CN112287791A - Intelligent violence and terrorism behavior detection method based on equipment side - Google Patents
Intelligent violence and terrorism behavior detection method based on equipment side Download PDFInfo
- Publication number
- CN112287791A CN112287791A CN202011129340.6A CN202011129340A CN112287791A CN 112287791 A CN112287791 A CN 112287791A CN 202011129340 A CN202011129340 A CN 202011129340A CN 112287791 A CN112287791 A CN 112287791A
- Authority
- CN
- China
- Prior art keywords
- terrorism
- violence
- model
- behavior
- sudden
- 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.)
- Pending
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/50—Context or environment of the image
- G06V20/52—Surveillance or monitoring of activities, e.g. for recognising suspicious objects
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR 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; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- General Physics & Mathematics (AREA)
- Evolutionary Computation (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- General Engineering & Computer Science (AREA)
- Computing Systems (AREA)
- Software Systems (AREA)
- Molecular Biology (AREA)
- Computational Linguistics (AREA)
- Biophysics (AREA)
- Biomedical Technology (AREA)
- Mathematical Physics (AREA)
- General Health & Medical Sciences (AREA)
- Health & Medical Sciences (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Biology (AREA)
- Multimedia (AREA)
- Alarm Systems (AREA)
Abstract
The invention provides an intelligent violence and terrorism behavior detection method based on an equipment terminal. Which comprises the following steps: s1, classifying the violence and terrorism behaviors; s2, selecting a lightweight network MobileNet model, wherein the model refers to the idea of separable convolution in the convolution operation, and dividing the standard convolution operation into depthwise and pointwise; s3, obtaining a model by using the violence and terrorism data marked by the network training, and compressing the model by using a pruning mode, namely removing parameters with small contribution to output by sequencing the contribution degree of neurons in the neural network; s4, carrying out logic relation combination on the sudden and violent elements detected by the model, and defining conditions for judging sudden and violent behaviors; and S5, detecting the current frame image, judging the sudden and terrorist behavior when the detection result accords with the logical relationship, sending an alarm to inform the nearest security personnel, and judging the behavior to be normal if the detection result does not accord with the logical relationship.
Description
Technical Field
The invention relates to an intelligent violence and terrorism behavior detection method based on an equipment terminal, and belongs to the technical field of violence and terrorism detection and deep learning.
Background
The edge end deployment of the deep learning model enables equipment to be intelligent, outdoor safety events occur frequently in recent years, and rescue is missed due to the fact that alarm is given out at the first time, so that the edge end intelligent deployment can replace manual video monitoring, targets in a monitoring area can be monitored and distinguished in real time, accident alarm rate is improved, and rescue speed is improved. The traditional riot and terrorist detection mostly depends on the abnormal behavior detection of limbs, and the target detection can also be applied to the field of riot and terrorist detection.
Disclosure of Invention
The invention aims to solve the problem of intelligently detecting the sudden and terrorist behaviors by side equipment, and provides an intelligent sudden and terrorist behavior detection method based on an equipment side.
In order to achieve the purpose, the invention is realized by the following technical scheme:
an intelligent violence and terrorism behavior detection method based on an equipment terminal comprises the following steps:
s1, classifying the violence and terrorism behaviors;
s2, selecting a lightweight network MobileNet model, wherein the model refers to the idea of separable convolution in the convolution operation, and dividing the standard convolution operation into depthwise and pointwise;
s3, obtaining a model by using the violence and terrorism data marked by the network training, and compressing the model by using a pruning mode, namely removing parameters with small contribution to output by sequencing the contribution degree of neurons in the neural network;
s4, carrying out logic relation combination on the sudden and violent elements detected by the model, and defining conditions for judging sudden and violent behaviors;
and S5, detecting the current frame image, judging the behavior as a sudden terrorism when the detection result accords with the logical relation, sending an alarm to inform the nearest security personnel, and judging the behavior as a normal behavior if the detection result does not accord with the logical relation.
On the basis of the intelligent violence and terrorism behavior detection method based on the equipment side, the dimensionality of input data is assumed to beThe number of combined convolution parameters of depthwise + Pointwise is:w is the image width, h is the image, c is the imageAnd n is the number of convolution kernels.
On the basis of the intelligent violence and terrorism behavior detection method based on the equipment side, the violence and terrorism behaviors in 6 are artificially defined by referring to the labeling format of the PASCAL VOC data set in s 1: bloody fish, explosions, car accidents, guns, knives and terrorists, mark data and participate in training.
The invention has the advantages that:
in outdoor public places such as squares, stations and the like, the angle of the camera is fixed, so that the image background is kept unchanged, and under the condition, a riot and terrorist detection model can be carried at the camera, so that the purpose of monitoring the monitoring video in real time is achieved.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
An intelligent violence and terrorism behavior detection method based on an equipment terminal comprises the following steps:
s1, artificially defining the violent behavior in 6 by referring to the labeling format of the PASCAL VOC data set: bloody fish, explosions, car accidents, guns, knives and terrorists, mark data and participate in training;
s2, selecting a lightweight network MobileNet model, wherein the model uses the idea of separable convolution for reference in convolution operation, and divides standard convolution operation into depthwise and pointwise parts for the purpose of eliminating redundant parameters and reducing parameter quantity and calculated quantity, and supposing that the dimensionality of input data isThe number of combined convolution parameters of depthwise + Pointwise is as follows, w is the image width, h is the image, c is the image channel, and n is the number of convolution kernels;
s3, obtaining a model by using the violence and terrorism data marked by the network training, and compressing the model by using a pruning mode, namely removing parameters with small contribution to output by sequencing the contribution degree of neurons in the neural network;
s4, carrying out logic relation combination on the sudden and violent elements detected by the model, and defining conditions for judging sudden and violent behaviors;
and S5, detecting the current frame image, judging the behavior as a sudden terrorism when the detection result accords with the logical relation, sending an alarm to inform the nearest security personnel, and judging the behavior as a normal behavior if the detection result does not accord with the logical relation. For example, single behaviors such as bloody smell, explosion and car accidents are classified as violence, and once an alarm needs to be given immediately, a management department and the nearest security personnel are notified to be rescued on site. Secondly, the judgment of the combined elements, such as the occurrence of police and firearms which do not represent terrorist behaviors, needs to be carried out by adding other elements such as terrorists and the like.
Finally, it should be noted that: although the present invention has been described in detail with reference to the foregoing embodiments, it will be apparent to those skilled in the art that changes may be made in the embodiments and/or equivalents thereof without departing from the spirit and scope of the invention. Any modification, equivalent replacement, or improvement made within the spirit and principle of the present invention should be included in the protection scope of the present invention.
Claims (3)
1. An intelligent violence and terrorism behavior detection method based on an equipment terminal is characterized by comprising the following steps:
s1, classifying the violence and terrorism behaviors;
s2, selecting a lightweight network MobileNet model, wherein the model refers to the idea of separable convolution in the convolution operation, and dividing the standard convolution operation into depthwise and pointwise;
s3, obtaining a model by using the violence and terrorism data marked by the network training, and compressing the model by using a pruning mode, namely removing parameters with small contribution to output by sequencing the contribution degree of neurons in the neural network;
s4, carrying out logic relation combination on the sudden and violent elements detected by the model, and defining conditions for judging sudden and violent behaviors;
and S5, detecting the current frame image, judging the behavior as a sudden terrorism when the detection result accords with the logical relation, sending an alarm to inform the nearest security personnel, and judging the behavior as a normal behavior if the detection result does not accord with the logical relation.
2. The device-side-based intelligent violence and terrorism behavior detection method according to claim 1, characterized in that: assuming that the dimension of input data is, the combined convolution parameter number of depthwise + Pointwise is:w is the image width, h is the image, c is the image channel, and n is the number of convolution kernels.
3. The device-side-based intelligent violence and terrorism behavior detection method according to claim 1, characterized in that: in s1, referring to the label format of the PASCAL VOC data set, the violence terrorist behavior in 6 is artificially defined: bloody fish, explosions, car accidents, guns, knives and terrorists, mark data and participate in training.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202011129340.6A CN112287791A (en) | 2020-10-21 | 2020-10-21 | Intelligent violence and terrorism behavior detection method based on equipment side |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202011129340.6A CN112287791A (en) | 2020-10-21 | 2020-10-21 | Intelligent violence and terrorism behavior detection method based on equipment side |
Publications (1)
Publication Number | Publication Date |
---|---|
CN112287791A true CN112287791A (en) | 2021-01-29 |
Family
ID=74423165
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202011129340.6A Pending CN112287791A (en) | 2020-10-21 | 2020-10-21 | Intelligent violence and terrorism behavior detection method based on equipment side |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN112287791A (en) |
Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109635790A (en) * | 2019-01-28 | 2019-04-16 | 杭州电子科技大学 | A kind of pedestrian's abnormal behaviour recognition methods based on 3D convolution |
CN110751214A (en) * | 2019-10-21 | 2020-02-04 | 山东大学 | Target detection method and system based on lightweight deformable convolution |
CN110781912A (en) * | 2019-09-10 | 2020-02-11 | 东南大学 | Image classification method based on channel expansion inverse convolution neural network |
-
2020
- 2020-10-21 CN CN202011129340.6A patent/CN112287791A/en active Pending
Patent Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109635790A (en) * | 2019-01-28 | 2019-04-16 | 杭州电子科技大学 | A kind of pedestrian's abnormal behaviour recognition methods based on 3D convolution |
CN110781912A (en) * | 2019-09-10 | 2020-02-11 | 东南大学 | Image classification method based on channel expansion inverse convolution neural network |
CN110751214A (en) * | 2019-10-21 | 2020-02-04 | 山东大学 | Target detection method and system based on lightweight deformable convolution |
Non-Patent Citations (2)
Title |
---|
廖毅雄: "基于深度学习的手势识别及人体行为识别算法研究", 《中国优秀博硕士学位论文全文数据库(硕士) 信息科技辑》 * |
邢艳芳等: "基于Mobilenet的敏感图像识别系统设计", 《器件与设计》 * |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN107862437B (en) | Public area crowd gathering early warning method and system based on risk probability assessment | |
CN107911653B (en) | Intelligent video monitoring module, system, method and storage medium for residence | |
CN106097346A (en) | A kind of video fire hazard detection method of self study | |
CN100583128C (en) | Real time intelligent control method based on natural video frequency | |
CN211293956U (en) | AI-based identification and alarm system for abnormal agent on construction site | |
CN111488803A (en) | Airport target behavior understanding system integrating target detection and target tracking | |
CN112287816A (en) | Dangerous working area accident automatic detection and alarm method based on deep learning | |
CN104700586B (en) | Intelligent area fire fighting monitoring and alarming platform | |
CN111257507A (en) | Gas concentration detection and accident early warning system based on unmanned aerial vehicle | |
CN109460744A (en) | A kind of video monitoring system based on deep learning | |
CN104765307A (en) | Aerial photography system of unmanned aerial vehicle | |
CN112107074A (en) | Underground terminal intelligent safety helmet based on multi-sensor fusion | |
Yang et al. | Early detection of forest fire based on unmaned aerial vehicle platform | |
CN108256447A (en) | A kind of unmanned plane video analysis method based on deep neural network | |
CN114971409A (en) | Smart city fire monitoring and early warning method and system based on Internet of things | |
CN111476979A (en) | Intelligent security and stability maintenance method and system based on multi-model analysis | |
CN115394036A (en) | Monitoring and early warning method and system for building fire | |
CN112287791A (en) | Intelligent violence and terrorism behavior detection method based on equipment side | |
CN117151478B (en) | Chemical enterprise risk early warning method and system based on convolutional neural network | |
CN110930632A (en) | Early warning system based on artificial intelligence | |
CN104574729A (en) | Alarming method, device and system | |
CN109447878A (en) | City big data security administration platform | |
CN113628172A (en) | Intelligent detection algorithm for personnel handheld weapons and smart city security system | |
CN117789398A (en) | Campus emergency guiding management system | |
CN112863105A (en) | Fire-fighting early warning system based on fire-fighting relevance |
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 | ||
RJ01 | Rejection of invention patent application after publication |
Application publication date: 20210129 |
|
RJ01 | Rejection of invention patent application after publication |