CN109145823A - A kind of market monitoring device - Google Patents
A kind of market monitoring device Download PDFInfo
- Publication number
- CN109145823A CN109145823A CN201810961756.0A CN201810961756A CN109145823A CN 109145823 A CN109145823 A CN 109145823A CN 201810961756 A CN201810961756 A CN 201810961756A CN 109145823 A CN109145823 A CN 109145823A
- Authority
- CN
- China
- Prior art keywords
- image
- layer
- indicate
- processing module
- processing
- 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.)
- Withdrawn
Links
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
- G06V40/20—Movements or behaviour, e.g. gesture recognition
-
- 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
- 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
- G06V10/44—Local feature extraction by analysis of parts of the pattern, e.g. by detecting edges, contours, loops, corners, strokes or intersections; Connectivity analysis, e.g. of connected components
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N7/00—Television systems
- H04N7/18—Closed-circuit television [CCTV] systems, i.e. systems in which the video signal is not broadcast
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Multimedia (AREA)
- General Physics & Mathematics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Data Mining & Analysis (AREA)
- General Engineering & Computer Science (AREA)
- Signal Processing (AREA)
- Evolutionary Biology (AREA)
- Bioinformatics & Computational Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Artificial Intelligence (AREA)
- Life Sciences & Earth Sciences (AREA)
- Evolutionary Computation (AREA)
- Health & Medical Sciences (AREA)
- General Health & Medical Sciences (AREA)
- Psychiatry (AREA)
- Social Psychology (AREA)
- Human Computer Interaction (AREA)
- Closed-Circuit Television Systems (AREA)
Abstract
The present invention provides a kind of market monitoring devices, including image capture device, information transmission equipment and monitor supervision platform；Described image acquires equipment and information transmission equipment communication connection；The information transmission equipment and the monitor supervision platform communicate to connect；Described image acquisition equipment real-time image acquisition simultaneously sends it to the information transmission equipment；The image received is sent to the monitor supervision platform by the information transmission equipment；The monitor supervision platform is for detecting the act of violence in image.The invention has the benefit that providing a kind of market monitoring device, the act of violence in image can be detected, discovery is fought in time.
Description
Technical field
The present invention relates to monitoring technology fields, and in particular to a kind of market monitoring device.
Background technique
With science and technology be constantly progressive and continuous improvement of people's living standards, monitoring technology have become guarantee quotient
One of the essential technological means of field safety.
Summary of the invention
In view of the above-mentioned problems, the present invention provides a kind of market monitoring device.
The purpose of the present invention is realized using following technical scheme:
Provide a kind of market monitoring device, including image capture device, information transmission equipment and monitor supervision platform；
Described image acquires equipment and information transmission equipment communication connection；
The information transmission equipment and the monitor supervision platform communicate to connect；
Described image acquisition equipment real-time image acquisition simultaneously sends it to the information transmission equipment；
The image received is sent to the monitor supervision platform by the information transmission equipment；
The monitor supervision platform is for detecting the act of violence in image.
The invention has the benefit that providing a kind of market monitoring device, the act of violence in image can be carried out
Detection, discovery is fought in time.
Detailed description of the invention
The present invention will be further described with reference to the accompanying drawings, but the embodiment in attached drawing is not constituted to any limit of the invention
System, for those of ordinary skill in the art, without creative efforts, can also obtain according to the following drawings
Other attached drawings.
Fig. 1 is structural schematic diagram of the invention；
Appended drawing reference:
Image capture device 1, information transmission equipment 2, monitor supervision platform 3.
Specific embodiment
The invention will be further described with the following Examples.
Referring to Fig. 1, a kind of market monitoring device of the present embodiment, including image capture device 1,2 and of information transmission equipment
Monitor supervision platform 3；
Described image acquires equipment 1 and information transmission equipment 2 communicates to connect；
The information transmission equipment 2 is communicated to connect with the monitor supervision platform 3；
Described image acquisition 1 real-time image acquisition of equipment simultaneously sends it to the information transmission equipment 2；
The image received is sent to the monitor supervision platform 3 by the information transmission equipment 2；
The monitor supervision platform 3 is for detecting the act of violence in image.
A kind of market monitoring device is present embodiments provided, the act of violence in image can be detected, is sent out in time
Now fight.
Preferably, the monitor supervision platform 3 include first processing module, Second processing module, third processing module, the everywhere
Module, the 5th processing module, the 6th processing module are managed, the first processing module is for inputting image detected, and described the
Two processing modules are used to extract the global characteristics of image, and the third processing module is used for extracted image overall feature
In depth network model, the fourth processing module determines violence testing result based on third processing module for fusion, described the
Five processing modules are used to export the violence of optimization for optimizing fourth processing module violence testing result, the 6th processing module
Testing result.
This preferred embodiment monitor supervision platform 3 effectively raises the accuracy rate of violence detection.
Preferably, the Second processing module includes the first process layer, second processing layer, third process layer, fourth process
Layer；First process layer pre-processes the image of input；The second processing layer is filtered to image and convolution
Operation；The output result of second processing layer is done Nonlinear Mapping by the third process layer；The fourth process layer is for compressing
Image after Nonlinear Mapping；
In second processing layer, by convolution operation to pretreated image zooming-out local neighborhood feature, by multilayer
Iteration extracts the global characteristics of image by two-dimensional convolution:
In formula,Indicate the activation value in i-th layer of j-th of Feature Mapping at the position (x, y), this activation value
It is exactly the two-dimentional global characteristics of image；YW () indicates activation primitive, wherein H, W respectively indicate the height of two-dimensional convolution core, width
The size of degree；Indicate the weight of convolution kernel,Indicate (i-1)-th layer of d-th of Feature Mapping at (x, y)
Activation value, EMijIndicate that bias vector, i indicate that the convolutional layer that image is currently located, j indicate the Feature Mapping quantity of this layer.
This preferred embodiment Second processing module by two-dimensional convolution can easily abstract image spatial information, letter
Just, application range is most wide for folk prescription, but is not sufficient to carry out expressed intact to video merely with these appearance features, can make video
It is lacked.
Preferably, the two-dimensional convolution core in Second processing module is generated three by spatial spread by the third processing module
Convolution kernel is tieed up, the Three dimensional convolution at pixel (x, y, z) calculates is defined as:
In formula,Indicate the activation value in i-th layer of j-th of Feature Mapping at the position (x, y, z)；This activation
Value is exactly the three-dimensional global characteristics of image；YW () indicates activation primitive, wherein H, W, T respectively indicate the height of three dimensional convolution kernel
Size on degree, width and time dimension；Indicate the weight of convolution kernel,Indicate -1 layer of d of jth
Activation value of a Feature Mapping at (x, y, z), EMijIndicate that bias vector, i indicate that the convolutional layer that image is currently located, j indicate
The Feature Mapping quantity of this layer.
This preferred embodiment third processing module is compared with two-dimensional convolution formula, and Three dimensional convolution is to convolution kernel and pixel
Expression on both increase time dimension.After convolution kernel is extended to three-dimensional space, when carrying out convolution to image sequence, convolution
Operation will carried out spatially and temporally simultaneously, and in this way after the operation of convolution sum pondization, the characteristic pattern of output remains image
Sequence can be very good to retain the space time information in video.By the feature extraction of multiple Three dimensional convolutions, so that it may extract view
The global space-time characteristic of frequency.
Preferably, the fourth processing module be based on third processing module use tri- second processing layers of C1, C2, C3, C1,
The three dimensional convolution kernel size that C2 and C3 are used is respectively 7 × 7 × 5,5 × 5 × 5 and 3 × 3 × 3 pixels；Fourth processing module it is defeated
Enter the image segments X to be made of 40 frame consecutive images；Picture frame is normalized to 60 × 90 pixel sizes after pretreatment
And be converted to grayscale image；Scalar Y scalar is exported, for indicating testing result that model inputs image, for trained mould
Type, if in test image including Violent scene, output Y is 1, and otherwise exporting result is 0；
The fourth processing module carries out pondization operation, Chi Huatong to the characteristic pattern that the first two second processing layer is calculated
Cross following formula calculating:
In formula,Indicate y-th of characteristic pattern of x layer,Indicate y-th of characteristic pattern of x-1 layer, θ and B are respectively
The biasing of multiplying property and additivity biasing,Indicate y-th of the multiplying property biasing of x layer,Indicate y-th of the additivity biasing of x layer, δTTo sample letter
Number,Wherein, t is the time, and T is the sampling period, and n ∈ [0 ,+∞] and n are positive integer；
The pondization operation does not carry out input feature vector graphic sequence in time dimension down-sampled using two-dimentional pondization operation
Operation, the pond factor are set to 3 × 3 and 2 × 2 pixels；
The fourth processing module is used during model training using minor function as cost function:
In formula, G is pattern function, and θ is model parameter, and X is training sample, and N is sample size, andIt is sample reality
Border label, k ∈ [1, N], N ∈ [1 ,+∞], CS1(X, θ) indicates fourth processing module cost function；The smaller table of cost function value
It is better that bright model is fitted with training set；
On the one hand this preferred embodiment fourth processing module can be further reduced network parameter, on the other hand also give
The characteristics such as characteristic pattern translation invariant and invariable rotary, so that the feature acquired is more robust.
Preferably, the 5th processing module is based on fourth processing module, and input is that 40 frames of 128 × 128 pixels connect
Continuous image, consecutive image are Three Channel Color image；
Three dimensional convolution kernel is uniformly set as to 3 × 3 × 3 pixels, in convolution operation, the 5th processing module carries out characteristic pattern
Padding, so that the size before the characteristic pattern obtained after convolution and calculating holding；Also using three-dimensional during pond
Pondization operation carries out down-sampled operation to input feature vector graphic sequence in time dimension, the pond factor is set as 2 × 2 × 2 pixels；
Cost function of 5th processing module during model training is chosen with minor function:
In formula, G is pattern function, and θ is model parameter, XkFor k-th of training sample, m is classification number, and N is every class
Sample number,It is k-th of data physical tags；K ∈ [1, N], N ∈ [1 ,+∞], l ∈ [1, m], m ∈ [1 ,+∞], CS2(X, θ)
Indicate the 5th processing module cost function.
The 5th processing module of this preferred embodiment uses more complicated structure, therefore the image data dimension handled can
With higher, the extraction of image temporal information can be accelerated in this way, remove bulk redundancy therein.
Violence detection is carried out using market monitoring device of the present invention, 5 markets is chosen and is tested, respectively market 1, quotient
Field 2, market 3, market 4, market 5 count Detection accuracy and detection speed, compare compared with monitoring device, generate
Have the beneficial effect that shown in table:
| Detection accuracy improves | Speed is detected to improve | |
| Market 1 | 29% | 27% |
| Market 2 | 27% | 26% |
| Market 3 | 26% | 26% |
| Market 4 | 25% | 24% |
| Market 5 | 24% | 22% |
Through the above description of the embodiments, those skilled in the art can be understood that it should be appreciated that can
To realize the embodiments described herein with hardware, software, firmware, middleware, code or its any appropriate combination.For hardware
It realizes, processor can be realized in one or more the following units: specific integrated circuit (ASIC), digital signal processor
(DSP), digital signal processing appts (DSPD), programmable logic device (PLD), field programmable gate array (FPGA), processing
Device, controller, microcontroller, microprocessor, other electronic units designed for realizing functions described herein or combinations thereof.
For software implementations, some or all of embodiment process can instruct relevant hardware to complete by computer program.
When realization, above procedure can be stored in computer-readable medium or as the one or more on computer-readable medium
Instruction or code are transmitted.Computer-readable medium includes computer storage media and communication media, wherein communication media packet
It includes convenient for from a place to any medium of another place transmission computer program.Storage medium can be computer can
Any usable medium of access.Computer-readable medium can include but is not limited to RAM, ROM, EEPROM, CD-ROM or other
Optical disc storage, magnetic disk storage medium or other magnetic storage apparatus or can be used in carry or store have instruction or data
The desired program code of structure type simultaneously can be by any other medium of computer access.
Finally it should be noted that the above embodiments are merely illustrative of the technical solutions of the present invention, rather than the present invention is protected
The limitation of range is protected, although explaining in detail referring to preferred embodiment to the present invention, those skilled in the art are answered
Work as understanding, it can be with modification or equivalent replacement of the technical solution of the present invention are made, without departing from the reality of technical solution of the present invention
Matter and range.
Claims (8)
1. a kind of market monitoring device, which is characterized in that including image capture device, information transmission equipment and monitor supervision platform；
Described image acquires equipment and information transmission equipment communication connection；
The information transmission equipment and the monitor supervision platform communicate to connect；
Described image acquisition equipment real-time image acquisition simultaneously sends it to the information transmission equipment；
The image received is sent to the monitor supervision platform by the information transmission equipment；
The monitor supervision platform is for detecting the act of violence in image.
2. market monitoring device according to claim 1, which is characterized in that the monitor supervision platform includes the first processing mould
Block, Second processing module, third processing module, fourth processing module, the 5th processing module, the 6th processing module, described first
For processing module for inputting image detected, the Second processing module is used to extract the global characteristics of image, and described the
Three processing modules are used for by extracted image overall Fusion Features in depth network model, and the fourth processing module is based on
Third processing module determines violence testing result, and the 5th processing module is for optimizing fourth processing module violence detection knot
Fruit, the 6th processing module are used to export the violence testing result of optimization.
3. market monitoring device according to claim 2, which is characterized in that the Second processing module includes the first processing
Layer, second processing layer, third process layer, fourth process layer；First process layer pre-processes the image of input；It is described
Second processing layer is filtered to image and convolution operation；The third process layer does the output result of second processing layer non-
Linear Mapping；The fourth process layer is for the image after compressive non-linearity mapping.
4. market monitoring device according to claim 3, which is characterized in that in second processing layer, pass through convolution operation
To pretreated image zooming-out local neighborhood feature, by Multilevel Iteration, the overall situation for extracting image by two-dimensional convolution is special
Sign:
In formula,Indicate the activation value in i-th layer of j-th of Feature Mapping at the position (x, y), this activation value is exactly to scheme
The two-dimentional global characteristics of picture；YW () indicate activation primitive, wherein H, W respectively indicate the height of two-dimensional convolution core, width it is big
It is small；Indicate the weight of convolution kernel,Indicate activation value of (i-1)-th layer of d-th of the Feature Mapping at (x, y),
EMijIndicate that bias vector, i indicate that the convolutional layer that image is currently located, j indicate the Feature Mapping quantity of this layer.
5. market monitoring device according to claim 4, which is characterized in that the third processing module is by second processing mould
Two-dimensional convolution core in block generates three dimensional convolution kernel by spatial spread, and the Three dimensional convolution at pixel (x, y, z) calculates fixed
Justice are as follows:
In formula,Indicate the activation value in i-th layer of j-th of Feature Mapping at the position (x, y, z)；This activation value is just
It is the three-dimensional global characteristics of image；YW () indicates activation primitive, wherein H, W, T respectively indicate the height of three dimensional convolution kernel, width
Size on degree and time dimension；Indicate the weight of convolution kernel,Indicate (i-1)-th layer of d-th of feature
The activation value being mapped at (x, y, z), EMijIndicate that bias vector, i indicate that the convolutional layer that image is currently located, j indicate this layer
Feature Mapping quantity.
6. market monitoring device according to claim 5, which is characterized in that the fourth processing module is based on third processing
Module uses tri- second processing layers of C1, C2, C3, and the three dimensional convolution kernel size that C1, C2 and C3 are used is respectively 7 × 7 × 5,5 ×
5 × 5 and 3 × 3 × 3 pixels.
7. market monitoring device according to claim 6, which is characterized in that the input of the fourth processing module is by 40
The image segments X that frame consecutive image is constituted；Picture frame is normalized to 60 × 90 pixel sizes and is converted to after pretreatment
Grayscale image；Scalar Y scalar is exported, for indicating testing result that model input image, for trained model, if survey
Attempt comprising Violent scene as in, then output Y is 1, otherwise exporting result is 0.
8. market monitoring device according to claim 7, which is characterized in that the fourth processing module is to the first two second
The characteristic pattern that process layer is calculated carries out pondization operation, and pond is calculate by the following formula:
In formula,Indicate y-th of characteristic pattern of x layer,Indicate that y-th of characteristic pattern of x-1 layer, θ and B are respectively that multiplying property is inclined
It sets and is biased with additivity,Indicate y-th of the multiplying property biasing of x layer,Indicate y-th of the additivity biasing of x layer, δTFor sampling function,Wherein, t is the time, and T is the sampling period, and n ∈ [0 ,+∞] and n are positive integer.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810961756.0A CN109145823A (en) | 2018-08-22 | 2018-08-22 | A kind of market monitoring device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810961756.0A CN109145823A (en) | 2018-08-22 | 2018-08-22 | A kind of market monitoring device |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN109145823A true CN109145823A (en) | 2019-01-04 |
Family
ID=64790916
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810961756.0A Withdrawn CN109145823A (en) | 2018-08-22 | 2018-08-22 | A kind of market monitoring device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109145823A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110502995A (en) * | 2019-07-19 | 2019-11-26 | 南昌大学 | Driver based on subtle facial action recognition yawns detection method |
-
2018
- 2018-08-22 CN CN201810961756.0A patent/CN109145823A/en not_active Withdrawn
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110502995A (en) * | 2019-07-19 | 2019-11-26 | 南昌大学 | Driver based on subtle facial action recognition yawns detection method |
| CN110502995B (en) * | 2019-07-19 | 2023-03-14 | 南昌大学 | Driver yawning detection method based on fine facial action recognition |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN104202547B (en) | Method, projection interactive approach and its system of target object are extracted in projected picture | |
| CN105405150B (en) | Anomaly detection method and device based on fusion feature | |
| CN107016357A (en) | A kind of video pedestrian detection method based on time-domain convolutional neural networks | |
| CN109522790A (en) | Human body attribute recognition approach, device, storage medium and electronic equipment | |
| CN111291633A (en) | Real-time pedestrian re-identification method and device | |
| CN106778645A (en) | A kind of image processing method and device | |
| CN113591968A (en) | Infrared weak and small target detection method based on asymmetric attention feature fusion | |
| CN108596041A (en) | A kind of human face in-vivo detection method based on video | |
| CN111325051A (en) | Face recognition method and device based on face image ROI selection | |
| CN107169994A (en) | Correlation filtering tracking based on multi-feature fusion | |
| CN108680796A (en) | Electromagnetic information leakage detecting system and method for computer display | |
| CN109902613A (en) | A kind of human body feature extraction method based on transfer learning and image enhancement | |
| CN106297755A (en) | A kind of electronic equipment for musical score image identification and recognition methods | |
| CN109145822A (en) | A kind of violence detection system of deep learning | |
| CN108171689A (en) | A kind of identification method, device and the storage medium of the reproduction of indicator screen image | |
| CN110111347A (en) | Logos extracting method, device and storage medium | |
| CN108471497A (en) | A kind of ship target real-time detection method based on monopod video camera | |
| CN109784230A (en) | A kind of facial video image quality optimization method, system and equipment | |
| CN109299660A (en) | A kind of Intelligent campus early warning platform | |
| CN114529462A (en) | Millimeter wave image target detection method and system based on improved YOLO V3-Tiny | |
| CN112801037A (en) | Face tampering detection method based on continuous inter-frame difference | |
| CN111950457A (en) | Oil field safety production image identification method and system | |
| CN111860457A (en) | Fighting behavior recognition early warning method and recognition early warning system thereof | |
| CN110163095B (en) | Loop detection method, loop detection device and terminal equipment | |
| CN108229281A (en) | The generation method and method for detecting human face of neural network, device and electronic equipment |
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 | ||
| WW01 | Invention patent application withdrawn after publication | ||
| WW01 | Invention patent application withdrawn after publication |
Application publication date: 20190104 |