CN104657205A - Virtualization-based video content analyzing method and system - Google Patents
Virtualization-based video content analyzing method and system Download PDFInfo
- Publication number
- CN104657205A CN104657205A CN201410830093.0A CN201410830093A CN104657205A CN 104657205 A CN104657205 A CN 104657205A CN 201410830093 A CN201410830093 A CN 201410830093A CN 104657205 A CN104657205 A CN 104657205A
- Authority
- CN
- China
- Prior art keywords
- task
- scheduling strategy
- analysis
- target video
- analysis task
- 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.)
- Granted
Links
Landscapes
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The invention discloses a virtualization-based video content analyzing method and system. The method and the system are characterized by comprising main parts, namely resource requirement analysis, task scheduling and equilibrium control allocation of analysis of video content. According to the method and the system, the content significance in the video is assessed; different content analyzing methods in the video are grouped into three types; a resource tank for analyzing the video content is managed and scheduled; seven scheduling modes are customized; multiple different scheduling policies are selected for different video content analyzing types, thereby ensuring the reliable analysis of different content in the video, implementing efficient utilization of treatment resources and reducing the waste and loss of the system resources.
Description
Technical field
The invention belongs to Video content analysis technique field, particularly relate to a kind of based on virtualized video content analysis method and system.
Background technology
Along with the continuous construction of wisdom safe city, require that intelligent monitor system is constantly perfect.In order to realize supervision of the cities region without dead angle, a large amount of monitoring cameras is installed to each corner in city.But a large amount of camera moment monitors each corner and will produce massive video data, use single physical main frame or the video data of single high-performance server to these magnanimity to analyze, its efficiency can not meet the demand of reality far away.
Summary of the invention
The object of the invention is to provide a kind of based on virtualized video content analysis method and system, for the efficiency requirements of the efficient video content analysis of satisfying magnanimity.
Based on the embodiment of the present invention, the invention provides a kind of based on virtualized video content analysis method, the method comprises:
Obtain the priority attribute information of target video, determine the task priority of target video being carried out to video content analysis according to the priority attribute information of target video;
Green energy conservation scheduling strategy is used for the analysis task not meeting priority requirement;
For the analysis task meeting priority requirement, obtain the categorical attribute information of target video, classify according to the analysis task of categorical attribute information to target video, determine the type of analysis task;
According to the corresponding relation of analysis task type and preset scheduling strategy, determine the scheduling strategy of the analysis task of target video;
Dispatch according to the analysis task of determined scheduling strategy to target video, distribute virtual resources, perform described analysis task.
Further, described categorical attribute information is for characterizing the algorithm characteristic of target video content analysis and target video content analysis to the demand of resource, and the type of described analysis task comprises: aspect ratio is to class, motion detection type, on-line study class.
Further, the corresponding relation of described preset scheduling strategy and scheduling strategy and analysis task type comprises:
Performance scheduling strategy, is applied to aspect ratio to class and on-line study alanysis task;
Efficiency scheduling strategy, is applied to aspect ratio to class, motion detection type and on-line study alanysis task;
Network topology scheduling strategy, is applied to on-line study alanysis task;
Algorithmic dispatching strategy, is applied to aspect ratio to class, motion detection type and on-line study alanysis task;
Environmental suitability scheduling strategy, is applied to aspect ratio to class, motion detection type and on-line study alanysis task;
Resource expansion scheduling strategy, is applied to motion detection type and on-line study alanysis task.
Further, from the scheduling strategy that each analysis task type is corresponding, select a kind of, multiple combination or all scheduling of resource carried out to one or more video analytic tasks.
Further, form cloud platform by multiple data center, scheduling is carried out to the analysis task of target video and refers to dispatch on cloud platform, share the virtual resources in multiple data center.
Based on the embodiment of the present invention, the present invention also provides a kind of based on virtualized video content analysis system, and this system comprises:
Interface arrangement, for target video data being sent to scheduler from external reception target video data, exports from the result data of scheduler receiving and analyzing task; The attribute information query interface of target video is provided;
Attribute acquisition device, for obtaining priority attribute information, the categorical attribute information of target video by interface arrangement;
Determining device, for determining the task priority of target video being carried out to video content analysis according to the priority attribute information of target video, and judge whether to use green energy conservation scheduling strategy to analyze target video according to task priority, the analysis task notice sorter meeting priority requirement is classified, uses green energy conservation scheduling strategy to carry out task scheduling for the analysis task notice scheduler not meeting priority requirement;
Sorter, for classifying to the analysis task meeting priority requirement according to categorical attribute information, determines the type of analysis task;
Scheduler, for the corresponding relation according to analysis task type and preset scheduling strategy, determines the scheduling strategy of the analysis task of target video; Dispatch according to the analysis task of determined scheduling strategy to target video, distribute virtual resources;
Virtual resources pond, for providing the virtual resources of video content analysis required by task, execution analysis task.
Further, algorithm characteristic and the demand of target video content analysis to resource of the target video content analysis that described sorter characterizes according to described categorical attribute information are classified to analysis task, and the type of described analysis task comprises: aspect ratio is to class, motion detection type, on-line study class.
Further, the sub-scheduler of green energy conservation, for dispatching the analysis task not meeting priority requirement;
The sub-scheduler of comparing class, for dispatching analysis task based on performance scheduling strategy, efficiency scheduling strategy, algorithmic dispatching strategy, environmental suitability scheduling strategy;
The sub-scheduler of motion detection type, for dispatching analysis task based on efficiency scheduling strategy, algorithmic dispatching strategy, environmental suitability scheduling strategy, resource expansion scheduling strategy;
The sub-scheduler of on-line study class, for dispatching analysis task based on performance scheduling strategy, efficiency scheduling strategy, algorithmic dispatching strategy, environmental suitability scheduling strategy, resource expansion scheduling strategy.
Further, a kind of, the multiple combination in corresponding scheduling strategy or whole selected by the sub-scheduler of described comparing class, the sub-scheduler of motion detection type and the sub-scheduler of on-line study class, carries out scheduling of resource to one or more video analytic tasks
Further, the shared virtual resources pond of described virtual resources pond for being made up of the virtual resources pond in multiple data center;
Described scheduler is dispatched based on the analysis task of cloud platform to target video be made up of multiple data center.
Video content analysis method and system based on Intel Virtualization Technology provided by the invention, comprise the resource requirement analysis of video content analysis, task scheduling and Balance route and distribute two major parts.The present invention makes assessment to content importance in video, different content analytical approach in video is divided into three classes, the resource pool that can be used for video content analysis is realized to management and dispatches and customized seven kinds of scheduling methods, multiple different scheduling strategy is chosen to different video content analysis types, thus ensure while different content fail-safe analysis in video, realize the efficiency utilization of process resource, reduce the waste loss of system resource.
Accompanying drawing explanation
The one that Fig. 1 provides for the embodiment of the present invention is based on virtualized video content analysis method step process flow diagram;
The analysis task type that Fig. 2 embodiment of the present invention provides and the corresponding relation schematic diagram of scheduling strategy;
The one that Fig. 3 embodiment of the present invention provides is based on virtualized video content analysis system structural representation.
Embodiment
Analyzing efficiently massive video content, is a very necessary technology.Usually the video of different importance and different content is carried out analyzing and have different resource use amounts and the requirement of analysis efficiency in actual applications, along with the lifting of video resolution and the complicated of video processnig algorithms make the requirement of Real-time video content analysis to computing power more and more higher, the video content analysis system that the existing physical server based on non-virtualized realizes can't bear the heavy load.
In order to realize the maximum using to existing resource, a kind of reasonably dynamic dispatching and Balance route technology are also selected to massive video content analysis with regard to particular importance to the classification technology of video content analysis.But rationally, dynamically scheduling and Balance route technology are analyzed massive video content also not have one at present.
For solving the problems of the prior art, the present invention proposes a kind of video content analysis method and system based on Intel Virtualization Technology, the present invention carries out virtual to the resource of video content analysis, build resource pool, virtual resources is managed and dispatches, seven kinds of resource dispatching strategies are provided, importance rate (or claiming priority) is carried out for video content divide, for the video content that importance rate is high, according to the feature of the algorithm that video content analysis uses, dynamic resource scheduling distribution and Balance route balance mode are divided into three classes, the resource dispatching strategy different to different classes of employing, if video content importance rate is not high, then directly adopt green resource scheduling method.
The one that Fig. 1 provides for the embodiment of the present invention, based on virtualized video content analysis method step process flow diagram, is specially:
The priority attribute information of step 100, acquisition target video, determines the task priority of target video being carried out to video content analysis according to the priority attribute information of target video;
In one embodiment of the invention, the priority attribute information of the video content of definition includes but not limited to: task grade, guarded region, storage and monitoring time segment etc.;
Such as, determine in the embodiment of the present invention that the method for target video being carried out to the priority of video content analysis is: for accident, task grade can be set to senior, and the property parameters such as monitoring place, storage and monitoring time segment that setting is relevant to event, video content analysis system provided by the invention can determine the priority to the analysis task that the video content of current input is analyzed automatically according to above-mentioned video content priority attribute information.Also key area and/or time period can be defined as the attribute information determining priority, from key area and/or time more close to the analysis priority of video content higher.
The priority threshold that step 102, basis are preset judges, carries out video content analysis whether use green energy conservation scheduling strategy to target video, if then perform step 108, otherwise performs step 104;
In the embodiment of the present invention, preset a priority threshold, such as priority is divided into 9 grades, priority threshold is set to 6 grades, when the priority of video content analysis task is more than 6 grades, do not use green energy conservation scheduling strategy, when the priority of video content analysis task is less than or equal to 6 grades, use green energy conservation scheduling strategy.Described green energy conservation scheduling strategy, see the definition in subsequent step, does not repeat herein.
In another specific embodiment of the present invention, the configuration interface of the priority attribute information of target video is provided, allow user to configure the priority attribute information of video content, priority number and the priority threshold of special configuration interface configuration video content analysis task also can be provided.Such as, suppose priorities attribute information only includes task grade one, video analytic tasks priority is divided into three grades, corresponding senior, intermediate, rudimentary three task grades respectively, user is senior by the priority attribute information of configuration interface Offered target video 1, and configuration preference level thresholding is secondary, namely task grade can use the mode determination scheduling strategy of step 104 ~ 106 higher than the video analytic tasks of middle rank, and the following task of middle rank all uses green energy conservation scheduling strategy.Further, the embodiment of the present invention is revised after user can be allowed target video to be analyzed to be carried out to the configuration of the priority attribute information of batch, and such as, video analytic tasks grade batch setting for a collection of target video meeting region and temporal characteristics is senior, middle rank or rudimentary.
The categorical attribute information of step 104, acquisition target video, classifies according to the analysis task of categorical attribute information to target video, determines the type of analysis task.
The embodiment of the present invention, according to the difference of analyzed video content, the algorithm analyzed video content and algorithm use the difference of stock number, classify to the analysis task of target video, the type of the analysis task of target video comprises following three classes:
(1) aspect ratio is to class: the video content analysis algorithm in this generic task needs a large amount of computational resources in the training stage, less resource is needed in the identification comparison stage, do not rely on the movable information of frame before and after video at the picture of training and testing process employing, therefore can split video;
In the present invention one specific embodiment, adopt histograms of oriented gradients (Histogram of OrientedGradient, HOG) feature extraction target sample picture HOG feature, then support vector machine (Support Vector Machine is used, SVM, a kind of trainable machine learning method) the HOG feature templates (or being called sorter) that training study obtains target is carried out to the HOG feature of a large amount of target sample pictures, the picture to be detected of HOG feature templates and input is finally used to carry out aspect ratio pair, thus judge whether comprise intended target in video content.The feature of this generic task needs a large amount of computational resources in HOG feature extraction and HOG tagsort training stage, but little computational resource is needed in identification comparison process, all adopt picture in training and testing process, do not rely on frame movable information before and after video, can split video.
(2) motion detection type: the video content analysis algorithm in this generic task depends on the front and back frame movable information of video content, can not carry out fractionation to analyze video to video, resource use amount is large;
In the present invention one specific embodiment, this generic task adopts Gaussian Background modeling method, and the method first will obtain the frame (such as the picture of 200 ~ 300 frames) of predetermined number from video, then sets up background model by mixed Gauss model.Finally, in detection process of moving target, when new frame of video occurs time, more new capital of background model parameters depends on the model parameter value of previous frame corresponding pixel points, when target video is HD video (such as 1080p), the renewal doing Gaussian Background reconstruction and background just needs a large amount of computational resources.In another specific embodiment of the present invention, this generic task also can adopt frame difference method, and current frame motion target detection relies on the picture pixels information of front 1 ~ 2 frame.
Due to, the video content analysis algorithm in these class methods relies on the front and back frame movable information of video content, and therefore, video, when analyzing video content, can not split by this generic task.
(3) on-line study class: the video content analysis algorithm in this generic task had both relied on the movable information of frame before and after video content, needed on-line study again, therefore needed a large amount of computational resources;
In the present invention one specific embodiment, this generic task adopts follows the tracks of study detection (TrackingLearning Detection, TLD) algorithm, it is by online real-time learning, upgrade moving object feature, a large amount of computer resources is needed, to ensure the coherent and fluency followed the tracks of in regeneration characteristics process.Video content analysis algorithm in this generic task had both relied on the movable information of frame before and after video content, and needed on-line study, needed a large amount of resources.
In the present invention one specific embodiment, the configuration interface of categorical attribute information is provided, allow user to carry out the configuration of categorical attribute information for single target video or the target video that meets pre-conditioned batch, video content analysis system provided by the invention is according to the type of the categorical attribute information determination target video analysis task of the target video content obtained.Such as, can be that user compares the type such as the recognition of face of easy understand, vehicle sails into, particular persons is followed the tracks of, particular vehicle tracking by categorical attribute information definition, system be according to the type of analysis task of the categorical attribute information of target video preset and the corresponding relation determination target video of previously defined analysis task type.
Step 106, corresponding relation according to analysis task type and preset scheduling strategy, determine the scheduling strategy of the analysis task of target video.
In one embodiment of the invention, the data center being used for analyzing target video is built based on Intel Virtualization Technology, if this data center is positioned at high in the clouds, then also cloud computing platform can be referred to as, in this data center, comprise and form server cluster by a large amount of physical servers for target video analysis, for storing the memory device of data and the network equipment for interconnecting, by Intel Virtualization Technology to the computational resource in this data center, storage resources and Internet resources carry out virtual, and on virtualized basis, by the corresponding relation of task dispatcher based target video analytic tasks type and preset scheduling strategy, determine the scheduling strategy of current analysis task, and dynamically distribute virtual resources according to the analysis task that scheduling strategy is target video, described virtual resources comprises computational resource, storage resources and Internet resources.
In one embodiment of the invention, task dispatcher is prefixed task scheduling strategy in following 7:
(1) performance scheduling strategy a: analysis task is split into several subtasks by this scheduling strategy, point subtask split out is assigned on the multiple computational resources in server cluster, perform in a parallel fashion, thus meet the performance requirement of fast processing.
(2) efficiency scheduling strategy: this scheduling strategy is shared multiple analysis task on a different server as far as possible, ensures the stability run, and reduces physical server load, ensures the treatment effeciency to many analysis task and success ratio.
(3) topological scheduling strategy: this scheduling strategy object is to reduce via net loss when analysis task performs on different server, when multiple server executed in parallel analysis task, the executing state by each subtask of Network Synchronization and transmission result data may be needed, therefore can to network build-up of pressure, in addition, when forming cloud platform by multiple data center, and on cloud platform when shared resource, network delay between the server at different pieces of information center crosses the efficiency that conference affects tasks carrying, therefore, the scheduling of analysis task needs the problem considering network topology and topological time delay, this scheduling strategy topology Network Based and network management information are selected in network topology contiguous as far as possible, multiple servers that time delay is low perform a parallel analysis task.
(4) algorithmic dispatching strategy: this scheduling strategy is according to the analytical algorithm of different video contents, and corresponding different video segment patterns, and Streaming Media cloud transmission mode, the processor calling different ability carries out algorithm operation.
(5) environmental suitability scheduling strategy: this scheduling strategy obtains present circumstances by sensor, described environment includes but not limited to temperature, humidity, uninterrupted power source indication information etc., whether the ambient conditions decision-making according to data center's zones of different uses the virtual resources being positioned at this region, such as when certain server region temperature is too high, reducing the Resourse Distribute in this region as far as possible.
(6) resource expansion scheduling strategy: this scheduling strategy is according to the loading condition at current data center, the use amount of flexible increase and reduction resource, such as when the analysis task of pre-treatment is less, the virtual machine quantity participating in analysis task is reduced by scheduler (or instruction virtual machine management platform), when analysis task amount is large, increase the virtual machine quantity participating in analysis task.Even can increase cloudy management on virtual basis, resource can be temporarily transferred on different physical region, realize expansion and the contraction of resource between cloud and cloud.
(7) green energy conservation scheduling strategy: this scheduling strategy is target with energy-conservation, reduces physical server shared by analysis task as much as possible, allows eachly to be maximized by the physical server utilization factor used, and reduces resource fragmentation.When fully loaded time, then open next physical server, improve energy utilization rate, energy-saving and cost-reducing.
The embodiment of the present invention uses green energy conservation scheduling strategy to the video content analysis task not reaching priority threshold requirement, to reaching analysis task that priority threshold requires based on the corresponding relation of preset analysis task type and scheduling strategy, determine the scheduling strategy of the analysis task of target video.
Described analysis task type and the corresponding relation of scheduling strategy as shown in Figure 2, describe according to analysis task type below respectively:
Aspect ratio is to class: the feature of this analysis task type needs a large amount of computer resources in feature extraction and classification based training stage, but few resources is needed in identification comparison process, all picture is adopted in training and testing process, do not rely on frame movable information before and after video, can split video.For improving computing velocity, this generic task carries out task scheduling by algorithmic dispatching strategy, performance scheduling strategy, efficiency scheduling strategy, environmental suitability scheduling strategy.Thus the efficiency utilization reached in training stage realization process resource, reduce the waste loss of system resource.
Motion detection type, the feature of this analysis task type is that video content analysis algorithm relies on the front and back frame movable information of video content, resource use amount is large and video can not be split video content analysis, therefore, efficiency scheduling strategy, algorithmic dispatching strategy and resource expansion scheduling strategy are chosen to this generic task.Choose efficiency scheduling strategy by algorithmic dispatching strategy to analyze video content, if resource requirement can not meet the real-time demand of this generic task, also on different physical region, temporarily transfer resource by resource expansion scheduling strategy, the expansion of resource and contraction between realizing.
On-line study class, the feature of this analysis task type is the movable information both having relied on frame before and after video content, and needs on-line study, needs a large amount of resources.Performance scheduling strategy, efficiency scheduling strategy, network topology scheduling strategy, algorithmic dispatching strategy, environmental suitability scheduling strategy and resource expansion scheduling strategy are chosen to these class methods.On-line study alanysis task, the method characteristic of the characteristics and comparison generic task type of its learning process is consistent, choose high performance processor by algorithmic dispatching strategy, network topology scheduling strategy and efficiency scheduling strategy to train, matching identification stage Algorithms of Selecting scheduling strategy works in coordination with efficiency scheduling strategy.
In the embodiment of the present invention, for the analysis task reaching priority threshold requirement, in order to ensure that the reliability service of hardware device all have chosen environmental suitability scheduling strategy.
It should be noted that, the scheduling strategy that above-mentioned each task type can use is stackable, and scheduler can according to circumstances therefrom select a kind of, multiple combination or whole scheduling strategy to carry out scheduling of resource to one or more video analytic tasks.
Step 108, be that the analysis task of target video distributes virtual resources according to determined scheduling strategy, perform described analysis task.
The present invention is by classifying to different content analytical approach in video, rational dynamic resource scheduling and Balance route are done to sorted video content analysis task, in guarantee is to video while different content fail-safe analysis, realize the efficiency utilization of process resource, reduce the waste loss of system resource.
A kind of structural representation based on virtualized video content analysis system that Fig. 3 provides for the embodiment of the present invention, this system 400 comprises:
Interface arrangement 460, this device is used for, from external reception target video data, target video data is sent to scheduler 440, exports from the result data of scheduler 440 receiving and analyzing task; The attribute information query interface of target video is provided;
Attribute acquisition device 410, the attribute information query interface that this device is used for being provided by interface arrangement 460 obtains priority attribute information, the categorical attribute information of target video;
Determining device 420, for determining the task priority of target video being carried out to video content analysis according to the priority attribute information of target video, and judge whether to use green energy conservation scheduling strategy to analyze target video according to task priority, the analysis task notice sorter 430 meeting priority requirement is classified, uses green energy conservation scheduling strategy to carry out task scheduling for the analysis task notice scheduler 440 not meeting priority requirement;
Sorter 430, for classifying to the analysis task meeting priority requirement according to categorical attribute information, determines the type of analysis task;
Scheduler 440, for the corresponding relation according to analysis task type and preset scheduling strategy, determines the scheduling strategy of the analysis task of target video; Dispatch according to the analysis task of determined scheduling strategy to target video, distribute virtual resources;
Virtual resources pond, for providing the virtual resources of video content analysis required by task, execution analysis task.
The video content analysis system realized based on Intel Virtualization Technology that the embodiment of the present invention provides, concerning user, can be regarded as the cloud platform of a video analysis, this cloud platform is a framework constructed by software and hardware that can carry out distributed analysis process to massive video data.This cloud platform has reliably, efficient, telescopic nature, wherein scheduler 440 is very important assemblies in cloud platform, its major function is that the idling-resource in system is distributed to each operation according to certain scheduling strategy, and it plays vital effect for the distribution of whole system virtual resource and Job execution.
In the video content analysis system that the embodiment of the present invention provides, algorithm characteristic and the demand of target video content analysis to resource of the target video content analysis that sorter 430 characterizes according to categorical attribute information are classified to analysis task, wherein, the type of analysis task comprises: aspect ratio is to class, motion detection type, on-line study class.
In the video content analysis system that the embodiment of the present invention provides, scheduler 440 comprises following sub-scheduler further, and the corresponding relation between each sub-scheduler and scheduling strategy is as follows:
The sub-scheduler 441 of green energy conservation, for dispatching the analysis task not meeting priority requirement;
The sub-scheduler 442 of comparing class, for dispatching analysis task based on performance scheduling strategy, efficiency scheduling strategy, algorithmic dispatching strategy, environmental suitability scheduling strategy;
The sub-scheduler 443 of motion detection type, for dispatching analysis task based on efficiency scheduling strategy, algorithmic dispatching strategy, environmental suitability scheduling strategy, resource expansion scheduling strategy;
The sub-scheduler 444 of on-line study class, for dispatching analysis task based on performance scheduling strategy, efficiency scheduling strategy, algorithmic dispatching strategy, environmental suitability scheduling strategy, resource expansion scheduling strategy.
In the video content analysis system that the embodiment of the present invention provides, a kind of, the multiple combination in corresponding scheduling strategy or whole selected by the sub-scheduler of comparing class 442, the sub-scheduler of motion detection type 443 and the sub-scheduler of on-line study class 444, carries out scheduling of resource to one or more video analytic tasks
In the video content analysis system that the embodiment of the present invention provides, in order to improve handling property, virtual resources is shared in the cloud platform that can form in multiple data center comprising described system, form shared virtual resources pond by the virtual resources pond in multiple data center, scheduler 440 is dispatched based on the analysis task of cloud platform to target video be made up of multiple data center.
Through the above description of the embodiments, those skilled in the art can be well understood to the mode that the present invention can add required general hardware platform by software and realize, and can certainly pass through hardware, but in a lot of situation, the former is better embodiment.Based on such understanding, technical scheme of the present invention can embody with the form of software product the part that prior art contributes in essence in other words, this computer software product is stored in a storage medium, comprising some instructions in order to make a computer equipment (can be personal computer, server, or the network equipment etc.) perform method described in each embodiment of the present invention.It will be appreciated by those skilled in the art that accompanying drawing is the schematic diagram of a preferred embodiment, the module in accompanying drawing or flow process might not be that enforcement the present invention is necessary.It will be appreciated by those skilled in the art that the module in the device in embodiment can carry out being distributed in the device of embodiment according to embodiment description, also can carry out respective change and be arranged in the one or more devices being different from the present embodiment.The module of above-described embodiment can merge into a module, also can split into multiple submodule further.The invention described above embodiment sequence number, just to describing, does not represent the quality of embodiment.Be only several specific embodiment of the present invention above, but the present invention is not limited thereto, the changes that any person skilled in the art can think of all should fall into protection scope of the present invention.
Claims (10)
1., based on a virtualized video content analysis method, it is characterized in that, comprising:
Obtain the priority attribute information of target video, determine the task priority of target video being carried out to video content analysis according to the priority attribute information of target video;
Green energy conservation scheduling strategy is used for the analysis task not meeting priority requirement;
For the analysis task meeting priority requirement, obtain the categorical attribute information of target video, classify according to the analysis task of categorical attribute information to target video, determine the type of analysis task;
According to the corresponding relation of analysis task type and preset scheduling strategy, determine the scheduling strategy of the analysis task of target video;
Dispatch according to the analysis task of determined scheduling strategy to target video, distribute virtual resources, perform described analysis task.
2. the method for claim 1, is characterized in that,
Described categorical attribute information is for characterizing the algorithm characteristic of target video content analysis and target video content analysis to the demand of resource, and the type of described analysis task comprises: aspect ratio is to class, motion detection type, on-line study class.
3. method as claimed in claim 2, it is characterized in that, the corresponding relation of described preset scheduling strategy and scheduling strategy and analysis task type comprises:
Performance scheduling strategy, is applied to aspect ratio to class and on-line study alanysis task;
Efficiency scheduling strategy, is applied to aspect ratio to class, motion detection type and on-line study alanysis task;
Network topology scheduling strategy, is applied to on-line study alanysis task;
Algorithmic dispatching strategy, is applied to aspect ratio to class, motion detection type and on-line study alanysis task;
Environmental suitability scheduling strategy, is applied to aspect ratio to class, motion detection type and on-line study alanysis task;
Resource expansion scheduling strategy, is applied to motion detection type and on-line study alanysis task.
4. method as claimed in claim 3, is characterized in that,
From the scheduling strategy that each analysis task type is corresponding, select a kind of, multiple combination or all scheduling of resource carried out to one or more video analytic tasks.
5. the method for claim 1, is characterized in that,
Form cloud platform by multiple data center, scheduling is carried out to the analysis task of target video and refers to dispatch on cloud platform, share the virtual resources in multiple data center.
6. based on a virtualized video content analysis system, it is characterized in that, comprising:
Interface arrangement, for target video data being sent to scheduler from external reception target video data, exports from the result data of scheduler receiving and analyzing task; The attribute information query interface of target video is provided;
Attribute acquisition device, for obtaining priority attribute information, the categorical attribute information of target video by interface arrangement;
Determining device, for determining the task priority of target video being carried out to video content analysis according to the priority attribute information of target video, and judge whether to use green energy conservation scheduling strategy to analyze target video according to task priority, the analysis task notice sorter meeting priority requirement is classified, uses green energy conservation scheduling strategy to carry out task scheduling for the analysis task notice scheduler not meeting priority requirement;
Sorter, for classifying to the analysis task meeting priority requirement according to categorical attribute information, determines the type of analysis task;
Scheduler, for the corresponding relation according to analysis task type and preset scheduling strategy, determines the scheduling strategy of the analysis task of target video; Dispatch according to the analysis task of determined scheduling strategy to target video, distribute virtual resources;
Virtual resources pond, for providing the virtual resources of video content analysis required by task, execution analysis task.
7. system as claimed in claim 6, is characterized in that,
Algorithm characteristic and the demand of target video content analysis to resource of the target video content analysis that described sorter characterizes according to described categorical attribute information are classified to analysis task, and the type of described analysis task comprises: aspect ratio is to class, motion detection type, on-line study class.
8. system as claimed in claim 7, it is characterized in that, described scheduler comprises:
The sub-scheduler of green energy conservation, for dispatching the analysis task not meeting priority requirement;
The sub-scheduler of comparing class, for dispatching analysis task based on performance scheduling strategy, efficiency scheduling strategy, algorithmic dispatching strategy, environmental suitability scheduling strategy;
The sub-scheduler of motion detection type, for dispatching analysis task based on efficiency scheduling strategy, algorithmic dispatching strategy, environmental suitability scheduling strategy, resource expansion scheduling strategy;
The sub-scheduler of on-line study class, for dispatching analysis task based on performance scheduling strategy, efficiency scheduling strategy, algorithmic dispatching strategy, environmental suitability scheduling strategy, resource expansion scheduling strategy.
9. system as claimed in claim 8, is characterized in that,
A kind of, the multiple combination in corresponding scheduling strategy or whole selected by the sub-scheduler of described comparing class, the sub-scheduler of motion detection type and the sub-scheduler of on-line study class, carries out scheduling of resource to one or more video analytic tasks.
10. system as claimed in claim 6, is characterized in that,
The shared virtual resources pond of described virtual resources pond for being made up of the virtual resources pond in multiple data center;
Described scheduler is dispatched based on the analysis task of cloud platform to target video be made up of multiple data center.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410830093.0A CN104657205B (en) | 2014-12-26 | 2014-12-26 | A kind of video content analysis method and system based on virtualization |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410830093.0A CN104657205B (en) | 2014-12-26 | 2014-12-26 | A kind of video content analysis method and system based on virtualization |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN104657205A true CN104657205A (en) | 2015-05-27 |
| CN104657205B CN104657205B (en) | 2018-12-14 |
Family
ID=53248379
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201410830093.0A Active CN104657205B (en) | 2014-12-26 | 2014-12-26 | A kind of video content analysis method and system based on virtualization |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN104657205B (en) |
Cited By (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105448125A (en) * | 2015-12-28 | 2016-03-30 | 天维尔信息科技股份有限公司 | Rescue vehicle management method and device |
| CN105975513A (en) * | 2016-04-27 | 2016-09-28 | 天脉聚源(北京)传媒科技有限公司 | Method and apparatus for setting identifiers of video resources in batches |
| CN106878389A (en) * | 2017-01-04 | 2017-06-20 | 北京百度网讯科技有限公司 | Method and apparatus for carrying out scheduling of resource in cloud system |
| CN110187961A (en) * | 2019-04-25 | 2019-08-30 | 北京易华录信息技术股份有限公司 | A kind of video data processing system and method |
| CN111222739A (en) * | 2019-10-22 | 2020-06-02 | 岭澳核电有限公司 | Task allocation method and task allocation system of nuclear power station |
| CN112035256A (en) * | 2020-08-28 | 2020-12-04 | 北京字节跳动网络技术有限公司 | Resource allocation method, device, electronic equipment and medium |
| CN113542807A (en) * | 2021-09-14 | 2021-10-22 | 杭州博雅鸿图视频技术有限公司 | Resource management scheduling method and system based on digital retina platform |
| CN113836290A (en) * | 2021-09-17 | 2021-12-24 | 掌阅科技股份有限公司 | Content recognition method, electronic device, and storage medium |
| CN114584804A (en) * | 2022-03-11 | 2022-06-03 | 广州慧思软件科技有限公司 | Virtual reality video stream data processing system |
| CN115375208A (en) * | 2022-10-26 | 2022-11-22 | 中国铁塔股份有限公司 | Camera data analysis method and device, electronic equipment and storage medium |
| CN116108492A (en) * | 2023-04-07 | 2023-05-12 | 安羚科技(杭州)有限公司 | Laterally expandable data leakage prevention system |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101105804A (en) * | 2006-07-10 | 2008-01-16 | 索尼株式会社 | Information management apparatus, image recording apparatus, information management method, and program |
| CN101211215A (en) * | 2006-12-29 | 2008-07-02 | Arm有限公司 | Performance of a processor in a managing data processing equipment |
| CN101821696A (en) * | 2007-08-28 | 2010-09-01 | Commvault系统公司 | Power management of data processing resources, such as power adaptive management of data storage operations |
| CN103023980A (en) * | 2012-11-21 | 2013-04-03 | 中国电信股份有限公司云计算分公司 | Method and system for processing user service request by cloud platform |
| CN103795804A (en) * | 2014-02-24 | 2014-05-14 | 华为技术有限公司 | Storage resource scheduling method and storage calculation system |
-
2014
- 2014-12-26 CN CN201410830093.0A patent/CN104657205B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101105804A (en) * | 2006-07-10 | 2008-01-16 | 索尼株式会社 | Information management apparatus, image recording apparatus, information management method, and program |
| CN101211215A (en) * | 2006-12-29 | 2008-07-02 | Arm有限公司 | Performance of a processor in a managing data processing equipment |
| CN101821696A (en) * | 2007-08-28 | 2010-09-01 | Commvault系统公司 | Power management of data processing resources, such as power adaptive management of data storage operations |
| CN103023980A (en) * | 2012-11-21 | 2013-04-03 | 中国电信股份有限公司云计算分公司 | Method and system for processing user service request by cloud platform |
| CN103795804A (en) * | 2014-02-24 | 2014-05-14 | 华为技术有限公司 | Storage resource scheduling method and storage calculation system |
Cited By (16)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105448125A (en) * | 2015-12-28 | 2016-03-30 | 天维尔信息科技股份有限公司 | Rescue vehicle management method and device |
| CN105975513A (en) * | 2016-04-27 | 2016-09-28 | 天脉聚源(北京)传媒科技有限公司 | Method and apparatus for setting identifiers of video resources in batches |
| CN106878389A (en) * | 2017-01-04 | 2017-06-20 | 北京百度网讯科技有限公司 | Method and apparatus for carrying out scheduling of resource in cloud system |
| CN106878389B (en) * | 2017-01-04 | 2020-02-07 | 北京百度网讯科技有限公司 | Method and device for resource scheduling in cloud system |
| US10778807B2 (en) | 2017-01-04 | 2020-09-15 | Beijing Baidu Netcom Science And Technology Co., Ltd. | Scheduling cluster resources to a job based on its type, particular scheduling algorithm,and resource availability in a particular resource stability sub-levels |
| CN110187961A (en) * | 2019-04-25 | 2019-08-30 | 北京易华录信息技术股份有限公司 | A kind of video data processing system and method |
| CN111222739B (en) * | 2019-10-22 | 2023-08-25 | 岭澳核电有限公司 | Nuclear power station task allocation method and nuclear power station task allocation system |
| CN111222739A (en) * | 2019-10-22 | 2020-06-02 | 岭澳核电有限公司 | Task allocation method and task allocation system of nuclear power station |
| CN112035256A (en) * | 2020-08-28 | 2020-12-04 | 北京字节跳动网络技术有限公司 | Resource allocation method, device, electronic equipment and medium |
| CN113542807A (en) * | 2021-09-14 | 2021-10-22 | 杭州博雅鸿图视频技术有限公司 | Resource management scheduling method and system based on digital retina platform |
| CN113542807B (en) * | 2021-09-14 | 2022-02-22 | 杭州博雅鸿图视频技术有限公司 | Resource management scheduling method and system based on digital retina platform |
| CN113836290A (en) * | 2021-09-17 | 2021-12-24 | 掌阅科技股份有限公司 | Content recognition method, electronic device, and storage medium |
| CN114584804A (en) * | 2022-03-11 | 2022-06-03 | 广州慧思软件科技有限公司 | Virtual reality video stream data processing system |
| CN114584804B (en) * | 2022-03-11 | 2023-06-23 | 云南电信公众信息产业有限公司 | Virtual reality video stream data processing system |
| CN115375208A (en) * | 2022-10-26 | 2022-11-22 | 中国铁塔股份有限公司 | Camera data analysis method and device, electronic equipment and storage medium |
| CN116108492A (en) * | 2023-04-07 | 2023-05-12 | 安羚科技(杭州)有限公司 | Laterally expandable data leakage prevention system |
Also Published As
| Publication number | Publication date |
|---|---|
| CN104657205B (en) | 2018-12-14 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN104657205A (en) | Virtualization-based video content analyzing method and system | |
| Canel et al. | Scaling video analytics on constrained edge nodes | |
| Liu et al. | Resource preprocessing and optimal task scheduling in cloud computing environments | |
| CN110442428B (en) | Coordination method of Docker container | |
| CN117472587A (en) | Resource scheduling system of AI intelligent computation center | |
| Chen et al. | Distributed real-time object detection based on edge-cloud collaboration for smart video surveillance applications | |
| CN117389742B (en) | Edge computing method, device and storage medium for machine vision | |
| CN111506434A (en) | Task processing method and device and computer readable storage medium | |
| Zhou et al. | EVCT: An efficient VM deployment algorithm for a software-defined data center in a connected and autonomous vehicle environment | |
| CN117215764A (en) | Computing power resource processing method, device, equipment and storage medium | |
| CN117170886B (en) | Continuous learning resource scheduling method and device for large-connection video analysis | |
| Volkov et al. | IoT traffic prediction with neural networks learning based on SDN infrastructure | |
| Xu et al. | Edge Video Analytics: A Survey on Applications, Systems and Enabling Techniques | |
| CN116402318B (en) | Multi-stage computing power resource distribution method and device for power distribution network and network architecture | |
| Yang et al. | Multi-resource allocation for virtual machine placement in video surveillance cloud | |
| Ashouri et al. | Analyzing distributed deep neural network deployment on edge and cloud nodes in IoT systems | |
| CN106485412B (en) | A kind of substation's image intelligent level of analysis multiple task distributive calculation method | |
| Lee et al. | Towards quality aware collaborative video analytic cloud | |
| Moussa et al. | Service management in the edge cloud for stream processing of IoT data | |
| Wang et al. | Real-Time AI in Social Edge | |
| Pu et al. | An elastic framework construction method based on task migration in edge computing | |
| Yang et al. | Latency-Aware Microservice Deployment for Edge AI Enabled Video Analytics | |
| Kimovski et al. | Cloud—edge offloading model for vehicular traffic analysis | |
| Kabir et al. | A cloud-based robot framework for indoor object identification using unsupervised segmentation technique and convolution neural network (CNN) | |
| CN116866352B (en) | Cloud-edge-coordinated intelligent camera system |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |