络模型输出得到人物的未来运动轨迹信息，所述第二神经网络输出人物的未来活动标签和未来目的位置信息；根据输出的人物的未来运动轨迹信息、人物的未来活动标签和未来目的信息，判定该人物行为是否可疑。With the development of artificial intelligence (AI) technology, AI technology has been widely used. As more advanced network models and algorithms are continuously proposed, the accuracy of AI applications will also increase, and it will be possible to replace manual inspections in many fields in the future. Therefore, in order to overcome the problems caused by the background technology in regional behavior early warning, the embodiment of the present invention applies the most advanced video character action trajectory and behavior prediction to the regional behavior early warning, instead of the manual monitoring and statistical methods in the background technology, Not only can the identity of the person be determined, but also multiple deep neural networks can be trained through a large amount of training data, and the abnormal behavior of the person can be predicted in real time by multiple deep neural networks, so as to achieve early warning and prevent problems before they occur. Specifically, in the embodiment of the present invention, the image information of the characters collected in the area is input into the set first neural network, and the output is the scene information in the area, the body information of the characters, the relationship information between the characters and the scene, and The relationship information between the character and the objects in the scene; the obtained above information is abstracted into an information tensor, which is respectively input into the set memory network model and the set second neural network, and the output of the memory network model is obtained. Future movement track information, the second neural network outputs the character's future activity label and future destination position information; according to the output character's future movement track information, character's future activity label and future purpose information, determine whether the character's behavior is suspicious.
在这里，所述记忆网络模型为长短期记忆(LSTM)网络，神经网络采用CNN，以下以此为例进行详细说明。Here, the memory network model is a long short-term memory (LSTM) network, and the neural network adopts a CNN. The following is an example for detailed description.
这样，由于本发明实施例采用了多个神经网络结合准确预测人物未来活动相关信息，以此来判定人物行为是否可疑，从而提高预警准确度，并改善预警效果。In this way, since the embodiments of the present invention use multiple neural networks to accurately predict the future activity-related information of the person, to determine whether the person's behavior is suspicious, thereby improving the early warning accuracy and improving the early warning effect.
图1为本发明实施例提供的区域行为预警方法流程图，其具体步骤为：Fig. 1 is the flow chart of the regional behavior early warning method provided by the embodiment of the present invention, and its concrete steps are:
步骤101、将在区域内采集的人物图像信息输入到设置的第一CNN中，输出得到在区域内的场景信息、人物肢体信息、人物与场景之间的关系信息、及人物与场景中的对象之间的关系信息；Step 101: Input the person image information collected in the area into the set first CNN, and output the scene information in the area, the person's body information, the relationship information between the person and the scene, and the person and the object in the scene. relationship information;
步骤102、将在区域内的场景信息、人物肢体信息、人物与场景之间的关系信息、及人物与场景中的对象之间的关系信息抽象为一个信息张量；
步骤103、将所述抽象的信息张量输入到设置的LSTM网络和设置的第二CNN中，以使所述LSTM网络输出得到人物的未来运动轨迹信息，所述第二CNN输出人物的未来活动标签和未来目的位置信息；Step 103, input the abstract information tensor into the set LSTM network and the set second CNN, so that the LSTM network output obtains the character's future motion track information, and the second CNN outputs the character's future activity tags and future destination information;
步骤104、根据输出的人物的未来运动轨迹信息、人物的未来活动标签和未来目的信息，判定该人物行为是否可疑。Step 104: Determine whether the behavior of the character is suspicious according to the outputted future motion track information of the character, the character's future activity label and future purpose information.
在该方法中，所述在区域内采集人物图像信息之前，该方法还包括：In the method, before collecting the person image information in the area, the method further includes:
采用人脸及身份识别方式对出现在所设置的监控摄像中的人物进行识别，判定该人物的身份信息，所述身份信息包括内部人员或外部人员。The person appearing in the set surveillance camera is identified by means of face and identity recognition, and the identity information of the person is determined, and the identity information includes internal personnel or external personnel.
在这种情况下，所述判定该人物行为是否可疑还包括：In this case, the determining whether the character's behavior is suspicious also includes:
根据输出的人物的未来运动轨迹信息、人物的未来活动标签、未来目的信息及该人物的身份信息判定该人物行为是否可疑。Whether the behavior of the character is suspicious is determined according to the outputted future motion track information of the character, the character's future activity label, future purpose information and the character's identity information.
