CN109034020A - A kind of community's Risk Monitoring and prevention method based on Internet of Things and deep learning - Google Patents
A kind of community's Risk Monitoring and prevention method based on Internet of Things and deep learning Download PDFInfo
- Publication number
- CN109034020A CN109034020A CN201810765316.8A CN201810765316A CN109034020A CN 109034020 A CN109034020 A CN 109034020A CN 201810765316 A CN201810765316 A CN 201810765316A CN 109034020 A CN109034020 A CN 109034020A
- Authority
- CN
- China
- Prior art keywords
- community
- risk monitoring
- internet
- things
- deep learning
- 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
- 238000012544 monitoring process Methods 0.000 title claims abstract description 36
- 238000000034 method Methods 0.000 title claims abstract description 21
- 230000002265 prevention Effects 0.000 title claims abstract description 17
- 238000013135 deep learning Methods 0.000 title claims abstract description 13
- 230000004927 fusion Effects 0.000 claims abstract description 15
- 230000006399 behavior Effects 0.000 claims abstract description 11
- 238000004458 analytical method Methods 0.000 claims abstract description 7
- 238000013527 convolutional neural network Methods 0.000 claims abstract description 5
- 238000013528 artificial neural network Methods 0.000 claims description 17
- 230000013016 learning Effects 0.000 claims description 17
- 238000013461 design Methods 0.000 claims description 7
- 230000006870 function Effects 0.000 claims description 6
- 238000007405 data analysis Methods 0.000 claims description 5
- 230000010354 integration Effects 0.000 claims description 5
- 238000006243 chemical reaction Methods 0.000 claims description 2
- 238000004140 cleaning Methods 0.000 claims description 2
- 238000011478 gradient descent method Methods 0.000 claims description 2
- 230000008447 perception Effects 0.000 description 9
- 238000013138 pruning Methods 0.000 description 9
- 238000012549 training Methods 0.000 description 9
- 238000001514 detection method Methods 0.000 description 7
- 238000005516 engineering process Methods 0.000 description 7
- 238000005457 optimization Methods 0.000 description 7
- 230000009467 reduction Effects 0.000 description 5
- 239000013598 vector Substances 0.000 description 5
- 230000008569 process Effects 0.000 description 4
- 230000019771 cognition Effects 0.000 description 3
- 230000003993 interaction Effects 0.000 description 3
- 238000007726 management method Methods 0.000 description 3
- 238000012545 processing Methods 0.000 description 3
- 238000012546 transfer Methods 0.000 description 3
- 238000013459 approach Methods 0.000 description 2
- 230000003542 behavioural effect Effects 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 2
- 230000005540 biological transmission Effects 0.000 description 2
- 230000008859 change Effects 0.000 description 2
- 238000010276 construction Methods 0.000 description 2
- 238000011161 development Methods 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 230000006872 improvement Effects 0.000 description 2
- 238000010801 machine learning Methods 0.000 description 2
- 238000011160 research Methods 0.000 description 2
- 238000003860 storage Methods 0.000 description 2
- 238000012360 testing method Methods 0.000 description 2
- 241001269238 Data Species 0.000 description 1
- 230000009471 action Effects 0.000 description 1
- 230000008901 benefit Effects 0.000 description 1
- 238000013480 data collection Methods 0.000 description 1
- 238000013500 data storage Methods 0.000 description 1
- 230000007423 decrease Effects 0.000 description 1
- 238000010612 desalination reaction Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 238000011156 evaluation Methods 0.000 description 1
- 238000001914 filtration Methods 0.000 description 1
- 230000005484 gravity Effects 0.000 description 1
- 210000004209 hair Anatomy 0.000 description 1
- 238000005286 illumination Methods 0.000 description 1
- 238000003780 insertion Methods 0.000 description 1
- 230000037431 insertion Effects 0.000 description 1
- 238000012417 linear regression Methods 0.000 description 1
- 239000011159 matrix material Substances 0.000 description 1
- 230000007246 mechanism Effects 0.000 description 1
- 230000006855 networking Effects 0.000 description 1
- 230000001537 neural effect Effects 0.000 description 1
- 238000003062 neural network model Methods 0.000 description 1
- 210000000056 organ Anatomy 0.000 description 1
- 238000011084 recovery Methods 0.000 description 1
- 230000001629 suppression Effects 0.000 description 1
- 238000005303 weighing Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR 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
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/25—Fusion techniques
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR 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; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/40—Extraction of image or video features
- G06V10/46—Descriptors for shape, contour or point-related descriptors, e.g. scale invariant feature transform [SIFT] or bags of words [BoW]; Salient regional features
- G06V10/462—Salient features, e.g. scale invariant feature transforms [SIFT]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR 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
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR 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
- G06V40/172—Classification, e.g. identification
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
- G06V40/20—Movements or behaviour, e.g. gesture recognition
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L67/00—Network arrangements or protocols for supporting network services or applications
- H04L67/01—Protocols
- H04L67/12—Protocols specially adapted for proprietary or special-purpose networking environments, e.g. medical networks, sensor networks, networks in vehicles or remote metering networks
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Health & Medical Sciences (AREA)
- Physics & Mathematics (AREA)
- General Health & Medical Sciences (AREA)
- Multimedia (AREA)
- Business, Economics & Management (AREA)
- Oral & Maxillofacial Surgery (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Human Computer Interaction (AREA)
- Tourism & Hospitality (AREA)
- Data Mining & Analysis (AREA)
- Psychiatry (AREA)
- Life Sciences & Earth Sciences (AREA)
- Evolutionary Computation (AREA)
- Evolutionary Biology (AREA)
- Bioinformatics & Computational Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Social Psychology (AREA)
- Computer Security & Cryptography (AREA)
- Artificial Intelligence (AREA)
- Development Economics (AREA)
- Educational Administration (AREA)
- General Engineering & Computer Science (AREA)
- Economics (AREA)
- Human Resources & Organizations (AREA)
- Marketing (AREA)
- Primary Health Care (AREA)
- Strategic Management (AREA)
- General Business, Economics & Management (AREA)
- Computing Systems (AREA)
- Medical Informatics (AREA)
- Computer Networks & Wireless Communication (AREA)
- Signal Processing (AREA)
- Image Analysis (AREA)
Abstract
The invention discloses a kind of community's Risk Monitoring and prevention method based on Internet of Things and deep learning, comprising the following steps: S1: build " cloud-net-end " integrated community's Risk Monitoring and guarding network framework；S2: depth coding is carried out to community's Risk Monitoring big data；S3: the monitored personnel identity based on limit study and face characteristic identifies；S4: cascade timing convolutional neural networks are built and realize monitored human behavior analysis；S5: isomery decision information fusion of the depth under encoding is realized.
Description
Technical field
The present invention relates to the fields that Internet of Things and neural network manage community policy, more particularly to use " cloud-net-end "
The method that integrated network framework is combined with the intelligent cognition technology under deep learning promotes community's Risk Monitoring and prevention energy
Power.
Background technique
Currently, with the development of the social economy, the continuous aggravation of urbanization process, urban population largely increase, society at
Divide complexity, interpersonal relationships desalination provides crime chance to offender, the criminal case specific gravity of community is caused to increase.Therefore
Reinforce Community Police Affairs construction, reinforce community policy prevention, combine punishment and prevention, with emphasis on the latter, combine the efforts of both professionals and the masses, by the comprehensive of the masses
Close control policy, construction community's Risk Monitoring with prevention platform be solve the problems, such as community policy, having of maintaining social stability
Effect approach.Peaceful community is created, fundamentally improves local security condition, enhances the sense of security, the happiness of the people
With reduction area under one's jurisdiction case involving public security, criminal case incidence.
Summary of the invention
In view of this, the purpose of the present invention is to provide a kind of community's Risk Monitoring based on Internet of Things and deep learning and
Prevention method, it is intended to promote community's Risk Monitoring and prevention ability.
In order to achieve the above objectives, the invention provides the following technical scheme:
A kind of community's Risk Monitoring and prevention method based on Internet of Things and deep learning, comprising the following steps:
S1: " cloud-net-end " integrated community's Risk Monitoring and guarding network framework are built；
S2: depth coding is carried out to community's Risk Monitoring big data；
S3: the monitored personnel identity based on limit study and face characteristic identifies；
S4: cascade timing convolutional neural networks are built and realize monitored human behavior analysis；
S5: isomery decision information fusion of the depth under encoding is realized.
Further, the step S1 specifically: community's Risk Monitoring and guarding network framework include cloud, net and end；
The cloud: referring mainly to fusion police service, government affairs, estate management, comprehensive treatment integration cloud platform, in comprehensive cloud platform
Upper collection helps that interconnection Information Sharing Technology, semantic-based community's basic information be integrated and service mechanism, to community's Risk Monitoring
The ubiquitous information that Internet of Things obtains carries out Multi-source Information Fusion, provides service for upper layer information shared platform.
The net: being made of networks such as 3G/4G, NB-IoT, WiFi, and the heterogeneous networks at scene are adapted to using borde gateway,
It realizes that the interconnection of heterogeneous networks is integrated using IPv6 technology, realizes that the corporate management of heterogeneous network and inter-network are dispatched using SDN.
The end: being made of community's Risk Monitoring facility and Internet of things node, is acquired using Internet of things node and obtains covering
People from community, thing, the basic datas of object, feelings.
Further, the step S2 specifically: community's Risk Monitoring big data depth coding includes big data analysis
With data encoding；
The big data analysis: it by data pick-up, cleaning, conversion, loads and realizes big data pretreatment.HBase is utilized
HDFS is as big data storage system storing data, and wherein HBase is located at structured storage layer, and HDFS is provided for highly reliable
Property bottom storage.In HDFS, file is cut into data block, usual each data block 64MB~128MB, then every number
It is written into file system according to block, the different data block of same file is not necessarily stored on identical DataNode.It uses
Frame of the MapReduce as big data computing engines, wherein MapRduce includes two functions of Map and Reduce, is suitble to use
In batch processing, non real-time and data-intensive situation.Using the MLlib of Spark as big data analysis computing engines.
Wherein Spark is an open source Computational frame memory-based, can quickly handle the big data problem under several scenes, can be efficiently
The value in big data is excavated, to provide decision support for business development.MLlib is that Spark calculates common machine learning
The realization library of method, while including relevant test and Data Generator.MLlib supports four kinds of common machine learning to ask at present
Topic: binary classification returns, cluster and collaborative filtering, while also including the gradient decline optimization basic algorithm an of bottom.
The data encoding:, bus transfer load capacity limited problem big for multi-modal monitoring data total amount, research
Using depth from coding techniques as the data coding mode of core, pass through the equal hierarchical structure of depth self-encoding encoder input and output and close
System, training are suitable for the depth of multi-modal sensor from coding structure, extract perception data common feature, as encoded information,
By bus transfer to vehicle-mounted data interaction center, the dimension of transmission data is reduced, information traffic load is flat in guarantee bus
Weighing apparatus.Depth is from encoding lower data encoding structures, based on multilayer deep neural network structure, by the input of neural network and
Output is set to equal data and encodes to intermediate combination, using least square objective function:
Wherein, the corresponding perception input layer of x,It is then the result of perception reduction.W, b respectively correspond the power of the autoencoder network
Will set and biasing.The target of the network training is to allow output result to approach perception input results as far as possible, and use a large amount of sounds
The initial data such as video, temperature, stochastic gradient descent technology learn network connection, improve the Generalization Capability of weight.Root
According to the structure of depth self-encoding encoder, coding result can restore the result of perception by coding reduction zone processing.It is self-editing by this
Code device encodes perception data, and the load of the dimension and bus of transmission data can be greatly reduced, promote bus transfer
Efficiency.Simultaneously at data interaction center, it is only necessary to first encoding reduction zone is sent to data center, it can be by it to coding
As a result reconstruct obtains the result data of perception.
For the monitoring data interaction center problem not strong for coding information data real-time, research is optimized for sparse
The neural networks pruning technology of core, by beta pruning depth self-encoding encoder network structure, the lesser connection relationship of weighted value, greatly
Width reduces the connection structure of depth self-encoding encoder, and a small amount of perception data is finely adjusted the connection of remaining neural network, drops
Low biography encoding error improves the operational efficiency of depth self-encoding encoder.
Reduction is iterated using absolute value of the coordinate descending method to the weight of neural network, and nearly zero parameter carries out
Beta pruning, and retrieve connection relationship more sparse network.Identical weight is all used to all layers in neural metwork training
Sparse optimization algorithm usesSparse Optimized model optimizes weight connection W,
Wherein W' is the weighed combination after rarefaction, and the input that i is each layer, W is to train obtained weight from it is encoded through depth.On
The solution for stating sparse Optimized model is a convex optimization problem, optimization | | W'| |1With optimization | | W'| |0Result it is of equal value, therefore can root
It is right according to coordinate descent algorithm | | W'| |1The result optimized obtains a W ' and makes | | W ' | |0、Solution is wrapped
Containing W', i.e. W' the most sparse, nearly zero in W' is carried out to beta pruning in neural network can be obtained updated network knot
Structure.
The connection relationship of neural network will be more simple after beta pruning, accelerates the efficiency of operation, reduces noise jamming.Pass through
Perception data collection further finely tunes network, extensive from encoding model after can be obtained beta pruning.Sparse optimization beta pruning
Model is combined optimization to the result of product and the neural network output of connection relationship and input after beta pruning, to greatest extent
Coding restore accuracy and real-time between acquired balance.
Further, the step S3 specifically: monitored personnel identity identifies comprising the design of limit learning algorithm and face
Feature identification；
The limit learning algorithm design: limit learning algorithm is extensive Single hidden layer feedforward neural networks, is carrying out people
When face identifies, which has invariance to dull grey scale change and angle rotation, has to image change caused by uneven illumination
There is insensitivity.In identification process, the human face in target area is fitted with characteristic point, obtains characteristic point label
Position.Subgraph is intercepted in each characteristic point contiguous range, obtains human face adjacent features, it is finally that all characteristic points are neighbouring
Feature series connection constitutive characteristic point limit learning characteristic.The limit learning characteristic for counting each picture portion respectively is hidden as extensive list
The training set of layer feedforward neural network, the multiple extreme learning machines of training, the output of combination feedforward neural network as a result,
The face characteristic identification: it by radial symmetry transform coarse positioning face, is learned using supervision gradient descent method (SDM)
Acquistion establishes shaped Offset amount Δ x=x to current point to the optimal iterative vectorized of target point*The feature of-x and current shape xBetween linear regression model (LRM)Then current shape x and deformation vectors Δ x iteration are utilized
Obtain desired position vector x:=x+ Δ x.Construct the learning objective of SDM:
Wherein, k is the number of iterations, xkIndicate shape vector when iterating to kth time,Indicate in the shape vector
The coordinate of i point.Successive ignition study is carried out, obtains the true deviation of with actual boundary point at i-th point.Then calculus is used
Operator:
The organs such as eyes, nose, the mouth in face are accurately positioned, wherein GσIt (r) is smooth function, I (x, y) is image ash
Matrix is spent, (a, b) is the center of circle, and r is radius.
Further, the step S4 specifically: be intended to behavioural analysis for the personnel in community intelligent risk monitoring and control field
Demand designs the depth level with head and shoulder identification function using non-linear movement pattern study and the more case-based learnings of display model
Networking network realizes the human behavior intention analysis of layering association multiple target tracking learning strategy driving.Building has head and shoulder identification
Depth cascade network (HsNet), a series of candidates are intercepted according to pre- fixed step size using multi-scale sliding window mouth to every frame image
Segment (Patch), forms sample to be identified；These samples are sent into trained head and shoulder/non-head and shoulder identification model HsNet in advance
(three-level CNN cascade network) classifies.In specific assorting process, the Patch for being judged as negative sample directly gives up, and remains
The next stage that remaining sample goes successively to network carries out tightened up identification classification, so successively carries out three-level CNN network class mirror
Not；Head and shoulder frame height degree is extended to original for judging whether image Patch belongs to head and shoulder region by the output result of the network third level
3 times of corresponding sliding window obtain the whole body frame of monitored personnel's detection；For the same monitored personnel, will form multiple
Detection block, finally rejects extra detection block with non-maxima suppression strategy, and each position only retains a most probable detection
Frame-occupant detection recognition result.
Non-linear movement pattern study and the more examples of display model are introduced in layering association multiple target tracking learning strategy
Study forms path segment by carrying out the credible association of bottom to test object；Using non-linear movement pattern on-line study and
The more case-based learnings of display model effectively connect path segment, obtain reliable object trajectory.Rail is moved using from object
Multiple features are combined as feature and form more advanced semanteme to describe by the parameters such as speed, direction, the distance extracted in mark
Object behavior, to judge that monitored human behavior is intended to.
For the robustness for improving behavioral value, the passenger attitude frame being consecutively detected is fused into a complete action row
For.Design two posture fusion rules are as follows:
Wherein f (i, j) is fusion function, and 1 indicates to merge, and 0 indicates to merge,For two attitude detection frame registrations, TIoU=0.5 indicates weight
Right threshold value, ShisFor Histogram Matching score in two detection blocks, This=35 be Histogram Matching threshold value, TΔ=25 indicate two appearances
State time difference threshold value.
Further, the step S5 specifically: assuming that any two groups of multi-modal inputs are respectively x and y, input data dimension
Respectively n and j.Input 1, by independent neural network model, mentions input feature vector using full connection type with input 2
Take and then obtain the characteristic information of feature 1 and feature 2.By being pressed to feature 1 and the characteristic information of feature 2 using full connection type
According to Dynamic Weights objective making decision Rule decision information.To obtain the information of decision 1 Yu decision 2.Then according to decision 1 with
The information of decision 2 uses Weighted Fusion algorithm, fusion decision information obtained after being merged.According to fusion decision information,
Decision 1 is carried out respectively to connect recovery entirely with decision 2, then successively hidden layer characteristic information is restored, finally to input
Information carries out restoring the output information as the self-encoding encoder.During being updated for this from coding structure, by defeated
Enter information as evaluation condition, is carried out using back-propagation algorithm according to output information with the difference of input information corresponding to network
Weight is adjusted using gradient descent algorithm, obtains one group of preferable weight data.Then, positive according to depth confidence network
The thought of weight fine tuning, successively finely tunes the weight of the network, finally obtains the weight of one group of robust.Pass through this group power
Value, calculates different inputs, can be obtained fused value information.The isomery letter of policy-making authority is carried out by this method
Breath fusion can reach the policy-making authority effective integration for heterogeneous multi-source, multi-dimensional data, to obtain final decision information.
The beneficial effects of the present invention are: the intelligent cognition skill under depth integration technology of Internet of things of the present invention and deep learning
Art effectively promotes community's Risk Monitoring and prevention ability, creation safety in the helpfulness insertion Internet of Things framework of intelligent cognition
Community environment.
Detailed description of the invention
Fig. 1 is " cloud-net-end " integrated community's Risk Monitoring and guarding network architecture diagram.
Specific embodiment
In order to keep the purpose of the present invention, technical scheme and beneficial effects clearer, below in conjunction with Fig. 1, to " cloud-net-
The integrated community's Risk Monitoring in end " is described in detail with guarding network framework.
1, " cloud-net-end " integrated community's Risk Monitoring and guarding network framework are built, which includes following 3
Step:
(1) " end " is formed by community's Risk Monitoring facility and Internet of things node；
(2) " net " is formed by networks such as 3G/4G, NB-IoT, WiFi；
(3) above-mentioned steps are based on, police service, government affairs, estate management, comprehensive treatment integration information are merged, form cloud platform.
2, depth coding is carried out to community's Risk Monitoring big data, which includes following 5 steps:
(1) community's risk data information is collected using " end " in step 1.
(2) big data processing is carried out to community's risk information.
(3) multilayer deep neural network is built, training obtains depth self-encoding encoder.
(4) by the way that in beta pruning depth self-encoding encoder network structure, depth is greatly reduced certainly in the lesser connection relationship of weighted value
The connection structure of encoder improves the operational efficiency of depth self-encoding encoder.
(5) by treated, big data input depth self-encoding encoder is encoded.
3, the personnel identity that is monitored identifies, which includes following 3 steps:
(1) face picture collection is acquired.
(2) the extensive Single hidden layer feedforward neural networks based on extreme learning machine are designed.
(3) using the face pictures training extensive Single hidden layer feedforward neural networks, personnel identity discriminator is obtained.
4, be monitored human behavior analysis, which includes following 3 steps:
(1) pedestrian behavior pictures are acquired.
(2) three-level depth cascade network is built.
(3) using the pedestrian behavior pictures training three-level depth cascade network, human behavior analyzer is obtained.
5, multi-source heterogeneous information fusion and decision, the process include following 3 steps:
(1) multi-source heterogeneous neural network is built, and using the training of multi-source heterogeneous information, obtains information fusion and decision model
Type.
(2) using 2,3,4 acquired results information of above-mentioned steps as the input of information fusion and decision model, output is society
Area's Risk Results.
The above display describes basic principles and main features and advantages of the present invention of the invention.The technology people of the industry
Member is it should be appreciated that the present invention is not limited to the above embodiments, and the above embodiments and description only describe this hairs
Bright principle, without departing from the spirit and scope of the present invention, various changes and improvements may be made to the invention, these variations
It all fall within the protetion scope of the claimed invention with improvement.The claimed scope of the invention is by appended claims and its waits
Effect object defines.
Claims (6)
1. a kind of community's Risk Monitoring and prevention method based on Internet of Things and deep learning, comprising the following steps:
S1: " cloud-net-end " integrated community's Risk Monitoring and guarding network framework are built；
S2: depth coding is carried out to community's Risk Monitoring big data；
S3: the monitored personnel identity based on limit study and face characteristic identifies；
S4: cascade timing convolutional neural networks are built and realize monitored human behavior analysis；
S5: isomery decision information fusion of the depth under encoding is realized.
2. a kind of community's Risk Monitoring and prevention method based on Internet of Things and deep learning according to claim 1,
It is characterized in that step S1 specifically: community's Risk Monitoring and guarding network framework include cloud, net and end, the cloud: main
Refer to fusion police service, government affairs, estate management, comprehensive treatment integration cloud platform, the net: by nets such as 3G/4G, NB-IoT, WiFi
Network composition, the end: is made of community's Risk Monitoring facility and Internet of things node.
3. a kind of community's Risk Monitoring and prevention method based on Internet of Things and deep learning according to claim 1,
It is characterized in that the step S2 specifically: community's Risk Monitoring big data depth coding includes that big data analysis and data are compiled
The big data analysis: code by data pick-up, cleaning, conversion, loads and realizes big data pretreatment.
4. a kind of community's Risk Monitoring and prevention method based on Internet of Things and deep learning according to claim 1,
It is characterized in that the step S3 specifically: monitored personnel identity, which identifies, to be known comprising the design of limit learning algorithm with face characteristic
Not, the limit learning algorithm design: limit learning algorithm is extensive Single hidden layer feedforward neural networks, and the face characteristic is known
It is other: by radial symmetry transform coarse positioning face, current point to be obtained using supervision gradient descent method (SDM) study and arrives target point
It is optimal iterative vectorized.
5. a kind of community's Risk Monitoring and prevention method based on Internet of Things and deep learning according to claim 1,
It is characterized in that the step S4 is specially to use non-linear movement pattern study and the more case-based learnings of display model, design has head
The depth cascade network of shoulder identification function realizes the human behavior intention point of layering association multiple target tracking learning strategy driving
Analysis.
6. a kind of community's Risk Monitoring and prevention method based on Internet of Things and deep learning according to claim 1,
It is characterized in that the step S5 specifically: assuming that any two groups of multi-modal inputs are respectively x and y, input data dimension is respectively n
With j.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810765316.8A CN109034020A (en) | 2018-07-12 | 2018-07-12 | A kind of community's Risk Monitoring and prevention method based on Internet of Things and deep learning |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810765316.8A CN109034020A (en) | 2018-07-12 | 2018-07-12 | A kind of community's Risk Monitoring and prevention method based on Internet of Things and deep learning |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN109034020A true CN109034020A (en) | 2018-12-18 |
Family
ID=64641053
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810765316.8A Pending CN109034020A (en) | 2018-07-12 | 2018-07-12 | A kind of community's Risk Monitoring and prevention method based on Internet of Things and deep learning |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109034020A (en) |
Cited By (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110009539A (en) * | 2019-04-12 | 2019-07-12 | 烟台工程职业技术学院(烟台市技师学院) | A kind of student is in school learning state smart profile system and application method |
| CN110119779A (en) * | 2019-05-10 | 2019-08-13 | 北京科技大学 | Any dimension fusion method of across a network data and device based on self-encoding encoder |
| CN110225067A (en) * | 2019-07-24 | 2019-09-10 | 上海戎磐网络科技有限公司 | A kind of Internet of Things safety pre-warning system |
| CN110459027A (en) * | 2019-08-15 | 2019-11-15 | 青岛文达通科技股份有限公司 | A kind of Community Safety means of defence and system based on multi-source heterogeneous data fusion |
| CN111126324A (en) * | 2019-12-25 | 2020-05-08 | 深圳力维智联技术有限公司 | Method, device, product and medium for multi-source heterogeneous data fusion |
| CN111669396A (en) * | 2020-06-15 | 2020-09-15 | 绍兴文理学院 | Self-learning security defense method and system for software-defined Internet of things |
| CN112507312A (en) * | 2020-12-08 | 2021-03-16 | 电子科技大学 | Digital fingerprint-based verification and tracking method in deep learning system |
| CN113518114A (en) * | 2021-05-12 | 2021-10-19 | 江苏力行电力电子科技有限公司 | Artificial intelligence control method and system based on multi-mode ad hoc network |
| CN113919544A (en) * | 2021-08-30 | 2022-01-11 | 浙江警察学院 | Crime early warning method and device, computer equipment and storage medium |
| CN117725493A (en) * | 2023-10-16 | 2024-03-19 | 中国矿业大学(北京) | Landslide hazard result evaluation method based on improved Stacking fusion model |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101997921A (en) * | 2010-11-22 | 2011-03-30 | 青岛龙泰天翔通信科技有限公司 | Networking based novel intelligent network application system |
| CN102891867A (en) * | 2011-07-19 | 2013-01-23 | 苏州科技学院 | Community IoT (Internet of Things) system based on Zigbee and access control system |
| CN107992904A (en) * | 2017-12-22 | 2018-05-04 | 重庆邮电大学 | Forest Ecology man-machine interaction method based on Multi-source Information Fusion |
| CN108256307A (en) * | 2018-01-12 | 2018-07-06 | 重庆邮电大学 | A kind of mixing enhancing intelligent cognition method of intelligent business Sojourn house car |
-
2018
- 2018-07-12 CN CN201810765316.8A patent/CN109034020A/en active Pending
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101997921A (en) * | 2010-11-22 | 2011-03-30 | 青岛龙泰天翔通信科技有限公司 | Networking based novel intelligent network application system |
| CN102891867A (en) * | 2011-07-19 | 2013-01-23 | 苏州科技学院 | Community IoT (Internet of Things) system based on Zigbee and access control system |
| CN107992904A (en) * | 2017-12-22 | 2018-05-04 | 重庆邮电大学 | Forest Ecology man-machine interaction method based on Multi-source Information Fusion |
| CN108256307A (en) * | 2018-01-12 | 2018-07-06 | 重庆邮电大学 | A kind of mixing enhancing intelligent cognition method of intelligent business Sojourn house car |
Non-Patent Citations (3)
| Title |
|---|
| 徐大成: "《基于物联网和云计算的智慧园区信息系统的研究与实现》", 《中国优秀硕士学位论文全文数据库 信息科技辑》 * |
| 王梦来等: "《基于CNN的监控视频事件检测》", 《自动化学报》 * |
| 蔡良辰: "《"互联网+"社区治理研究—以滨湖世纪智慧社区建设为例》", 《中国优秀硕士学位论文全文数据库 社会科技Ⅰ辑》 * |
Cited By (15)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110009539A (en) * | 2019-04-12 | 2019-07-12 | 烟台工程职业技术学院(烟台市技师学院) | A kind of student is in school learning state smart profile system and application method |
| CN110119779B (en) * | 2019-05-10 | 2022-03-04 | 北京科技大学 | Cross-network data arbitrary dimension fusion method and device based on self-encoder |
| CN110119779A (en) * | 2019-05-10 | 2019-08-13 | 北京科技大学 | Any dimension fusion method of across a network data and device based on self-encoding encoder |
| CN110225067A (en) * | 2019-07-24 | 2019-09-10 | 上海戎磐网络科技有限公司 | A kind of Internet of Things safety pre-warning system |
| CN110225067B (en) * | 2019-07-24 | 2021-08-24 | 上海戎磐网络科技有限公司 | Internet of things safety early warning system |
| CN110459027A (en) * | 2019-08-15 | 2019-11-15 | 青岛文达通科技股份有限公司 | A kind of Community Safety means of defence and system based on multi-source heterogeneous data fusion |
| CN111126324A (en) * | 2019-12-25 | 2020-05-08 | 深圳力维智联技术有限公司 | Method, device, product and medium for multi-source heterogeneous data fusion |
| CN111126324B (en) * | 2019-12-25 | 2023-09-26 | 深圳力维智联技术有限公司 | Method, device, computer equipment and medium for multi-source heterogeneous data fusion |
| CN111669396A (en) * | 2020-06-15 | 2020-09-15 | 绍兴文理学院 | Self-learning security defense method and system for software-defined Internet of things |
| CN112507312A (en) * | 2020-12-08 | 2021-03-16 | 电子科技大学 | Digital fingerprint-based verification and tracking method in deep learning system |
| CN112507312B (en) * | 2020-12-08 | 2022-10-14 | 电子科技大学 | Digital fingerprint-based verification and tracking method in deep learning system |
| CN113518114A (en) * | 2021-05-12 | 2021-10-19 | 江苏力行电力电子科技有限公司 | Artificial intelligence control method and system based on multi-mode ad hoc network |
| CN113919544A (en) * | 2021-08-30 | 2022-01-11 | 浙江警察学院 | Crime early warning method and device, computer equipment and storage medium |
| CN113919544B (en) * | 2021-08-30 | 2022-05-27 | 浙江警察学院 | Crime early warning method and device, computer equipment and storage medium |
| CN117725493A (en) * | 2023-10-16 | 2024-03-19 | 中国矿业大学(北京) | Landslide hazard result evaluation method based on improved Stacking fusion model |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109034020A (en) | A kind of community's Risk Monitoring and prevention method based on Internet of Things and deep learning | |
| CN112418117B (en) | Small target detection method based on unmanned aerial vehicle image | |
| Wang et al. | A real-time collision prediction mechanism with deep learning for intelligent transportation system | |
| CN110458844B (en) | Semantic segmentation method for low-illumination scene | |
| Kurniawan et al. | Traffic congestion detection: learning from CCTV monitoring images using convolutional neural network | |
| CN110781838A (en) | Multi-modal trajectory prediction method for pedestrian in complex scene | |
| CN110084151B (en) | Video abnormal behavior discrimination method based on non-local network deep learning | |
| CN111079674B (en) | Target detection method based on global and local information fusion | |
| CN107133569A (en) | The many granularity mask methods of monitor video based on extensive Multi-label learning | |
| CN105354273A (en) | Method for fast retrieving high-similarity image of highway fee evasion vehicle | |
| CN106228137A (en) | A kind of ATM abnormal human face detection based on key point location | |
| CN109801260A (en) | The recognition methods of livestock number and device | |
| CN111476285B (en) | Training method of image classification model, image classification method and storage medium | |
| CN106682681A (en) | Recognition algorithm automatic improvement method based on relevance feedback | |
| CN112766280A (en) | Remote sensing image road extraction method based on graph convolution | |
| CN109376613A (en) | Video brainpower watch and control system based on big data and depth learning technology | |
| CN107977683A (en) | Joint SAR target identification methods based on convolution feature extraction and machine learning | |
| CN105930792A (en) | Human action classification method based on video local feature dictionary | |
| Shen et al. | An attention-based digraph convolution network enabled framework for congestion recognition in three-dimensional road networks | |
| Nawaratne et al. | A generative latent space approach for real-time road surveillance in smart cities | |
| Gorodnichev et al. | Research and Development of a System for Determining Abnormal Human Behavior by Video Image Based on Deepstream Technology | |
| CN114758178A (en) | Hub real-time classification and air valve hole positioning method based on deep learning | |
| Wang et al. | Vision-based highway traffic accident detection | |
| Vu et al. | Traffic incident recognition using empirical deep convolutional neural networks model | |
| Lukic et al. | Galaxy classifications with 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 | ||
| WD01 | Invention patent application deemed withdrawn after publication | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20181218 |