CN111461958A - System and method for controlling real-time detection and optimization processing of rapid multi-path data streams - Google Patents
System and method for controlling real-time detection and optimization processing of rapid multi-path data streams Download PDFInfo
- Publication number
- CN111461958A CN111461958A CN202010417390.8A CN202010417390A CN111461958A CN 111461958 A CN111461958 A CN 111461958A CN 202010417390 A CN202010417390 A CN 202010417390A CN 111461958 A CN111461958 A CN 111461958A
- Authority
- CN
- China
- Prior art keywords
- model
- picture
- detection
- data
- module
- 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
- 238000000034 method Methods 0.000 title claims abstract description 16
- 238000011897 real-time detection Methods 0.000 title claims abstract description 11
- 238000005457 optimization Methods 0.000 title claims abstract description 6
- 238000001514 detection method Methods 0.000 claims abstract description 41
- 238000003672 processing method Methods 0.000 claims description 3
- 238000004886 process control Methods 0.000 claims 1
- 230000007547 defect Effects 0.000 abstract description 5
- 238000010191 image analysis Methods 0.000 abstract description 2
- 238000012544 monitoring process Methods 0.000 abstract description 2
- 238000010586 diagram Methods 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000003203 everyday effect Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T1/00—General purpose image data processing
- G06T1/20—Processor architectures; Processor configuration, e.g. pipelining
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/0002—Inspection of images, e.g. flaw detection
- G06T7/0004—Industrial image inspection
- G06T7/0008—Industrial image inspection checking presence/absence
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2200/00—Indexing scheme for image data processing or generation, in general
- G06T2200/28—Indexing scheme for image data processing or generation, in general involving image processing hardware
Abstract
The invention discloses a system and a method for controlling the real-time detection and optimization processing of a rapid multi-path data stream, wherein a data distribution module comprises the following components: receiving a picture to be detected, distributing the picture to a processing module of each application, and counting the number of detected pictures per second of each model for a model scheduling module to use; a model scheduling module: acquiring the number of detected pictures per second of each model from the data distribution module, and starting a new data processing module to shunt the heavy-load models; redundant data processing modules are closed, and system resources are saved; a data processing module: the dynamic scheduling of multiple defect detection intelligent detection applications and the load balance of the multi-channel real-time monitoring video stream are realized, and the image analysis speed is increased. The invention meets the requirement of multipath large data volume detection and has the characteristics of high real-time performance, high reliability and the like.
Description
Technical Field
The invention relates to the technical field of data processing, in particular to a system and a method for controlling the real-time detection and optimization processing of a fast multi-path data stream.
Background
The power lines are numerous, and an unmanned aerial vehicle patrolling and examining every day can acquire massive high-definition pictures, and meanwhile, the defect types of each picture to be detected are numerous, so that high requirements on the speed of defect identification are provided. Fig. 3 is a prior art flow chart. In the prior art, the received image needs to be stored locally, then the depth model is called to calculate, and a result is returned. After a batch of images are received, the model is required to be called for calculation, the time consumption of calling the model is dozens of times of that of calculation, and parallel processing cannot be performed.
Disclosure of Invention
In order to solve the technical defects in the prior art, the invention aims to provide a system and a method for controlling the real-time detection and optimization processing of a fast multi-path data stream, aiming at improving the processing speed of the whole detection process and increasing the parallel processing capacity.
The technical solution adopted by the invention is as follows:
a fast multi-path data flow real-time detection optimizing processing control system is composed of the following modules:
the data distribution module: receiving a picture to be detected, distributing the picture to a processing module of each application, and counting the number of detected pictures per second of each model for a model scheduling module to use;
a model scheduling module: the number of pictures detected by each model per second is obtained from the data distribution module, a new data processing module is started to shunt for the heavy-load model under the condition of sufficient resources, and redundant data processing modules are closed under the condition of small data quantity, so that system resources are saved;
a data processing module: the method comprises the steps of receiving a picture to be detected, carrying out intelligent detection on the picture, enabling each application to have a corresponding data processing module, continuously loading a detection model into a GPU (graphics processing unit) by the data processing modules, directly sending the detection model into the model for intelligent detection without repeatedly loading the model after the picture is received, and returning a final result.
The number of the data processing modules is more than 1.
A real-time detection optimizing processing method for fast multi-path data stream includes the following steps:
(1) receiving an input picture to be detected, distributing the picture to a processing module of each application, and counting the number of the detected pictures per second of each model for a model scheduling module to use;
(2) detecting data of the number of pictures per second according to the obtained models, predicting whether the processing data amount in unit time is within a threshold range, starting new detection service for shunting of heavy-load models under the condition of sufficient resources, and closing redundant detection service under the condition of small data amount, so that system resources are saved;
(3) and receiving the distributed picture to be detected, intelligently detecting the picture, wherein each application is provided with a corresponding data processing module, the module continuously loads the detection model into the GPU, directly sends the detection model into the GPU for intelligent detection without repeatedly loading the model after receiving the picture, and returns a final result.
The invention has the beneficial effects that: the invention realizes the dynamic scheduling of various defect detection intelligent detection applications and the load balance of multi-channel real-time monitoring video streams based on an application service mode, improves the image analysis speed, averagely shortens the detection time of a single picture and a single application to be within 0.2 second, meets the requirement of multi-channel large data volume detection, and has the characteristics of high instantaneity, high reliability and the like.
Drawings
FIG. 1 is a diagram of a control system according to the present invention.
FIG. 2 is a flow chart of the method of the present invention.
Fig. 3 is a prior art flow chart.
Detailed Description
The invention will now be further described with reference to figures 1 and 2,
fig. 1 is a structural diagram of a control system of the present invention, which is composed of the following modules:
the data distribution module: receiving a picture to be detected, distributing the picture to a processing module of each application, and counting the number of detected pictures per second of each model for a model scheduling module to use;
a model scheduling module: the number of pictures detected by each model per second is obtained from the data distribution module, a new data processing module is started to shunt for the heavy-load model under the condition of sufficient resources, and redundant data processing modules are closed under the condition of small data quantity, so that system resources are saved;
a data processing module: the method comprises the steps of receiving a picture to be detected, carrying out intelligent detection on the picture, enabling each application to have a corresponding data processing module, continuously loading a detection model into a GPU (graphics processing unit) by the data processing modules, directly sending the detection model into the model for intelligent detection without repeatedly loading the model after the picture is received, and returning a final result.
The number of the data processing modules is more than 1.
A real-time detection optimizing processing method for fast multi-path data stream, fig. 2 is a flow chart of the method of the invention. The method comprises the following steps:
(1) receiving an input picture to be detected, distributing the picture to a processing module of each application, and counting the number of the detected pictures per second of each model for a model scheduling module to use;
(2) detecting data of the number of pictures per second according to the obtained models, predicting whether the processing data amount in unit time is within a threshold range, starting new detection service for shunting of heavy-load models under the condition of sufficient resources, and closing redundant detection service under the condition of small data amount, so that system resources are saved;
(3) and receiving the distributed picture to be detected, intelligently detecting the picture, wherein each application is provided with a corresponding data processing module, the module continuously loads the detection model into the GPU, directly sends the detection model into the GPU for intelligent detection without repeatedly loading the model after receiving the picture, and returns a final result.
According to the invention, the data detection module is packaged in an http service mode, so that the model is solidified in the GPU, reloading of the model during each detection is avoided, the processing time of the detection process is greatly reduced, and the model loading time is saved for each picture by about 5s compared with that of a traditional detection mode; and through the model scheduling module, GPU resources are reasonably distributed, so that more models are used for distributing more resources.
According to the method and the device, model scheduling is carried out according to the access amount, more service processes are started for the model with large access amount, and therefore the utilization rate of GPU resources is improved. And the depth model is continuously loaded in the GPU in a service mode, so that the condition that the model needs to be reloaded in each detection is avoided, the time of the detection process is greatly reduced, and the utilization rate of GPU resources is further improved.
The above description is only a preferred embodiment of the present invention, and the protection scope of the present invention is not limited to the above embodiments, and all technical solutions belonging to the idea of the present invention belong to the protection scope of the present invention. It should be noted that modifications and embellishments within the scope of the invention may occur to those skilled in the art without departing from the principle of the invention, and are considered to be within the scope of the invention.
Claims (3)
1. A fast multi-path data flow real-time detection optimizing processing control system is characterized by comprising the following modules:
(1) the data distribution module: receiving a picture to be detected, distributing the picture to a processing module of each application, and counting the number of detected pictures per second of each model for a model scheduling module to use;
(2) a model scheduling module: acquiring the number of detected pictures per second of each model from the data distribution module, and starting a new data processing module to shunt the heavy-load models; redundant data processing modules are closed, and system resources are saved;
(3) a data processing module: the method comprises the steps of receiving a picture to be detected, carrying out intelligent detection on the picture, enabling each application to have a corresponding data processing module, continuously loading a detection model into a GPU (graphics processing unit) by the data processing modules, directly sending the detection model into the model for intelligent detection without repeatedly loading the model after the picture is received, and returning a final result.
2. The fast multi-path data stream real-time detection optimizing process control system of claim 1, wherein: the number of the data processing modules is at least 1.
3. A real-time detection optimization processing method for fast multi-path data streams is characterized by comprising the following steps:
step 1: the data distribution module receives an input picture to be detected and distributes the picture to the processing module of each application, and counts the number of the detected pictures per second of each model for the model scheduling module to use;
step 2: detecting data of the number of pictures per second according to the obtained models, predicting whether the processing data amount in unit time is within a threshold range, starting new detection service for shunting of heavy-load models under the condition of sufficient resources, and closing redundant detection service under the condition of small data amount, so that system resources are saved;
and step 3: the data processing module receives the distributed picture to be detected and carries out intelligent detection on the picture, each application is provided with a corresponding data processing module, the detection model is continuously loaded in the GPU by the data processing module, after the picture is received, the model is directly sent to the model without repeatedly loading the model for intelligent detection, and a final result is returned.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010417390.8A CN111461958A (en) | 2020-05-18 | 2020-05-18 | System and method for controlling real-time detection and optimization processing of rapid multi-path data streams |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010417390.8A CN111461958A (en) | 2020-05-18 | 2020-05-18 | System and method for controlling real-time detection and optimization processing of rapid multi-path data streams |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN111461958A true CN111461958A (en) | 2020-07-28 |
Family
ID=71686068
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010417390.8A Pending CN111461958A (en) | 2020-05-18 | 2020-05-18 | System and method for controlling real-time detection and optimization processing of rapid multi-path data streams |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111461958A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116629711A (en) * | 2023-07-21 | 2023-08-22 | 苏州友得利金属制品有限公司 | Shunting processing method and system based on steel pipe quality detection |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108230317A (en) * | 2018-01-09 | 2018-06-29 | 北京百度网讯科技有限公司 | Steel plate defect detection sorting technique, device, equipment and computer-readable medium |
| US20180232321A1 (en) * | 2017-02-16 | 2018-08-16 | Qualcomm Incorporated | Optimizing network driver performance and power consumption in multi-core processor-based systems |
| US20190121673A1 (en) * | 2017-10-19 | 2019-04-25 | Pure Storage, Inc. | Data transformation caching in an artificial intelligence infrastructure |
| CN110765288A (en) * | 2019-09-04 | 2020-02-07 | 北京旷视科技有限公司 | Image information synchronization method, device and system and storage medium |
| CN110856018A (en) * | 2019-11-14 | 2020-02-28 | 武汉珞佳伟业科技有限公司 | Rapid transcoding method and system in monitoring system based on cloud computing |
-
2020
- 2020-05-18 CN CN202010417390.8A patent/CN111461958A/en active Pending
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20180232321A1 (en) * | 2017-02-16 | 2018-08-16 | Qualcomm Incorporated | Optimizing network driver performance and power consumption in multi-core processor-based systems |
| US20190121673A1 (en) * | 2017-10-19 | 2019-04-25 | Pure Storage, Inc. | Data transformation caching in an artificial intelligence infrastructure |
| CN108230317A (en) * | 2018-01-09 | 2018-06-29 | 北京百度网讯科技有限公司 | Steel plate defect detection sorting technique, device, equipment and computer-readable medium |
| CN110765288A (en) * | 2019-09-04 | 2020-02-07 | 北京旷视科技有限公司 | Image information synchronization method, device and system and storage medium |
| CN110856018A (en) * | 2019-11-14 | 2020-02-28 | 武汉珞佳伟业科技有限公司 | Rapid transcoding method and system in monitoring system based on cloud computing |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116629711A (en) * | 2023-07-21 | 2023-08-22 | 苏州友得利金属制品有限公司 | Shunting processing method and system based on steel pipe quality detection |
| CN116629711B (en) * | 2023-07-21 | 2023-10-27 | 苏州友得利金属制品有限公司 | Shunting processing method and system based on steel pipe quality detection |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107067365A (en) | The embedded real-time video stream processing system of distribution and method based on deep learning | |
| CN116260990B (en) | AI asynchronous detection and real-time rendering method and system for multipath video streams | |
| CN116010054A (en) | Heterogeneous edge cloud AI system task scheduling frame based on reinforcement learning | |
| CN111461958A (en) | System and method for controlling real-time detection and optimization processing of rapid multi-path data streams | |
| CN112132022A (en) | Face snapshot framework, face snapshot method, device, equipment and storage medium | |
| CN109842521A (en) | A kind of mobile terminal delay machine forecasting system and method | |
| CN110677623B (en) | Data processing method, device, equipment and storage medium | |
| CN108762936A (en) | Distributed computing system based on artificial intelligence image recognition and method | |
| CN113286175A (en) | Video stream processing method, device and storage medium | |
| CN116582693B (en) | Camera calling control method based on video resource pool | |
| CN112817753A (en) | Task processing method and device, storage medium and electronic device | |
| Lu et al. | Dynamic offloading on a hybrid edge–cloud architecture for multiobject tracking | |
| CN107707921B (en) | Dynamic image processing system | |
| CN116152744A (en) | Dynamic detection method and device for electric vehicle, computer equipment and storage medium | |
| CN114363185B (en) | Virtual resource processing method and device | |
| CN111147603A (en) | Method and device for networking reasoning service | |
| CN115941907A (en) | RTP data packet sending method, system, electronic equipment and storage medium | |
| CN113438417A (en) | Method, system, medium and device for capturing object to be identified by video | |
| CN114257618A (en) | Vehicle operation data real-time analysis system based on Internet of vehicles platform | |
| CN113743235A (en) | Electric power inspection image processing method, device and equipment based on edge calculation | |
| CN116860657B (en) | Method and device for processing pressure control, computer equipment and storage medium | |
| CN114640669A (en) | Edge calculation method and device | |
| CN114972550B (en) | Edge calculation method for real-time video stream analysis | |
| CN110020613B (en) | Front-end face real-time detection method based on Jetson TX1 platform | |
| CN110659571A (en) | Streaming video face detection acceleration method based on frame buffer queue |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20200728 |
|
| RJ01 | Rejection of invention patent application after publication |