在该方法中，所述将抽象的信息张量输入到设置的LSTM网络中，输出得到人物的未来运动轨迹信息为：In this method, the abstract information tensor is input into the set LSTM network, and the output of the character's future motion trajectory information is:
将所述抽象的信息张量进行LSTM网络的压缩编码，得到视觉特征张量后，采用LSTM网络进行解码，在实际坐标平面上输出人物的未来运动轨迹特征坐标信息。The abstract information tensor is compressed and encoded by the LSTM network, and after the visual feature tensor is obtained, the LSTM network is used for decoding, and the feature coordinate information of the character's future motion trajectory is output on the actual coordinate plane.
在该方法中，所述第一CNN实时更新；所述LSTM实时更新；所述第二CNN实时更新。这样，本发明实施例的多个深度神经网络实际上是一个可以不断学习优化的网络，对于已经记录的信息都可以用于这多个神经网络的训练过程，从而提高网络判断的准确性。In this method, the first CNN is updated in real time; the LSTM is updated in real time; the second CNN is updated in real time. In this way, the multiple deep neural networks in the embodiments of the present invention are actually a network that can be continuously learned and optimized, and the recorded information can be used in the training process of the multiple neural networks, thereby improving the accuracy of network judgment.
在该方法中，所述方法还包括：In the method, the method further includes:
如果判定该人物行为可疑，进行预警If it is judged that the person's behavior is suspicious, give an early warning
图2为本发明实施例提供的区域行为预警系统结构示意图，包括：人物行为模块、人物交互模块、轨迹生成模块及活动预测模块，其中，2 is a schematic structural diagram of a regional behavior early warning system provided by an embodiment of the present invention, including: a character behavior module, a character interaction module, a trajectory generation module, and an activity prediction module, wherein,
人物行为模块，用于设置第一CNN、设置LSTM及第二CNN；Character behavior module, used to set the first CNN, set the LSTM and the second CNN;
人物交互模块，用于将在区域内采集的人物图像信息输入到设置的第一神经网络CNN中，输出得到在区域内的场景信息、人物肢体信息、人物与场景之间的关系信息、及人物与场景中的对象之间的关系信息，将在区域内的场景信息、人物肢体信息、人物与场景之间的关系信息、及人物与场景中的对象之间的关系信息抽象为一个信息张量；The character interaction module is used to input the character image information collected in the area into the set first neural network CNN, and output the scene information in the area, the character body information, the relationship information between the character and the scene, and the character The relationship information with the objects in the scene, the scene information in the area, the body information of the characters, the relationship information between the characters and the scene, and the relationship information between the characters and the objects in the scene are abstracted into an information tensor ;
轨迹生成模块，用于将所述抽象的信息张量输入到设置的LSTM网络，以使所述LSTM网络输出得到人物的未来运动轨迹信息；A trajectory generation module, for inputting the abstract information tensor into the set LSTM network, so that the LSTM network output obtains the future motion trajectory information of the character;
活动预测模块，用于将所述抽象的信息张量输入到设置的第二CNN中，所述第二CNN输出人物的未来活动标签和未来目的位置信息；an activity prediction module, for inputting the abstract information tensor into the set second CNN, and the second CNN outputs the character's future activity label and future destination location information;
异常报警模块，用于根据输出的人物的未来运动轨迹信息、人物的未来活动标签和未来目的信息，判定该人物行为是否可疑。The abnormal alarm module is used to determine whether the behavior of the character is suspicious according to the output information of the character's future movement track, the character's future activity label and future purpose information.
在该系统中还包括：人脸及身份识别模块，用于对出现在所设置的监控摄像中的人物进行识别，判定该人物的身份信息，所述身份信息包括内部人员或外部人员。The system also includes: a face and identity recognition module for identifying a person appearing in the set surveillance camera, and determining the identity information of the person, where the identity information includes internal personnel or external personnel.
在该系统中，所述轨迹生成模块，还用于将所述抽象的信息张量进行LSTM网络的压缩编码，得到视觉特征张量后，采用LSTM网络进行解码，在实际坐标平面上输出人物的未来运动轨迹特征坐标信息。In this system, the trajectory generation module is also used to compress and encode the abstract information tensor with the LSTM network, and after obtaining the visual feature tensor, use the LSTM network for decoding, and output the character's data on the actual coordinate plane. Feature coordinate information of future motion trajectory.
