CN109068145A - Distribution type intelligent video analysis system, method, apparatus, equipment and storage medium - Google Patents
Distribution type intelligent video analysis system, method, apparatus, equipment and storage medium Download PDFInfo
- Publication number
- CN109068145A CN109068145A CN201810930648.7A CN201810930648A CN109068145A CN 109068145 A CN109068145 A CN 109068145A CN 201810930648 A CN201810930648 A CN 201810930648A CN 109068145 A CN109068145 A CN 109068145A
- Authority
- CN
- China
- Prior art keywords
- video
- intelligent video
- video analysis
- end processor
- identification information
- 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
- 238000004458 analytical method Methods 0.000 title claims abstract description 120
- 238000000034 method Methods 0.000 title abstract description 24
- 238000013135 deep learning Methods 0.000 claims abstract description 36
- 238000012545 processing Methods 0.000 claims abstract description 36
- 238000004422 calculation algorithm Methods 0.000 claims abstract description 34
- 238000003062 neural network model Methods 0.000 claims description 10
- 238000004590 computer program Methods 0.000 claims description 9
- 210000004218 nerve net Anatomy 0.000 claims description 2
- 230000008569 process Effects 0.000 description 9
- 230000008901 benefit Effects 0.000 description 5
- 238000010586 diagram Methods 0.000 description 4
- 238000013473 artificial intelligence Methods 0.000 description 3
- 238000004364 calculation method Methods 0.000 description 3
- 230000000694 effects Effects 0.000 description 3
- 230000008921 facial expression Effects 0.000 description 3
- 230000006870 function Effects 0.000 description 3
- 238000010801 machine learning Methods 0.000 description 3
- 238000005516 engineering process Methods 0.000 description 2
- 238000010191 image analysis Methods 0.000 description 2
- 241000287828 Gallus gallus Species 0.000 description 1
- 230000009471 action Effects 0.000 description 1
- 230000006978 adaptation Effects 0.000 description 1
- 238000004891 communication Methods 0.000 description 1
- 230000007812 deficiency Effects 0.000 description 1
- 239000006185 dispersion Substances 0.000 description 1
- 235000013399 edible fruits Nutrition 0.000 description 1
- 230000005611 electricity Effects 0.000 description 1
- 230000001815 facial effect Effects 0.000 description 1
- 238000001914 filtration Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000012544 monitoring process Methods 0.000 description 1
- 230000003068 static effect Effects 0.000 description 1
- 230000001360 synchronised effect Effects 0.000 description 1
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N21/00—Selective content distribution, e.g. interactive television or video on demand [VOD]
- H04N21/20—Servers specifically adapted for the distribution of content, e.g. VOD servers; Operations thereof
- H04N21/21—Server components or server architectures
- H04N21/218—Source of audio or video content, e.g. local disk arrays
- H04N21/21805—Source of audio or video content, e.g. local disk arrays enabling multiple viewpoints, e.g. using a plurality of cameras
-
- 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
-
- 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
-
- 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
-
- 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
- G06V40/174—Facial expression recognition
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N21/00—Selective content distribution, e.g. interactive television or video on demand [VOD]
- H04N21/20—Servers specifically adapted for the distribution of content, e.g. VOD servers; Operations thereof
- H04N21/23—Processing of content or additional data; Elementary server operations; Server middleware
- H04N21/231—Content storage operation, e.g. caching movies for short term storage, replicating data over plural servers, prioritizing data for deletion
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N21/00—Selective content distribution, e.g. interactive television or video on demand [VOD]
- H04N21/20—Servers specifically adapted for the distribution of content, e.g. VOD servers; Operations thereof
- H04N21/23—Processing of content or additional data; Elementary server operations; Server middleware
- H04N21/234—Processing of video elementary streams, e.g. splicing of video streams or manipulating encoded video stream scene graphs
- H04N21/23418—Processing of video elementary streams, e.g. splicing of video streams or manipulating encoded video stream scene graphs involving operations for analysing video streams, e.g. detecting features or characteristics
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Multimedia (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- General Health & Medical Sciences (AREA)
- Health & Medical Sciences (AREA)
- Signal Processing (AREA)
- Data Mining & Analysis (AREA)
- Oral & Maxillofacial Surgery (AREA)
- Evolutionary Computation (AREA)
- Human Computer Interaction (AREA)
- Artificial Intelligence (AREA)
- General Engineering & Computer Science (AREA)
- Life Sciences & Earth Sciences (AREA)
- Computational Linguistics (AREA)
- Software Systems (AREA)
- Mathematical Physics (AREA)
- Databases & Information Systems (AREA)
- Computing Systems (AREA)
- Molecular Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Biology (AREA)
- Biophysics (AREA)
- Biomedical Technology (AREA)
- Image Analysis (AREA)
- Image Processing (AREA)
Abstract
The present invention relates to intelligent video analysis fields, more particularly to a kind of distribution type intelligent video analysis system, method, apparatus, equipment and storage medium, the system includes: intelligent video analysis platform and multiple local video processing units, local video processing unit includes front end processor and IP video source, IP video source is for acquiring or storing the IP network video flowing for needing to carry out intelligent video analysis, front end processor is used to obtain the IP network video flowing of IP video source output, IP network video flowing is tentatively identified, and preliminary identification information is sent to intelligent video analysis platform；Intelligent video analysis platform handles preliminary identification information based on deep learning algorithm, obtains recognition result, and recognition result is back to front end processor for receiving preliminary identification information.Using the inexpensive front end processor of distributed deployment, while accurately identifying to video, cost is greatly saved, has given full play to the performance of intelligent video analysis, is effectively saved the performance resource of video processing.
Description
Technical field
The present invention relates to intelligent video analysis field more particularly to a kind of distribution type intelligent video analysis systems, method, dress
It sets, equipment and storage medium.
Background technique
With the mature of IP based network (Internet Protocol, agreement) in network video technique, currently
Emerge the sea largely using digital video monitoring as the IP Video Applications scene of representative, generated in these IP Video Applications scenes
Video image is measured, only relies on and manually carries out viewing identification, a large amount of manpowers and time will be expended, and be easy to produce mistakes and omissions problem.
Although intelligent IP camera has also been introduced at present to replace manually carrying out video identification, because of intelligent IP camera
SoC(System on Chip, system on chip) chip processing capabilities limitation, the reality of intelligent video pre-filtering is much not achieved
Border demand, thus people further introduce artificial intelligence depth learning technology carry out intelligent video analysis (such as face know
Not, object identification etc.).Wherein, most common intelligent video analysis system is the GPU(Graphics using high calculated performance
Processing Unit, graphics processor) server analyzes IP video source, the video analysis efficiency of the system and knowledge
Other precision is all very high, but for now, the system is configured before each IP video source, it is not only with high costs, and
The performance resource of the system can largely be wasted.
Summary of the invention
It is an object of the invention to overcome the deficiencies of existing technologies, a kind of reality of distribution type intelligent video analysis system is provided
Existing method, solves above-mentioned technical problem.
The technical solution of present invention realization above-mentioned purpose are as follows:
A kind of distribution type intelligent video analysis system, comprising: intelligent video analysis platform and multiple local video processing units；
The local video processing unit includes front end processor and IP video source；
The IP video source is for acquiring or storing the IP network video flowing for needing to carry out intelligent video analysis；
The front end processor is used to obtain the IP network video flowing of the IP video source output, flows into the IP network video
The preliminary identification of row, and preliminary identification information is sent to the intelligent video analysis platform；
The intelligent video analysis platform is for receiving the preliminary identification information, based on deep learning algorithm to the preliminary knowledge
Other information is handled, and obtains recognition result, and the recognition result is back to the front end processor；
The intelligent video analysis platform is used to updated pretreatment neural network model being transmitted to the front end processor.
The present invention also provides a kind of distribution type intelligent video analysis methods, are applied to intelligent video analysis platform, feature
It is, comprising:
Receive the preliminary identification information corresponding with IP network video flowing that front end processor uploads；
The preliminary identification information is handled based on deep learning algorithm, obtains recognition result；
The recognition result is back to the front end processor.
The present invention also provides a kind of distribution type intelligent video analytical equipments, are applied to intelligent video analysis platform, feature
It is, comprising:
Receiving unit, for receiving the preliminary identification information corresponding with IP network video flowing of front end processor upload；
Processing unit obtains recognition result for handling based on deep learning algorithm the preliminary identification information；
Feedback unit is also used to for the recognition result to be back to the front end processor by updated pretreatment nerve net
Network model is back to the front end processor.
The present invention also provides a kind of computer equipment, including memory and processor, calculating is stored in the memory
Machine program, when the computer program is executed by the processor, so that the processor executes a kind of above-mentioned distributed intelligence
The step of video analysis method；The present invention also provides a kind of computer readable storage mediums, which is characterized in that the computer can
It reads to be stored with computer program on storage medium, when the computer program is executed by processor, so that the processor executes
Described in claim 7 the step of a kind of distribution type intelligent video analysis method.
It is that the present invention obtains the utility model has the advantages that using can distributed deployment inexpensive front end processor computing capability, first to view
Frequency carries out simple preliminary analysis, and then intelligent video analysis platform carries out advanced treating to Preliminary Analysis Results, is used only one
Platform intelligent video analysis platform with high performance, so that it may realize the face in the video information exported to multiple IP video sources
It is accurately identified, each IP video source compared to the prior art needs the technical side of one intelligent video analysis platform of setting
While capable of obtaining same video recognition effect, cost is greatly saved in case, and intelligent video point is more fully utilized
Analyse the performance resource of platform.
Detailed description of the invention
Fig. 1 shows distribution type intelligent video analysis system implementation environment figure provided in an embodiment of the present invention；
Fig. 2 shows a kind of structural schematic diagrams of distribution type intelligent video analysis system provided in an embodiment of the present invention；
Fig. 3 shows a kind of distribution type intelligent video analysis method flow chart provided by the invention；
Fig. 4 shows a kind of structural schematic diagram of distribution type intelligent video analytical equipment provided in an embodiment of the present invention.
Specific embodiment
In order to make the objectives, technical solutions, and advantages of the present invention clearer, with reference to the accompanying drawings and embodiments, right
The present invention is further elaborated.It should be appreciated that the specific embodiments described herein are merely illustrative of the present invention, and
It is not used in the restriction present invention.
The embodiment of the present invention is concentrated by intelligent video analysis platform to the local video processing unit output to dispersion
Preliminary Analysis Results carry out advanced treating, can not only obtain higher video identification effect, and cost is greatly saved, and fill
The performance resource of intelligent video analysis platform is utilized with dividing.
Embodiment one
Fig. 1 is shown suitable for distribution type intelligent video analysis system implementation environment figure provided in an embodiment of the present invention, for convenient for saying
It is bright, illustrate only part related with the embodiment of the present invention.
Intelligent video analysis platform is at least connect with 2 front end processors by IP network, 1 front end processor and one or more IP
Video source communication connection；Intelligent video analysis platform is stored with the unpaired message data acquisition system of specific occasion needs, including but not
It is limited to recognition of face information data set, Car license recognition information data set；Intelligent video analysis platform is by IP network with before
The machine of setting is communicated, including internet or local area network；Front end processor transmits signal by data line with IP video source.
Embodiment two
Fig. 2 shows a kind of structural schematic diagram of distribution type intelligent video analysis system provided in an embodiment of the present invention, the systems
Include: intelligent video analysis platform and multiple local video processing units, wherein local video processing unit include front end processor and
IP video source；
IP video source is for acquiring or storing the IP network video flowing for needing to carry out intelligent video analysis；
Front end processor is used to obtain the IP network video flowing of IP video source output, is regarded using pretreatment neural network model to IP network
Frequency stream is tentatively identified, and preliminary identification information is sent to the intelligent video analysis platform；
Intelligent video analysis platform for receiving preliminary identification information, based on deep learning algorithm to preliminary identification information at
Reason, obtains recognition result, and recognition result is back to front end processor；
Intelligent video analysis platform is also used to updated pretreatment neural network model being transmitted to the front end processor.
In embodiments of the present invention, local video processing unit is to refer to simply locate video at video acquisition end
The device of reason, including IP video source and front end processor, front end processor are referred to using pretreatment neural network model to video image
The common industrial personal computer (such as the Intel industrial personal computer that can use tetra- core CPU of dominant frequency 3.0GHz) for carrying out preliminary analysis, can incite somebody to action
The characteristic information for meeting preset standard in IP video flowing intercepts, and is sent to intellectualized analysis platform；Wherein meet pre- bidding
Quasi- characteristic information refers to that a kind of standard being arranged inside program, equipment in advance, agreement (such as are existed by the standard, agreement
Character graphics etc. are obtained in video image, meeting the standard, the character graphics of the judgment criteria of agreement can believe as feature
Breath is intercepted), the standard, agreement are pretreatment neural network model, and machine goes scanning, acquisition to meet according to the standard agreement
Characteristic information, the information for meeting the standard can be opened with other graphics distinctions in video, for example, in the scene of recognition of face
In, the characteristic information for meeting preset standard is exactly facial contour, in Car license recognition scene, meets the characteristic information of preset standard just
It is license plate outline etc., and other figures, background in image will be filtered；IP video source is to refer to acquire or store IP network
The equipment of network video, and send IP network video to front end processor, including but not limited to IP camera, digital hard disc video recorder etc.
Equipment.
Local video processing unit application scenarios are relatively broad, the screens prison such as example small shop of many personal use scenes, family
It can all be used in control, but the video process performance of these video process apparatus is not high, recognition efficiency and accuracy are all very low, right
For ordinary user, it is based on cost consideration, generally will not specially configure the very high image identification system of process performance, this is just
Cause these local video processing units to seem very chicken ribs, performs practically no function.
And in the present invention, preliminary analysis is first carried out to video by front end processor, preliminary identification refers to relative to intelligent video
Through deep learning algorithm for video carries out advanced treating, front end processor only needs to run simple video analysis analysis platform
Software carries out the interception of characteristic information to video, i.e., when the characteristic information for occurring meeting preset standard in video pictures, to institute
State characteristic information to be intercepted, form the preliminary identification information, for example, in recognition of face scene, front end processor only need by
The picture capturing for occurring face in video pictures, which comes out, forms image, and is uploaded to intelligent video analysis as preliminary identification information
Platform.
Then the preliminary analysis that simple process obtains is passed through by the front end processor that intelligent video analysis platform will be dispersed in various regions
As a result it is focused on, obtains more accurate video recognition result, video recognition result is then back to front end processor.
Intelligent video analysis platform can run deep learning algorithm and analyze image, and deep learning is machine learning
Middle a kind of based on the method for carrying out representative learning to data, observation (such as piece image) can be used various ways and carry out table
Show, such as the vector of each pixel intensity value, or be more abstractively expressed as a series of sides, the region of specific shape etc., use this
A little specific representation methods are easier to be identified (for example, recognition of face or human facial expression recognition) to specific image recognition；
Intelligent video analysis platform in the present invention is using the clothes that can be handled by deep learning algorithm video image information
Business device has very high recognition efficiency and accuracy of identification with stronger video processing capabilities, which generally at least configures
There are 12 core CPU and also at least configures two panels GPU.
When intelligent video analysis platform is handled the preliminary identification information by the deep learning algorithm, identification
Minutia in the preliminary identification information searches for the number to match with the minutia from preset information aggregate
According to, recognition result is formed, and recognition result is back to front end processor, for example, in above-mentioned recognition of face scene, intelligent video point
After analysing image of the platform reception containing face information, preliminary analysis information is handled by deep learning algorithm, i.e., to this
Frame shows that highlighting for face information carries out depth analysis identification, the including but not limited to five features of face, skeleton character, then
It will identify that the face information set prestored in obtained information and date library compares, whether find out has the face to match letter
Breath if so, then show the face information, while showing that the face information corresponds to the identity information of people, and by the identity information
It is back to front end processor, face is accurately identified to realize.
Updated pretreatment neural network model can also be transmitted to front end processor by intelligent video analysis platform, in order to
Front end processor can preferably adapt to the preliminary analysis of image.
As soon as a profession Intel server with high performance is used only as intelligent video analysis platform in the present embodiment,
The face that may be implemented in the video information acquired in real time to multiple IP cameras accurately identifies, and can reach at this stage
Hardware calculation of performance indicators under conditions of, single server can front end processor with no less than 12 in system of the present invention
It is built into distribution type intelligent video analysis system, it is no longer necessary to high performance server be set to video at each video acquisition end
Depth recognition is carried out, is not only only capable of obtaining preferable video identification effect, has also greatly saved cost, and an intelligent video
Analysis platform just can satisfy the video analysis demand of at least 12 IP video sources, give full play to intelligent video analysis platform
Performance resource.
The intellectual analysis of HD video of the processing not less than 96 road 1080P simultaneously, greatly improves the intelligence of whole system
While energy video analysis processing capacity and power system capacity, the cost that face accurately identifies is greatly reduced；Take full advantage of height
The process performance of performance professional Intel server, does not reproduce into the unnecessary performance wasting of resources.
Embodiment three
Fig. 3 shows a kind of distribution type intelligent video analysis method flow chart provided by the invention, is applied to intelligent video analysis
Platform, the specific steps of which are as follows:
In step S301, the preliminary identification information corresponding with IP network video flowing that front end processor uploads is received.
Intelligent video analysis platform for receiving preliminary identification information, based on deep learning algorithm to preliminary identification information into
Row processing, obtains recognition result, and recognition result is back to front end processor.
Intelligent video analysis platform can run deep learning algorithm and analyze image, and deep learning is machine learning
Middle a kind of based on the method for carrying out representative learning to data, observation (such as piece image) can be used various ways and carry out table
Show, such as the vector of each pixel intensity value, or be more abstractively expressed as a series of sides, the region of specific shape etc., use this
A little specific representation methods are easier to be learnt (for example, recognition of face or human facial expression recognition) to specific image recognition；
Carrying out processing to preliminary identification information based on deep learning algorithm is referred to through deep learning algorithm to video image information
The server handled has very high recognition efficiency and accuracy of identification, generally all can with stronger video processing capabilities
Using at least be configured with 12 core CPU and also at least configure two panels GPU computing capability professional Intel server.
Front end processor is to refer to carry out video image the common industrial personal computer of preliminary analysis (such as to use dominant frequency
Tetra- core CPU(Central Processing Unit of 3.0GHz, central processing unit) Intel industrial personal computer), IP can be regarded
The characteristic information for meeting preset standard in frequency stream intercepts, and is sent to intellectualized analysis platform；Wherein meet preset standard
Characteristic information refers to a kind of standard being arranged inside program, equipment in advance, agreement (such as by the standard, agreement in video
Character graphics etc. are obtained in image, characteristic information quilt can be regarded by meeting the standard, the character graphics of the judgment criteria of agreement
Interception), machine goes scanning according to the standard agreement, obtains the characteristic information met, and the information for meeting the standard can be in video
In opened with other graphics distinctions, for example, the characteristic information for meeting preset standard is exactly face wheel in the scene of recognition of face
Exterior feature, in Car license recognition scene, the characteristic information for meeting preset standard is exactly license plate outline etc., and other figures, back in image
Scape will be filtered；IP video source is the equipment for referring to acquire or store IP network video, and IP network video is sent to
Front end processor, the including but not limited to equipment such as IP camera, digital hard disc video recorder.
Preliminary identification, which refers to, carries out advanced treating to video by deep learning algorithm relative to intelligent video analysis platform
For, front end processor only needs to run the interception that simple video analysis software carries out characteristic information to video, that is, works as video pictures
When the middle characteristic information for occurring meeting preset standard, the characteristic information is intercepted, forms the preliminary identification information, example
Such as, in recognition of face scene, front end processor only needs to come out the picture capturing for occurring face in video pictures, by face
The picture capturing divided forms a frame image.
In embodiments of the present invention, front end processor can be the Intel industrial personal computer or other of tetra- core CPU of dominant frequency 3.0GHz
Performance is similar or better industrial personal computer, IP network video flowing include but is not limited to IP camera acquire in real time video information,
The video information of digital hard disc video recorder storage.
It is illustrated so that the present invention is applied to recognition of face scene as an example below, IP camera believes the video acquired in real time
Breath is transmitted to industrial personal computer, and industrial personal computer carries out preliminary analysis to video, specifically: operation video analysis software judges video pictures
In whether there is face, if there is face, then the frame picture for face occur is intercepted from video flowing and is come out, intercepted simultaneously
The picture of face part out forms a frame picture as preliminary analysis information, is uploaded to server.
In step s 302, the preliminary identification information is handled based on deep learning algorithm, obtains recognition result.
In the present embodiment, intelligent video analysis platform using 12 core CPU and configures the special of two panels GPU computing capability
Industry Intel server can run complicated artificial intelligence deep learning algorithm.
In above-mentioned recognition of face scene, after server receives preliminary analysis information, by deep learning algorithm to first
Step analysis information is handled, i.e., shows that highlighting for face information carries out depth analysis identification, including but not limited to people to the frame
Then the five features of face, skeleton character carry out the face information set prestored in the obtained information and date library of identification pair
Than whether have the face information that matches, if so, then display is the face information, while showing the face information pair if finding out
The identity information of people is answered, face is accurately identified to realize.
In step S303, the recognition result is back to the front end processor.
In above-mentioned recognition of face scene, server accurately identifies face information, and matches corresponding identity letter
After breath, identity information is back to front end processor, directly displays face recognition result in front end.
A profession Intel server with high performance is used only in the present embodiment, so that it may realize and image to multiple IP
Face in the video information that acquires in real time of head accurately identified, at this stage attainable hardware calculation of performance indicators
Under the conditions of, single server can be built into distribution type intelligent video with no less than 12 front end processors in system of the present invention
Analysis system, while the intellectual analysis of the HD video not less than 96 road 1080P is handled, greatly improve the intelligence of whole system
While energy video analysis processing capacity and power system capacity, the cost that face accurately identifies is greatly reduced；By server institute
The function of the intelligent video-image analysis undertaken, is largely divided by being distributed in the front end processor near the IP video source of front end
Load carries out identifying processing by the video that multiple IP cameras acquire by a server, takes full advantage of high-performance profession
The process performance of Intel server does not reproduce into the unnecessary performance wasting of resources.
Example IV
Fig. 4 shows a kind of structural schematic diagram of distribution type intelligent video analytical equipment provided in an embodiment of the present invention, is applied to
Intelligent video analysis platform only shows part related to the present invention for convenience of description.
In embodiments of the present invention, distribution type intelligent video analytical equipment include receiving unit 410, processing unit 420 with
And return unit 430.
Receiving unit 410, for receiving the preliminary identification information corresponding with IP network video flowing of front end processor upload；
Intelligent video analysis platform for receiving preliminary identification information, based on deep learning algorithm to preliminary identification information at
Reason, obtains recognition result, and recognition result is back to front end processor.
Intelligent video analysis platform can run deep learning algorithm and analyze image, and deep learning is machine learning
Middle a kind of based on the method for carrying out representative learning to data, observation (such as piece image) can be used various ways and carry out table
Show, such as the vector of each pixel intensity value, or be more abstractively expressed as a series of sides, the region of specific shape etc., use this
A little specific representation methods are easier to be learnt (for example, recognition of face or human facial expression recognition) to specific image recognition；
Carrying out processing to preliminary identification information based on deep learning algorithm is referred to through deep learning algorithm to video image information
The server handled has very high recognition efficiency and accuracy of identification, generally all can with stronger video processing capabilities
Using at least be configured with 12 core CPU and also at least configure two panels GPU computing capability professional Intel server.
Front end processor is to refer to carry out video image the common industrial personal computer of preliminary analysis (such as to use dominant frequency
Tetra- core CPU(Central Processing Unit of 3.0GHz, central processing unit) Intel industrial personal computer), IP can be regarded
The characteristic information for meeting preset standard in frequency stream intercepts, and is sent to intellectualized analysis platform；Wherein meet preset standard
Characteristic information refers to a kind of standard being arranged inside program, equipment in advance, agreement (such as by the standard, agreement in video
Character graphics etc. are obtained in image, characteristic information quilt can be regarded by meeting the standard, the character graphics of the judgment criteria of agreement
Interception), machine goes scanning according to the standard agreement, obtains the characteristic information met, and the information for meeting the standard can be in video
In opened with other graphics distinctions, for example, the characteristic information for meeting preset standard is exactly face wheel in the scene of recognition of face
Exterior feature, in Car license recognition scene, the characteristic information for meeting preset standard is exactly license plate outline etc., and other figures, back in image
Scape will be filtered；IP video source is the equipment for referring to acquire or store IP network video, and IP network video is sent to
Front end processor, the including but not limited to equipment such as IP camera, digital hard disc video recorder.
Preliminary identification, which refers to, carries out advanced treating to video by deep learning algorithm relative to intelligent video analysis platform
For, front end processor only needs to run the interception that simple video analysis software carries out characteristic information to video, that is, works as video pictures
When the middle characteristic information for occurring meeting preset standard, the characteristic information is intercepted, forms the preliminary identification information, example
Such as, in recognition of face scene, front end processor only needs to come out the picture capturing for occurring face in video pictures, by face
The picture capturing divided forms a frame image.
In embodiments of the present invention, front end processor can be the Intel industrial personal computer or other of tetra- core CPU of dominant frequency 3.0GHz
Performance is similar or better industrial personal computer, IP network video flowing include but is not limited to IP camera acquire in real time video information,
The video information of digital hard disc video recorder storage.
It is illustrated so that the present invention is applied to recognition of face scene as an example below, IP camera believes the video acquired in real time
Breath is transmitted to industrial personal computer, and industrial personal computer carries out preliminary analysis to video, specifically: operation video analysis software judges video pictures
In whether there is face, if there is face, then the frame picture for face occur is intercepted from video flowing and is come out, intercepted simultaneously
The picture of face part out forms a frame picture as preliminary analysis information, is uploaded to server.
Processing unit 420 obtains identification knot for handling based on deep learning algorithm the preliminary identification information
Fruit；
In the present embodiment, intelligent video analysis platform uses 12 core CPU and configures the profession of two panels GPU computing capability
Intel server can run complicated artificial intelligence deep learning algorithm.
In above-mentioned recognition of face scene, after server receives preliminary analysis information, by deep learning algorithm to first
Step analysis information is handled, i.e., shows that highlighting for face information carries out depth analysis identification, including but not limited to people to the frame
Then the five features of face, skeleton character carry out the face information set prestored in the obtained information and date library of identification pair
Than whether have the face information that matches, if so, then display is the face information, while showing the face information pair if finding out
The identity information of people is answered, face is accurately identified to realize.
Feedback unit 430 is also used to for the recognition result to be back to the front end processor by updated pretreatment
Neural network model is transmitted to the front end processor.
In above-mentioned recognition of face scene, server accurately identifies face information, and matches corresponding identity letter
After breath, identity information is back to front end processor, directly displays face recognition result in front end；When pretreatment neural network model more
After new, feedback unit 430 is responsible for updated pretreatment neural network model being transmitted to front end processor, in order to which front end processor is more preferable
Adaptation image preliminary analysis.
A profession Intel server with high performance is used only in the present embodiment, so that it may realize and image to multiple IP
Face in the video information that acquires in real time of head accurately identified, at this stage attainable hardware calculation of performance indicators
Under the conditions of, single server can be built into distribution type intelligent video with no less than 12 front end processors in system of the present invention
Analysis system, while the intellectual analysis of the HD video not less than 96 road 1080P is handled, greatly improve the intelligence of whole system
While energy video analysis processing capacity and power system capacity, the cost that face accurately identifies is greatly reduced；By server institute
The function of the intelligent video-image analysis undertaken, is largely divided by being distributed in the front end processor near the IP video source of front end
Load carries out identifying processing by the video that multiple IP cameras acquire by a server, takes full advantage of high-performance profession
The process performance of Intel server does not reproduce into the unnecessary performance wasting of resources.
Embodiment five
The present invention also provides a kind of computer equipment, including memory and processor, computer journey is stored in the memory
Sequence, when the computer program is executed by the processor, so that the processor executes a kind of distributed intelligence in embodiment
The step of video analysis method:
Receive the preliminary identification information corresponding with IP network video flowing that front end processor uploads；
The preliminary identification information is handled based on deep learning algorithm, obtains recognition result；
The recognition result is back to the front end processor.
In one embodiment, a kind of computer readable storage medium is provided, is stored on computer readable storage medium
Computer program, when computer program is executed by processor, so that processor executes following steps:
Receive the preliminary identification information corresponding with IP network video flowing that front end processor uploads；
The preliminary identification information is handled based on deep learning algorithm, obtains recognition result；
The recognition result is back to the front end processor.
Although should be understood that various embodiments of the present invention flow chart in each step according to arrow instruction successively
It has been shown that, but these steps are not that the inevitable sequence according to arrow instruction successively executes.Unless expressly state otherwise herein,
There is no stringent sequences to limit for the execution of these steps, these steps can execute in other order.Moreover, each embodiment
In at least part step may include that perhaps these sub-steps of multiple stages or stage are not necessarily multiple sub-steps
Completion is executed in synchronization, but can be executed at different times, the execution in these sub-steps or stage sequence is not yet
Necessarily successively carry out, but can be at least part of the sub-step or stage of other steps or other steps in turn
Or it alternately executes.
Those of ordinary skill in the art will appreciate that realizing all or part of the process in above-described embodiment method, being can
It is completed with instructing relevant hardware by computer program, it is readable that the program can be stored in a non-volatile computer
It takes in storage medium, the program is when being executed, it may include such as the process of the embodiment of above-mentioned each method.Wherein, the application is mentioned
Any reference used in each embodiment supplied to memory, storage, database or other media, may each comprise non-volatile
Property and/or volatile memory.Nonvolatile memory may include that read-only memory (ROM), programming ROM (PROM), electricity can
Programming ROM (EPROM), electrically erasable ROM(EEPROM) or flash memory.Volatile memory may include random access memory
Device (RAM) or external cache.By way of illustration and not limitation, RAM is available in many forms, such as static state RAM
(SRAM), dynamic ram (DRAM), synchronous dram (SDRAM), double data rate sdram (DDRSDRAM), enhanced SDRAM
(ESDRAM), synchronization link (Synchlink) DRAM(SLDRAM), memory bus (Rambus) direct RAM(RDRAM), it is straight
Connect memory bus dynamic ram (DRDRAM) and memory bus dynamic ram (RDRAM) etc..
Each technical characteristic of embodiment described above can be combined arbitrarily, for simplicity of description, not to above-mentioned reality
It applies all possible combination of each technical characteristic in example to be all described, as long as however, the combination of these technical characteristics is not deposited
In contradiction, all should be considered as described in this specification.
The embodiments described above only express several embodiments of the present invention, and the description thereof is more specific and detailed, but simultaneously
Limitations on the scope of the patent of the present invention therefore cannot be interpreted as.It should be pointed out that for those of ordinary skill in the art
For, without departing from the inventive concept of the premise, various modifications and improvements can be made, these belong to guarantor of the invention
Protect range.Therefore, the scope of protection of the patent of the invention shall be subject to the appended claims.
Claims (10)
1. a kind of distribution type intelligent video analysis system characterized by comprising intelligent video analysis platform and multiple local views
Frequency processing device；
The local video processing unit includes front end processor and IP video source；
The IP video source is for acquiring or storing the IP network video flowing for needing to carry out intelligent video analysis；
The front end processor is used to obtain the IP network video flowing of the IP video source output, flows into the IP network video
The preliminary identification of row, and preliminary identification information is sent to the intelligent video analysis platform；
The intelligent video analysis platform is for receiving the preliminary identification information, based on deep learning algorithm to the preliminary knowledge
Other information is handled, and obtains recognition result, and the recognition result is back to the front end processor；
The intelligent video analysis platform is used to updated pretreatment neural network model being transmitted to the front end processor.
2. a kind of distribution type intelligent video analysis system according to claim 1, which is characterized in that the IP video source
Are as follows: IP video capture device or storage equipment.
3. a kind of distribution type intelligent video analysis system according to claim 1, which is characterized in that the front end processor is to institute
When stating IP network video flowing and tentatively being identified, when the characteristic information for occurring meeting preset standard in video pictures, to described
Characteristic information is intercepted, and the preliminary identification information is formed.
4. a kind of distribution type intelligent video analysis system according to claim 1, which is characterized in that the intelligent video point
When analysis platform is handled the preliminary identification information by the deep learning algorithm, identify in the preliminary identification information
Minutia, the data to match with the minutia are searched for from preset information aggregate, form recognition result.
5. a kind of distribution type intelligent video analysis method is applied to intelligent video analysis platform characterized by comprising
Receive the preliminary identification information corresponding with IP network video flowing that front end processor uploads；
The preliminary identification information is handled based on deep learning algorithm, obtains recognition result；
The recognition result is back to the front end processor.
6. a kind of distribution type intelligent video analysis method according to claim 5, which is characterized in that described to be regarded with IP network
Frequency is when to flow corresponding preliminary identification information be that front end processor tentatively identifies the IP network video flowing, in video pictures out
The information that the existing characteristic information for meeting preset standard is intercepted.
7. a kind of distribution type intelligent video analysis method according to claim 5, which is characterized in that described to be based on depth
Habit algorithm carries out processing to the preliminary identification information and specifically includes:
The minutia in the preliminary identification information is identified by the deep learning algorithm；
The data to match with the minutia are searched for from preset information aggregate, form recognition result.
8. a kind of distribution type intelligent video analytical equipment is applied to intelligent video analysis platform characterized by comprising
Receiving unit, for receiving the preliminary identification information corresponding with IP network video flowing of front end processor upload；
Processing unit obtains recognition result for handling based on deep learning algorithm the preliminary identification information；
Feedback unit is also used to for the recognition result to be back to the front end processor by updated pretreatment nerve net
Network model is back to the front end processor.
9. a kind of computer equipment, which is characterized in that including memory and processor, computer journey is stored in the memory
Sequence, when the computer program is executed by the processor, so that the processor perform claim requires any one of 5-7 described one
The step of kind distribution type intelligent video analysis method.
10. a kind of computer readable storage medium, which is characterized in that be stored with computer on the computer readable storage medium
Program, when the computer program is executed by processor, so that the processor perform claim requires the described one kind of any one of 5-7
The step of distribution type intelligent video analysis method.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810930648.7A CN109068145A (en) | 2018-08-15 | 2018-08-15 | Distribution type intelligent video analysis system, method, apparatus, equipment and storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810930648.7A CN109068145A (en) | 2018-08-15 | 2018-08-15 | Distribution type intelligent video analysis system, method, apparatus, equipment and storage medium |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN109068145A true CN109068145A (en) | 2018-12-21 |
Family
ID=64686036
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810930648.7A Pending CN109068145A (en) | 2018-08-15 | 2018-08-15 | Distribution type intelligent video analysis system, method, apparatus, equipment and storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109068145A (en) |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110855947A (en) * | 2019-11-26 | 2020-02-28 | 浙江大华技术股份有限公司 | Image snapshot processing method and device |
| CN111080637A (en) * | 2019-12-25 | 2020-04-28 | 深圳力维智联技术有限公司 | Cloud service-based advertisement remote method, device, system, product and medium |
| CN111708635A (en) * | 2020-06-16 | 2020-09-25 | 深圳天海宸光科技有限公司 | Video intelligent grading processing system and method |
| CN111836102A (en) * | 2019-04-23 | 2020-10-27 | 杭州海康威视数字技术股份有限公司 | Video frame analysis method and device |
| CN112215567A (en) * | 2020-09-28 | 2021-01-12 | 上海鸢安智能科技有限公司 | Production flow compliance checking method and system, storage medium and terminal |
| CN113566729A (en) * | 2021-07-27 | 2021-10-29 | 同济大学 | Method for realizing overwater and underwater integrated distributed high-speed video measurement system |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104637035A (en) * | 2015-02-15 | 2015-05-20 | 百度在线网络技术（北京）有限公司 | Method, device and system for generating cartoon face picture |
| CN105574506A (en) * | 2015-12-16 | 2016-05-11 | 深圳市商汤科技有限公司 | Intelligent face tracking system and method based on depth learning and large-scale clustering |
| US20160328606A1 (en) * | 2013-11-15 | 2016-11-10 | Facebook, Inc. | Pose-aligned networks for deep attribute modeling |
| CN106204780A (en) * | 2016-07-04 | 2016-12-07 | 武汉理工大学 | A kind of based on degree of depth study and the human face identification work-attendance checking system and method for cloud service |
| CN106982359A (en) * | 2017-04-26 | 2017-07-25 | 深圳先进技术研究院 | A kind of binocular video monitoring method, system and computer-readable recording medium |
| CN108009491A (en) * | 2017-11-29 | 2018-05-08 | 深圳火眼智能有限公司 | A kind of object recognition methods solved in fast background movement and system |
-
2018
- 2018-08-15 CN CN201810930648.7A patent/CN109068145A/en active Pending
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20160328606A1 (en) * | 2013-11-15 | 2016-11-10 | Facebook, Inc. | Pose-aligned networks for deep attribute modeling |
| CN104637035A (en) * | 2015-02-15 | 2015-05-20 | 百度在线网络技术（北京）有限公司 | Method, device and system for generating cartoon face picture |
| CN105574506A (en) * | 2015-12-16 | 2016-05-11 | 深圳市商汤科技有限公司 | Intelligent face tracking system and method based on depth learning and large-scale clustering |
| CN106204780A (en) * | 2016-07-04 | 2016-12-07 | 武汉理工大学 | A kind of based on degree of depth study and the human face identification work-attendance checking system and method for cloud service |
| CN106982359A (en) * | 2017-04-26 | 2017-07-25 | 深圳先进技术研究院 | A kind of binocular video monitoring method, system and computer-readable recording medium |
| CN108009491A (en) * | 2017-11-29 | 2018-05-08 | 深圳火眼智能有限公司 | A kind of object recognition methods solved in fast background movement and system |
Cited By (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111836102A (en) * | 2019-04-23 | 2020-10-27 | 杭州海康威视数字技术股份有限公司 | Video frame analysis method and device |
| CN110855947A (en) * | 2019-11-26 | 2020-02-28 | 浙江大华技术股份有限公司 | Image snapshot processing method and device |
| CN110855947B (en) * | 2019-11-26 | 2021-07-06 | 浙江大华技术股份有限公司 | Image snapshot processing method and device |
| CN111080637A (en) * | 2019-12-25 | 2020-04-28 | 深圳力维智联技术有限公司 | Cloud service-based advertisement remote method, device, system, product and medium |
| CN111708635A (en) * | 2020-06-16 | 2020-09-25 | 深圳天海宸光科技有限公司 | Video intelligent grading processing system and method |
| CN112215567A (en) * | 2020-09-28 | 2021-01-12 | 上海鸢安智能科技有限公司 | Production flow compliance checking method and system, storage medium and terminal |
| CN113566729A (en) * | 2021-07-27 | 2021-10-29 | 同济大学 | Method for realizing overwater and underwater integrated distributed high-speed video measurement system |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109068145A (en) | Distribution type intelligent video analysis system, method, apparatus, equipment and storage medium | |
| Luo et al. | Real-world image datasets for federated learning | |
| US10089556B1 (en) | Self-attention deep neural network for action recognition in surveillance videos | |
| CN112232293B (en) | Image processing model training method, image processing method and related equipment | |
| CN110889672B (en) | Student card punching and class taking state detection system based on deep learning | |
| CN109145759B (en) | Vehicle attribute identification method, device, server and storage medium | |
| CN107463888B (en) | Face emotion analysis method and system based on multi-task learning and deep learning | |
| WO2021082589A1 (en) | Content check model training method and apparatus, video content check method and apparatus, computer device, and storage medium | |
| Wang et al. | A benchmark for clothes variation in person re‐identification | |
| WO2018102880A1 (en) | Systems and methods for replacing faces in videos | |
| WO2021184754A1 (en) | Video comparison method and apparatus, computer device and storage medium | |
| WO2022213540A1 (en) | Object detecting, attribute identifying and tracking method and system | |
| CN107527045A (en) | A kind of human body behavior event real-time analysis method towards multi-channel video | |
| Acharya et al. | Workload-aware task placement in edge-assisted human re-identification | |
| RU2770748C1 (en) | Method and apparatus for image processing, device and data carrier | |
| CN112597824A (en) | Behavior recognition method and device, electronic equipment and storage medium | |
| CN108921023A (en) | A kind of method and device of determining low quality portrait data | |
| CN111582654B (en) | Service quality evaluation method and device based on deep cycle neural network | |
| CN108288025A (en) | A kind of car video monitoring method, device and equipment | |
| CN112766176B (en) | Training method of lightweight convolutional neural network and face attribute recognition method | |
| Li et al. | Object Detection for UAV Images Based on Improved YOLOv6 | |
| CN107783856A (en) | A kind of method of testing and system of image processor parameter | |
| Siregar et al. | Human face recognition using eigenface in cloud computing environment | |
| CN109063732B (en) | Image ranking method and system based on feature interaction and multi-task learning | |
| CN106657026B (en) | Video flowing Intellisense coprocessing system and its processing method based on GPU |
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 | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20181221 |