在该系统中，所述第一CNN实时更新；所述LSTM实时更新；所述第二CNN实时更新。In this system, the first CNN is updated in real time; the LSTM is updated in real time; the second CNN is updated in real time.
在该系统中，所述异常报警模块，用于判定该人物行为可疑时，进行预警。In this system, the abnormal alarm module is used to give an early warning when it is determined that the character's behavior is suspicious.
图3为本发明实施例提供的人脸及身份识别模块的执行过程示意图。人脸及身份识别模块一般设置在区域内，比如大型活动场所的主要通道和进出口，其主要的组成部分包括摄像单元、人脸采集单元、面部特征提取单元、人脸匹配单元及身份确认单元。具体地，如图3所示，其中：FIG. 3 is a schematic diagram of an execution process of a face and identity recognition module provided by an embodiment of the present invention. The face and identity recognition module is generally set up in the area, such as the main passage and entrance and exit of large-scale event venues. Its main components include a camera unit, a face acquisition unit, a facial feature extraction unit, a face matching unit and an identity confirmation unit. . Specifically, as shown in Figure 3, where:
步骤301、诸如高清摄像头的摄像单元采集区域内的主要通道和进出口的视频数据，发送给人脸采集单元；
步骤302、人脸采集单元采集视频数据中的人脸图片，将采集到的人脸图片发送给面部特征提取单元；
步骤303、面部特征提取单元对每一张人脸图片提取面部特征，将人脸图片的面部特征发送给人脸匹配单元；
步骤304、人脸匹配单元根据人脸图片的面部特征进行预设的区域内部人员的面部特征比对，确认是否相同，并将匹配结果发送给身份确认单元；
步骤305、身份确认单元根据匹配结果，确认是区域内部人员还是区域外部人员。Step 305: The identity confirmation unit confirms whether it is a person inside the area or a person outside the area according to the matching result.
在本发明实施例中，人物行为模块用于设置第一CNN、设置LSTM及第二CNN，在设置第一CNN、设置LSTM及第二CNN时，对于出现在监控摄像场景内的每个人的视觉信息进行编码，在这个过程中，针对的是人物外形和身体动作进行建模，而不是将人物抽象为一个点，如图4所示，图4为本发明实施例提供的人物行为模块的处理过程示意图。其中，为了对人物在区域场景中的外形进行建模，采用带有“RolAlign”的物体检测模型，来提取每个人物边界框的设定尺寸的CNN特征；为了获得人物的身体动作模型，使用了一个微软公司的图像识别(MSCOCO)数据集上训练LSTM网络，来提取人物动作关键点信息。In the embodiment of the present invention, the character behavior module is used for setting the first CNN, setting the LSTM and the second CNN. When setting the first CNN, setting the LSTM and the second CNN, the visual perception of each person appearing in the surveillance camera scene is The information is encoded. In this process, the modeling is aimed at the appearance and body movements of the characters, rather than abstracting the characters as a point, as shown in FIG. 4 , which is the processing of the character behavior module provided by the embodiment of the present invention. Process schematic. Among them, in order to model the appearance of the character in the regional scene, the object detection model with "RolAlign" is used to extract the CNN features of the set size of the bounding box of each character; in order to obtain the body motion model of the character, use An LSTM network is trained on a Microsoft Image Recognition (MSCCOCO) dataset to extract key point information of character actions.
在本发明实施例中，人物交互模块负责查看人物与周围环境的交互，包括人物与场景及人物与场景中的对象之间的交互。其中，人物与场景的交互是为了对人物附近场景进行识别。如图5所示，图5为本发明实施例提供的人物交互模块的处理流程示意图。首先使用训练的场景分割模型导出每一帧的像素级别场景语义分类，划分出场景中的道路及人行道等部分；然后选取设定的尺寸大小确定模型需要识别的环境区域。例如设定的尺寸大小为3，表示选取人物周围3*3大小的范围作为观察区域将以上不同时刻获取的信息输入到LSTM中，最终得到人物与场景之间的关系信息，然后将在区域内的场景信息、人物肢体信息、人物与场景之间的关系信息、及人物与场景中的对象之间的关系信息抽象为一个信息张量。In this embodiment of the present invention, the character interaction module is responsible for checking the interaction between the character and the surrounding environment, including the interaction between the character and the scene and between the character and objects in the scene. Among them, the interaction between the character and the scene is to identify the scene near the character. As shown in FIG. 5 , FIG. 5 is a schematic diagram of a processing flow of a character interaction module provided by an embodiment of the present invention. First, the trained scene segmentation model is used to derive the pixel-level scene semantic classification of each frame, and the roads and sidewalks in the scene are divided; then the set size is selected to determine the environmental area that the model needs to identify. For example, the set size is 3, which means that a 3*3 size area around the character is selected as the observation area, and the information obtained at different times above is input into the LSTM, and finally the relationship information between the character and the scene is obtained, and then the information in the area is obtained. The scene information, character body information, relationship information between characters and scenes, and relationship information between characters and objects in the scene are abstracted into an information tensor.
在本发明实施例中，轨迹生成模块将所述抽象的信息张量输入到设置的LSTM网络，以使所述LSTM网络输出得到人物的未来运动轨迹信息，将这些信息由LSTM网络编码器压缩成视觉特征张量Q，然后使用LSTM网络的解密器解密，得到人物的未来运动轨迹信息，在LSTM网络中采用了注意力机制，其关键点就是将多个特征投射到相关空间中，在这个空间中，辨别特征更容易被这种注意力机制捕获，焦点注意力对不同特征的关系进行建模，并将它们汇总到一个低维度向量中。In the embodiment of the present invention, the trajectory generation module inputs the abstract information tensor into the set LSTM network, so that the LSTM network outputs the information about the future movement trajectory of the character, and compresses the information into the LSTM network encoder into The visual feature tensor Q is then decrypted by the decryptor of the LSTM network to obtain the future motion trajectory information of the character. The attention mechanism is adopted in the LSTM network. The key point is to project multiple features into the relevant space. In this space , discriminative features are more easily captured by this attention mechanism, and focal attention models the relationship of different features and summarizes them into a low-dimensional vector.
在本发明实施例中，活动预测模块有两个任务，确定人物未来活动的类型和发生地点。相应地，它包括两个部分，曼哈顿网格的活动位置预测和活动标签预测。活动标签预测的作用是猜出画面中的人物最后的目的是什么，预测未来某个瞬间的活动。活动标签在某一时刻并不限于一种，比如一个人物可以同时走路和携带物品。而活动位置预测的功能，是为轨迹生成模块进行纠错，它确定人物的最终目的地，以弥补轨迹生成模块和活动标签预测之间的偏差。包括位置分类和位置回归两个任务。位置分类的目的是预测最终位置坐标所在的网格块，如图6所示，图6为本发明实施例提供的活动预测模块的结构示意图。位置回归的目标是预测网格块中心(图中的圆点)与最终位置坐标(箭头的末端)的偏差。In this embodiment of the present invention, the activity prediction module has two tasks, which are to determine the type and location of the character's future activity. Correspondingly, it consists of two parts, active location prediction and active label prediction for the Manhattan grid. The function of activity label prediction is to guess what the final purpose of the characters in the picture is, and to predict the activity of a certain moment in the future. Active tags are not limited to one at a time, for example a character can walk and carry items at the same time. The function of the activity location prediction is to correct errors for the trajectory generation module, which determines the final destination of the character to compensate for the deviation between the trajectory generation module and the activity label prediction. It includes two tasks, location classification and location regression. The purpose of the location classification is to predict the grid block where the final location coordinates are located, as shown in FIG. 6 , which is a schematic structural diagram of an activity prediction module provided by an embodiment of the present invention. The goal of position regression is to predict the deviation of the grid block center (the dot in the figure) from the final position coordinate (the end of the arrow).
本发明实施例中的异常报警模块可以根据人脸及身份识别模块和活动预测模块的结果，分析出此人行为是否可疑，从而对可疑人员的可疑行为进行提前预警，如图7所示，图7为本发明实施例提供的异常报警模块的处理流程示意图，其具体步骤为：The abnormal alarm module in the embodiment of the present invention can analyze whether the behavior of the person is suspicious according to the results of the face and the identity recognition module and the activity prediction module, so as to give an early warning to the suspicious behavior of the suspicious person, as shown in FIG. 7 . 7 is a schematic diagram of the processing flow of the abnormal alarm module provided by the embodiment of the present invention, and its specific steps are:
步骤701、从人脸及身份识别模块得到人员身份信息；Step 701, obtain personnel identity information from the face and identity recognition module;
步骤702、从活动预测模块得到人物未来活动的类型和发生地点；
步骤703、将步骤701和步骤702得到的信息进行收集；
步骤704、确定人物的行为是否异常，如果是，执行步骤705；否则，返回步骤703继续执行；Step 704, determine whether the behavior of the character is abnormal, if so, execute
步骤705、确定人物是否为区域外部人员，如果是，在执行步骤706；如果否，执行步骤707；
步骤706、进行预警，提醒安保人员进行密切观察；Step 706: Carry out an early warning and remind security personnel to observe closely;
步骤707、记录视频片段，用于后续分析。Step 707: Record the video clip for subsequent analysis.
采用本发明实施例，被监控人员包括区域内部人员和外来人员。对于小区环境包括业主、物业人员和外部人员等；对于别墅环境包括主人、访客、维修工人、保安、未知人员等；对于库房环境包括保管员、取件员、送件员、外来人员等；对于银行环境包括工作人员、办理业务人员、保安、可疑人员等；对于商场环境包括客人、店员、保洁人员、保安等。By adopting the embodiment of the present invention, the monitored personnel include the personnel inside the area and the outside personnel. For the residential environment, it includes owners, property personnel and external personnel; for the villa environment, it includes the owner, visitors, maintenance workers, security guards, unknown personnel, etc.; The banking environment includes staff, business personnel, security guards, suspicious personnel, etc.; the shopping mall environment includes guests, shop assistants, cleaning staff, security guards, etc.
采用本发明实施例，预测得到的人物未来活动信息可以包括：外部人员尾随业主进入楼道，驻足观察小区住户作息规律，查看楼房结构、监控摄像位置及小区出入通道；翻入院墙，撬开门锁等盗窃破坏行为；非法进入，盗窃机密材料、技术、文件等行为；在银行、商场实施盗窃抢劫的行为且不限于以上可疑行为。Using the embodiment of the present invention, the predicted future activity information of the characters may include: external personnel follow the owner to enter the corridor, stop to observe the work and rest rules of the residents of the community, check the building structure, the location of surveillance cameras and the access channels of the community; turn over the entrance wall, pick the door lock, etc. Theft and sabotage; illegal entry, theft of confidential materials, technologies, documents, etc.; theft and robbery in banks and shopping malls and are not limited to the above suspicious behaviors.
采用本发明实施例对犯罪行为预警的通知对象包含安保人员、民警、业主且不限于上述人员。实时监控预警方法包含短信预警、电话预警、警铃预警、物联网(IOT)设备预警、视频预警等常见预警方法，但并不局限于上述预警方法。The notification objects for early warning of criminal behaviors using the embodiments of the present invention include security personnel, policemen, and owners, and are not limited to the above-mentioned personnel. The real-time monitoring and early warning methods include common early warning methods such as SMS warning, telephone warning, alarm bell warning, Internet of Things (IOT) device warning, and video warning, but are not limited to the above warning methods.
采用本发明实施例，区域包含且不限于小区环境，别墅环境，库房环境，银行环境以及商场环境。对于其他人员繁杂，容易出现盗窃抢劫的环境同样适用于区域的概念。Using the embodiment of the present invention, the area includes but is not limited to a community environment, a villa environment, a warehouse environment, a bank environment, and a shopping mall environment. For other environments that are crowded with people and prone to theft and robbery, the concept of area is also applicable.
以下举几个具体例子进行说明The following are some specific examples to illustrate
实施例一：将区域定为小区。采用本发明实施例，在小区环境中，提前预警外部人员尾随业主进入楼道，驻足观察小区住户作息规律，查看楼房结构、监控摄像位置及小区出入通道等异常行为。Embodiment 1: Determining an area as a cell. Using the embodiment of the present invention, in the community environment, outsiders are warned in advance to follow the owner to enter the corridor, stop to observe the work and rest rules of the community residents, and check the abnormal behaviors such as building structure, monitoring camera positions, and community access passages.
具体过程为：The specific process is:
第一个步骤，在小区的入口或主要干道通过人脸及身份识别模块检测到外部人员进入小区。The first step is to detect the entry of outsiders into the community through the face and identity recognition module at the entrance or main road of the community.
第二个步骤，根据视频监控，通过人物行为模块，人物交互模块，轨迹生成模块和活动预测模块预测外部人员的未来行动路径和行为。In the second step, according to the video surveillance, the future action paths and behaviors of external personnel are predicted through the character behavior module, the character interaction module, the trajectory generation module and the activity prediction module.
第三个步骤，异常报警模块根据以上信息，发现其出现尾随业主走向楼道，驻足观察小区住户作息规律，或是查看楼房结构、监控摄像位置及小区出入通道等异常行为。提前报警，提醒安保人员密切观察外部人员后续动作。必要时出动人员进行抓捕盘问。In the third step, according to the above information, the abnormal alarm module finds that it follows the owner to the corridor, stops to observe the work and rest rules of the residents in the community, or checks the abnormal behavior of the building structure, the location of surveillance cameras, and the access channels of the community. Call the police in advance to remind security personnel to closely observe the follow-up actions of external personnel. When necessary, personnel will be dispatched for arrest and questioning.
实施例二：将区域定为别墅，采用本发明实施例，在独栋别墅外，对于翻入院墙，撬开门锁等盗窃破坏行为，进行预警。Embodiment 2: The area is designated as a villa, and by adopting the embodiment of the present invention, outside the single-family villa, early warning is given for theft and sabotage behaviors such as turning over the courtyard wall and picking the door lock.
具体过程为：The specific process is:
第一个步骤，在独栋别墅的高处墙壁上设置摄像头，通过人脸识别判断在别墅院墙外的人员是否为主人。The first step is to set up a camera on the high wall of the single-family villa, and determine whether the person outside the courtyard wall of the villa is the owner through face recognition.
第二个步骤，根据视频的监控，通过人物行为模块，人物交互模块，轨迹生成模块和活动预测模块预测该人员的未来行动路径和行为。The second step is to predict the future action path and behavior of the person through the character behavior module, the character interaction module, the trajectory generation module and the activity prediction module according to the monitoring of the video.
第三个步骤，异常报警模块根据以上信息，如发现有异常行为，可通过手机端的APP提前向主人和安保人员报警，并开启自动播放录像功能。也可同时启动别墅内的IOT音响，播放警告语音。In the third step, according to the above information, if abnormal behavior is found, the abnormal alarm module can alarm the owner and security personnel in advance through the APP on the mobile phone, and enable the automatic video playback function. You can also start the IOT audio in the villa at the same time to play the warning voice.
实施例三：将区域设定为工厂库房内，采用本发明实施例对于非法进入，盗窃机密材料、技术、文件等行为，进行预警。Embodiment 3: The area is set as the factory warehouse, and the embodiment of the present invention is used to provide early warning for illegal entry, theft of confidential materials, technologies, documents, and the like.
具体过程为：The specific process is:
第一个步骤，在库房的入口或主要干道通过人脸识别检测到可疑人员进入，首先判断该人身份。The first step is to detect the entry of a suspicious person at the entrance of the warehouse or the main road through face recognition, and first determine the identity of the person.
第二个步骤，根据视频的监控，通过人物行为模块，人物交互模块，轨迹生成模块和活动预测模块预测该人员的未来行动路径和行为。The second step is to predict the future action path and behavior of the person through the character behavior module, the character interaction module, the trajectory generation module and the activity prediction module according to the monitoring of the video.
第三个步骤，异常报警模块根据以上信息，如发现其有盗窃机密材料、技术、文件的嫌疑，提前报警，提醒安保人员密切观察该处摄像头情况，并封堵库房出入口，实施抓捕。In the third step, the abnormal alarm module, based on the above information, if it finds that it is suspected of stealing confidential materials, technologies, or documents, it will call the police in advance, remind the security personnel to closely observe the situation of the camera, and block the entrance and exit of the warehouse to implement arrest.
实施例四：将区域设定为在银行内，采用本发明实施例对于将要实施抢劫和偷窃的行为，进行提前预警，提醒保安注意，必要时向公安报警。Embodiment 4: The area is set to be in the bank, and the embodiment of the present invention is used to give an early warning to the behavior of robbery and theft, remind the security guard to pay attention, and report the police to the police when necessary.
具体过程为：The specific process is:
第一个步骤，在银行的出入口和主要角落通过人脸识别检测进入银行的每个人，首先判断该人身份，如果是记录在案的在逃人员，立即触发报警。The first step is to detect everyone who enters the bank through face recognition at the entrances, exits and main corners of the bank, first determine the identity of the person, and immediately trigger the alarm if it is a fugitive on record.
第二个步骤，根据视频的监控，通过人物行为模块，人物交互模块，轨迹生成模块和活动预测模块预测该人员的未来行动路径和行为。The second step is to predict the future action path and behavior of the person through the character behavior module, the character interaction module, the trajectory generation module and the activity prediction module according to the monitoring of the video.
第三个步骤，异常报警模块根据以上信息，如发现其有实施抢劫和偷窃的嫌疑，提前连通安保人员身上的无线设备，提醒安保人员密切观察该人动向。并根据不同的安全等级，必要时直接通知警方出警。In the third step, according to the above information, if the abnormal alarm module is found to be suspected of robbery and theft, it will connect the wireless device on the security personnel in advance, and remind the security personnel to closely observe the movement of the person. And according to different security levels, directly notify the police if necessary.
实施例五：将区域设定大型商场内，采用本发明实施例对于将要实施抢劫和偷窃的行为，进行提前预警，提醒保安注意，必要时向公安报警。Embodiment 5: The area is set in a large shopping mall, and the embodiment of the present invention is used to give an early warning to the behavior of robbery and theft, remind the security guard to pay attention, and report the police to the police when necessary.
具体过程为：The specific process is:
第一个步骤，在大型商场的出入口和主要角落通过人脸识别检测进入商场的每个人，首先判断该人身份，如果是记录在案的在逃人员，立即触发报警。The first step is to detect everyone entering the mall through face recognition at the entrances, exits and main corners of large shopping malls, first determine the identity of the person, and immediately trigger the alarm if it is a fugitive on record.
第二个步骤，根据视频的监控，通过人物行为模块，人物交互模块，轨迹生成模块和活动预测模块预测该人员的未来行动路径和行为。The second step is to predict the future action path and behavior of the person through the character behavior module, the character interaction module, the trajectory generation module and the activity prediction module according to the monitoring of the video.
第三步骤，异常报警模块根据以上信息，如发现其有实施抢劫和偷窃的嫌疑，提前连通安保人员身上的无线设备，提醒安保人员对该人员进行跟踪观察。In the third step, according to the above information, if the abnormal alarm module is found to be suspected of robbery and theft, it will connect the wireless device on the security personnel in advance, and remind the security personnel to track and observe the personnel.
第四个步骤，如预警过程中，该人员已经实施了犯罪行为，可通过联网的商场大屏幕实时播放该人的行踪，提醒商场各处保安进行围堵。In the fourth step, if the person has committed a crime during the early warning process, the person's whereabouts can be broadcast in real time through the networked shopping mall big screen, reminding the security guards all over the mall to contain them.
可以看出，本发明实施例了区域内的智能监控，不仅提供了更先进有效的可疑人员未来行为预测及提前预警，而且也减轻了区域内安保人员查看多个监控录像的繁重工作量。It can be seen that the embodiment of the present invention implements intelligent monitoring in the area, which not only provides more advanced and effective future behavior prediction and early warning of suspicious persons, but also reduces the heavy workload of security personnel in the area to view multiple surveillance videos.
以上所述仅为本发明的较佳实施例而已，并不用以限制本发明，凡在本发明的精神和原则之内，所做的任何修改、等同替换、改进等，均应包含在本发明保护的范围之内。The above descriptions are only preferred embodiments of the present invention, and are not intended to limit the present invention. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principles of the present invention shall be included in the present invention. within the scope of protection.
Claims (10)
Translated fromChinesePriority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910921181.4ACN110674761B (en) | 2019-09-27 | 2019-09-27 | A regional behavior early warning method and system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910921181.4ACN110674761B (en) | 2019-09-27 | 2019-09-27 | A regional behavior early warning method and system |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN110674761A CN110674761A (en) | 2020-01-10 |
| CN110674761Btrue CN110674761B (en) | 2022-07-15 |
Family
ID=69079531
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910921181.4AActiveCN110674761B (en) | 2019-09-27 | 2019-09-27 | A regional behavior early warning method and system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110674761B (en) |
Families Citing this family (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113449558A (en)* | 2020-03-26 | 2021-09-28 | 上海依图网络科技有限公司 | Method and device for monitoring abnormal behaviors of personnel |
| CN111935461A (en)* | 2020-09-11 | 2020-11-13 | 合肥创兆电子科技有限公司 | Based on intelligent security control system |
| CN113376656A (en)* | 2021-06-07 | 2021-09-10 | 重庆大学 | Multi-robot enclosure system based on LSTM prediction |
| CN115294515B (en)* | 2022-07-05 | 2023-06-13 | 南京邮电大学 | A comprehensive anti-theft management method and system based on artificial intelligence |
| CN115273370B (en)* | 2022-09-28 | 2023-01-03 | 苏州美集供应链管理股份有限公司 | System and method for monitoring personnel in real time based on visual scanning and track recognition |
| CN116052077A (en)* | 2022-12-30 | 2023-05-02 | 惠州学院 | A method and system for image recognition and alarm based on artificial intelligence |
| CN116993777A (en)* | 2023-07-19 | 2023-11-03 | 中国电信股份有限公司 | Object monitoring information display method, device, computer equipment and storage medium |
| CN118629137B (en)* | 2024-08-08 | 2024-11-05 | 江苏蓝盾智能科技有限公司 | Artificial intelligence-oriented access control monitoring system |
| CN118644094B (en)* | 2024-08-19 | 2024-12-24 | 杭州万斛泉科技有限公司 | Intelligent community user behavior intelligent monitoring and early warning method and system |
Family Cites Families (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108509880A (en)* | 2018-03-21 | 2018-09-07 | 南京邮电大学 | A kind of video personage behavior method for recognizing semantics |
| CN109410496B (en)* | 2018-10-25 | 2022-04-01 | 北京交通大学 | Intrusion early warning method and device and electronic equipment |
- 2019
- 2019-09-27CNCN201910921181.4Apatent/CN110674761B/enactiveActive
Also Published As
| Publication number | Publication date |
|---|---|
| CN110674761A (en) | 2020-01-10 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110674761B (en) | A regional behavior early warning method and system | |
| US20050168574A1 (en) | Video-based passback event detection | |
| CN103986910A (en) | A method and system for counting passenger flow based on intelligent analysis camera | |
| US12253605B2 (en) | Individual identification and tracking via combined video and LiDAR systems | |
| CN110852148A (en) | Visitor destination verification method and system based on target tracking | |
| Patil et al. | Suspicious movement detection and tracking based on color histogram | |
| CN118334559A (en) | Intelligent campus security management method and system based on face recognition | |
| CN115294519A (en) | An abnormal event detection and early warning method based on lightweight network | |
| CN113469080A (en) | Individual, group and scene interactive collaborative perception method, system and equipment | |
| CN118942159A (en) | A park visitor safety warning method, system and medium based on digital twin | |
| CN116798176A (en) | Data management system based on big data and intelligent security | |
| CN109511090A (en) | A kind of interactive mode tracing and positioning anticipation system | |
| Zin et al. | A Markov random walk model for loitering people detection | |
| Shivthare et al. | Suspicious activity detection network for video surveillance using machine learning | |
| Arshad et al. | Smart Surveillance System Using Machine Learning | |
| Masood et al. | Identification of anomaly scenes in videos using graph neural networks | |
| Rahangdale et al. | Event detection using background subtraction for surveillance systems | |
| Ho et al. | Public space behavior modeling with video and sensor analytics | |
| Sahana et al. | IOT Based Burglar Detection and Alarming System Using Raspberry Pi | |
| Prezioso et al. | Integrating object detection and advanced analytics for smart city crowd management | |
| CN114299448A (en) | A method and system for recognizing behavior of personnel illegally carrying mobile phones into sensitive places | |
| Shrivastav et al. | Integrated Approach for Real-time Human Counting, Tracking, and Direction Estimation using Advanced Algorithms | |
| KR20220031258A (en) | A method for providing active security control service based on learning data corresponding to counseling event | |
| Kushbu et al. | Crowd analysis using deepSORT | |
| CN118917980A (en) | Community management system combining image analysis and big data deep learning |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |