CN116468990B - Task random dispatch intelligent management system and method based on centralized judgment chart - Google Patents

Task random dispatch intelligent management system and method based on centralized judgment chart Download PDF

Info

Publication number
CN116468990B
CN116468990B CN202310671306.9A CN202310671306A CN116468990B CN 116468990 B CN116468990 B CN 116468990B CN 202310671306 A CN202310671306 A CN 202310671306A CN 116468990 B CN116468990 B CN 116468990B
Authority
CN
China
Prior art keywords
task
representing
image
module
judging
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.)
Active
Application number
CN202310671306.9A
Other languages
Chinese (zh)
Other versions
CN116468990A (en
Inventor
邵雅琪
刘筠
赵新宇
金金
阮博
左海波
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
CHINA INTELLECTECH (BEIJING) CORP
Original Assignee
CHINA INTELLECTECH (BEIJING) CORP
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by CHINA INTELLECTECH (BEIJING) CORP filed Critical CHINA INTELLECTECH (BEIJING) CORP
Priority to CN202310671306.9A priority Critical patent/CN116468990B/en
Publication of CN116468990A publication Critical patent/CN116468990A/en
Application granted granted Critical
Publication of CN116468990B publication Critical patent/CN116468990B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/94—Hardware or software architectures specially adapted for image or video understanding
    • G06V10/95—Hardware or software architectures specially adapted for image or video understanding structured as a network, e.g. client-server architectures
    • 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
    • G06N3/04—Architecture, e.g. interconnection topology
    • G06N3/0464—Convolutional networks [CNN, ConvNet]
    • 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
    • G06N3/04—Architecture, e.g. interconnection topology
    • G06N3/048—Activation functions
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q10/00—Administration; Management
    • G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
    • G06Q10/063—Operations research, analysis or management
    • G06Q10/0631—Resource planning, allocation, distributing or scheduling for enterprises or organisations
    • G06Q10/06311—Scheduling, planning or task assignment for a person or group
    • G06Q10/063112—Skill-based matching of a person or a group to a task
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q10/00—Administration; Management
    • G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
    • G06Q10/063—Operations research, analysis or management
    • G06Q10/0633—Workflow analysis
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/20—Image preprocessing
    • G06V10/26—Segmentation of patterns in the image field; Cutting or merging of image elements to establish the pattern region, e.g. clustering-based techniques; Detection of occlusion
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/40—Extraction of image or video features
    • G06V10/44—Local feature extraction by analysis of parts of the pattern, e.g. by detecting edges, contours, loops, corners, strokes or intersections; Connectivity analysis, e.g. of connected components
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/40—Extraction of image or video features
    • G06V10/52—Scale-space analysis, e.g. wavelet analysis
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
    • G06V10/77—Processing image or video features in feature spaces; using data integration or data reduction, e.g. principal component analysis [PCA] or independent component analysis [ICA] or self-organising maps [SOM]; Blind source separation
    • G06V10/80—Fusion, i.e. combining data from various sources at the sensor level, preprocessing level, feature extraction level or classification level
    • G06V10/806—Fusion, i.e. combining data from various sources at the sensor level, preprocessing level, feature extraction level or classification level of extracted features
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
    • G06V10/82—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using neural networks

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Business, Economics & Management (AREA)
  • General Physics & Mathematics (AREA)
  • Human Resources & Organizations (AREA)
  • Multimedia (AREA)
  • Evolutionary Computation (AREA)
  • Software Systems (AREA)
  • Entrepreneurship & Innovation (AREA)
  • Artificial Intelligence (AREA)
  • Computing Systems (AREA)
  • General Health & Medical Sciences (AREA)
  • Strategic Management (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Economics (AREA)
  • Health & Medical Sciences (AREA)
  • Educational Administration (AREA)
  • Biophysics (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • General Engineering & Computer Science (AREA)
  • Mathematical Physics (AREA)
  • Data Mining & Analysis (AREA)
  • Computational Linguistics (AREA)
  • Databases & Information Systems (AREA)
  • Development Economics (AREA)
  • Biomedical Technology (AREA)
  • Molecular Biology (AREA)
  • Game Theory and Decision Science (AREA)
  • Medical Informatics (AREA)
  • Marketing (AREA)
  • Operations Research (AREA)
  • Quality & Reliability (AREA)
  • Tourism & Hospitality (AREA)
  • General Business, Economics & Management (AREA)
  • Image Analysis (AREA)

Abstract

The application provides a task random dispatch intelligent management system and method based on a centralized judgment chart, which are characterized in that images in a task are acquired, the images are uniformly divided into image blocks with equal length and width, an image contour model is constructed, and image segmentation is carried out; extracting the characteristic information of the task according to the image contour model, sending the extracted characteristic information to a processing terminal, reducing the workload of the processing terminal, setting up a difficulty level and a state level of the processing terminal, randomly distributing the task according to the corresponding level, selecting proper graph judging seat groups for tasks with different difficulties, randomly distributing the task to be distributed within a limited range, improving the task processing efficiency, avoiding the task backlog problem caused by unreasonable task distribution, and simultaneously reducing the transmission quantity of task data, thereby having small bandwidth requirements and avoiding generating network congestion.

Description

Task random dispatch intelligent management system and method based on centralized judgment chart
Technical Field
The application relates to the field of centralized graph judgment, in particular to a task random dispatch intelligent management system and method based on centralized graph judgment.
Background
In the centralized graph judging system, the existing task allocation modes mainly comprise two implementation modes of centralized allocation and distributed allocation. The centralized distribution is realized by a special scheduling module, the scheduling module is connected with all security check devices and all graph judging workstations in the centralized graph judging system, and security check tasks generated by the security check devices are optimally distributed in all the graph judging workstations. In the task allocation process, huge task data needs to be transmitted in the whole network of the centralized graph judging system, the requirement on network bandwidth is high, and once the bandwidth is limited, the problems of transmission delay and task allocation failure can occur.
Chinese patent application number: CN202010718488.7, publication date: 2022.01.25 discloses a task allocation method, device, system and computer storage medium, the method comprises: when the target security inspection equipment generates a task to be distributed, determining a graph judging workstation in an idle state in a target group associated with the target security inspection equipment; and distributing target graph judging workstations for the tasks to be distributed according to the preset priority order of the graph judging workstations in the idle state. The application can avoid task allocation in the whole network range of the centralized graph judging system, and has the advantages of small bandwidth requirement on the backbone network of the system and suitability for the situation of limited bandwidth of the backbone network.
However, in the process of implementing the technical scheme of the embodiment of the application, the inventor discovers that the above technology has at least the following technical problems: the prior art does not integrate a large number of tasks into zero, so that backlog of the tasks is caused, the workload of graph judging seats is heavy, and the error rate is easily increased.
Disclosure of Invention
The embodiment of the application solves the problem that the prior art does not integrate a large number of tasks into zero, which results in backlog of the tasks, the graph judging seat has heavy workload and easily causes the increase of error rate, thereby greatly reducing the computational complexity, accelerating the task transmission, reducing the workload of the graph judging seat and improving the task processing efficiency.
The application provides a task random dispatch intelligent management system and method based on a centralized judgment chart, which concretely comprises the following technical scheme:
an intelligent management system for task random dispatch based on a centralized judgment chart comprises the following parts:
the system comprises a task generating module, a task processing module, a task simplifying module, a broadcasting module, a grading module, a task distributing module, a task management module and a transmission module;
the task processing module is used for acquiring an image to be interpreted in the graph judging task, carrying out grey-scale processing on the image to be interpreted, dividing the grey-scale image, and connecting the task processing module with the task simplifying module and the grading module in a data transmission mode;
the task simplification module is used for extracting the characteristic information of the graph judging task according to the image contour model, wherein the extracting mode is double-channel distributed parallel processing, the extracted characteristic information is sent to the graph judging seat, and the task simplification module is connected with the transmission module in a data transmission mode;
the grading module is used for counting the number of the image blocks to be identified in each image judging task, setting up image judging difficulty grades, and grading the image judging tasks into image judging tasks with different difficulty grades according to the number of the image blocks to be identified in the image judging tasks; the working completion efficiency and the idle state of each graph judging seat are subjected to state grading, and the grading module is connected with the task dispatch module in a data transmission mode;
the task distribution module is used for randomly distributing the task by selecting on-line graph judging seats in the state grades with the same grade according to the grade number of the difficulty, namely the grade number of the graph judging task is the same as the grade number of the state of the graph judging seats, and the task distribution module is connected with the task simplification module in a data transmission mode.
An intelligent management method for task random dispatch based on a centralized judgment chart comprises the following steps:
s1, acquiring an image to be interpreted in an image judging task, uniformly dividing the image to be interpreted into image blocks with equal length and width, constructing an image contour model, and carrying out image segmentation;
s2, extracting characteristic information of the graph judging task according to the image contour model, sending the extracted characteristic information to the graph judging seat, reducing the workload of the graph judging seat, setting a graph judging difficulty level and a state level of the graph judging seat, and carrying out task random distribution according to the corresponding level.
Preferably, the step S1 specifically includes:
uniformly dividing an image to be interpreted into R image blocks with equal length and width, and representing each image block as:
，
wherein ,representing the r-th image block,/->Representing Fourier basis-> and />Background pixels and object pixels representing the r-th image block, respectively, < >>And carrying out edge matching on the R image blocks to construct an image contour model.
Preferably, the step S2 specifically includes: and inputting the segmented image blocks into a feature extraction module, wherein the extraction mode is double-channel distributed parallel processing.
Preferably, the step S2 specifically includes: the first channel passes through the attention mechanism layerAnd then pass throughConvolution layers and pooling layers of size.
Preferably, the step S2 specifically includes: the second channel is subjected to a depth-separable convolution process and then to a processAnd the convolution layer and the pooling layer with the sizes are used for carrying out multi-scale feature fusion on the output of the two channels to obtain the final image block features.
Preferably, the step S2 specifically includes: the multi-scale feature fusion is carried out by oneThe depth of the (c) separable convolution layer, then passing through a 4-time up-sampling layer, and finally obtaining output after normalization operation.
Preferably, the step S2 specifically includes: carrying out state grade classification on the work completion efficiency and the idle state of each graph judging seat, wherein the graph judging difficulty grade is the same as the number of the state grades of the graph judging seats, and the specific formula of the state grade classification is as follows:
，
wherein ,representing a rating parameter,/->Indicating the amount of task left->Representing the total task amount of all the map judgment seats,representing the time parameter required to complete n tasks, < ->Representing the efficiency of the current map judging seat to complete the task with low difficulty level,and the efficiency of completing the task with high difficulty level by the current graph judging seat is represented, and the higher the level dividing parameter value is, the lower the state level is.
Preferably, the step S2 specifically includes:
and selecting the on-line graph judging seats in the state grades with the same grade according to the grade number of the difficulty to carry out task random dispatch, namely the grade number of the difficulty of the graph judging task is the same as the grade number of the state of the graph judging seats.
The beneficial effects are that:
the technical schemes provided by the embodiment of the application have at least the following technical effects or advantages:
1. the method adopts the linear combination of Fourier basis functions to better represent the smooth background area, enhances the connectivity of object pixels and background pixels, and maintains clear graphic outline.
2. The application adopts a double-channel feature extraction method of attention mechanism and depth separable convolution, belongs to linear operation, can greatly reduce the computational complexity, is beneficial to amplifying image features, accelerates task transmission and reduces the workload of judging the map seat; the multi-scale feature fusion is adopted, so that the purpose of improving the image feature extraction precision is achieved.
3. The application selects proper graph judging seat groups for tasks with different difficulties, randomly distributes the task to be distributed in a limited range, improves the task processing efficiency, avoids the problem of task backlog caused by unreasonable task distribution, and simultaneously reduces the transmission quantity of task data, thereby having small bandwidth requirement and avoiding network congestion.
4. The technical scheme of the application can effectively solve the problems that a large number of tasks are not integrated into zero in the prior art, so that the backlog phenomenon of the tasks is caused, the work load of a graph judging seat is heavy, the error rate is easy to increase, and the system or the method is subjected to a series of effect investigation, and finally, the calculation complexity is greatly reduced, the task transmission is accelerated, the work load of the graph judging seat is reduced, and the task processing efficiency is improved.
Drawings
FIG. 1 is a block diagram of a task random dispatch intelligent management system based on a centralized judgment chart according to the application;
fig. 2 is a flowchart of a task random dispatch intelligent management method based on a centralized judgment chart.
Detailed Description
The embodiment of the application provides an intelligent task random dispatch management system and method based on a centralized graph judgment, which solve the problem that the prior art does not integrate a large number of tasks into zero, so that the backlog phenomenon of the tasks is caused, the graph judgment seat has heavy workload, and the error rate is easy to increase.
The technical scheme in the embodiment of the application aims to solve the problems, and the overall thought is as follows:
the method adopts the linear combination of Fourier basis functions to better represent the smooth background area, enhances the connectivity of object pixels and background pixels, and keeps clear graphic outline, and has better image segmentation effect; the application adopts a double-channel feature extraction method of attention mechanism and depth separable convolution, belongs to linear operation, can greatly reduce the computational complexity, is beneficial to amplifying image features, accelerates task transmission and reduces the workload of judging the map seat; the multi-scale feature fusion is adopted, so that the purpose of improving the image feature extraction precision is achieved; the application selects proper graph judging seat groups for tasks with different difficulties, randomly distributes the task to be distributed in a limited range, improves the task processing efficiency, avoids the problem of task backlog caused by unreasonable task distribution, and simultaneously reduces the transmission quantity of task data, thereby having small bandwidth requirement and avoiding network congestion.
In order to better understand the above technical solutions, the following detailed description will refer to the accompanying drawings and specific embodiments.
Referring to fig. 1, the task random dispatch intelligent management system based on the centralized judgment chart provided by the application comprises the following parts:
a task generation module 10, a task processing module 20, a task simplification module 30, a broadcasting module 40, a ranking module 50, a task dispatch module 60, a task management module 70, and a transmission module 80.
The task generating module 10 is configured to receive an image from a corresponding scanning device, generate a corresponding graph judging task, and connect the task generating module 10 with the task processing module 20 in a data transmission manner;
the task processing module 20 is configured to obtain an image to be interpreted in the graph judging task, perform graying processing on the image to be interpreted, and segment the grayed image, where the task processing module 20 is connected to the task simplification module 30 and the grading module 50 by a data transmission manner;
the task simplification module 30 is configured to extract feature information of a graph judging task according to an image contour model, where the extracting mode is two-channel distribution parallel processing, and send the extracted feature information to a graph judging seat, and the task simplification module 30 is connected with the transmission module 80 through a data transmission mode;
the broadcasting module 40 is used for broadcasting the working state of the map judging seat at regular time, and the broadcasting module 40 is connected with the grading module 50 and the task management module 70 in a data transmission mode;
the grading module 50 is configured to count the number of image blocks to be identified in each image judgment task, set up an image judgment difficulty grade, and divide the image judgment task into image judgment tasks with different difficulty grades according to the number of image blocks to be identified in the image judgment task; the working completion efficiency and the idle state of each graph judging seat are classified by the state classification, and the classification module 50 is connected with the task dispatch module 60 in a data transmission mode;
the task distributing module 60 is configured to randomly distribute the task by selecting an online graph judging seat in the state levels of the same level according to the number of difficulty levels, that is, the number of difficulty levels of the graph judging task is the same as the number of state levels of the graph judging seat, and the task distributing module 60 is connected with the task simplifying module 30 through a data transmission manner;
the task management module 70 is configured to receive a task state fed back by the seat of the judgment chart, update the task state regularly, and ensure that each task can be distributed and processed in time;
the transmission module 80 is configured to transmit the feature information of the graph judging task to the graph judging seat.
Referring to fig. 2, the task random dispatch intelligent management method based on the centralized judgment chart provided by the application comprises the following steps:
s1, acquiring an image to be interpreted in an image judging task, uniformly dividing the image to be interpreted into image blocks with equal length and width, constructing an image contour model, and carrying out image segmentation;
the image interpretation is a process of comprehensively analyzing, comparing, reasoning and judging various features on the image and finally extracting target information. The centralized image judging step is to deploy the image judging seat workstations in independent places in a centralized way, connect with the image scanning equipment through a network, perform centralized and unified management on the images through the information system, distribute the images and related information to the image judging seat workstations according to a certain rule, and comprehensively judge the images and the information through the centralized image judging step.
The scanning equipment scans the object to generate an image, and sends the generated image to the task generating modules 10 in real time, wherein each task generating module 10 is configured to receive the image from the corresponding scanning equipment and generate a corresponding graph judging task; the task management module 70 will periodically update the task status to ensure that each task is dispatched and processed in a timely manner.
Each map judging seat can broadcast the working state of the map judging seat by the broadcasting module 40 at fixed time, and receives a task dispatching request to execute the map judging task. The working state comprises information such as the online state of the current judgment chart seat, the amount of the residual task to be processed, the average task processing speed and the like. Each graph seat feeds back the task status to the task management module 70 after processing one graph task.
The task generating module 10 transmits the graph judging task to the task processing module 20, the task processing module 20 acquires an image to be interpreted in the graph judging task, the image to be interpreted is subjected to grey-scale processing, the grey-scale image is segmented, the graph judging difficulty of the current image to be interpreted is determined according to the number of segmented image blocks to be identified, and the specific steps of the segmentation of the interpreted image are as follows:
in order to rapidly distinguish the difficulty coefficient of the graph judging task, an image segmentation algorithm with high operation speed and low calculation complexity is adopted, and a closed contour is formed by the segmented image, wherein a specific segmentation formula is as follows:
uniformly dividing an image to be interpreted into R image blocks with equal length and width, and representing each image block as:
，
wherein ,representing the r-th image block,/->Representing Fourier basis-> and />Background pixels and object pixels representing the r-th image block, respectively, < >>. Performing edge matching on R image blocks to construct an image contour model, wherein the model expression is as follows:
，
wherein ,representing an image contour model +.> and />Parameters inside and outside the contour, respectively, +.> and />Respectively representing the start point and the end point of the traversing pixel point in the outline, respectively +.> and />Respectively representing the start point and the end point of the out-of-contour traversal pixel, respectively>Representation->Pixel value of pixel point at position, < >>Representing the weight coefficient, ++> and />Representing pixel center values inside and outside the contour, respectively.
S2, extracting characteristic information of the graph judging task according to the image contour model, sending the extracted characteristic information to the graph judging seat, reducing the workload of the graph judging seat, setting a graph judging difficulty level and a state level of the graph judging seat, and carrying out task random distribution according to the corresponding level.
The task simplification module 30 extracts feature information of the graph judging task according to the image contour model, sends the extracted feature information to the graph judging seat, reduces the workload of the graph judging seat, and simultaneously accelerates the transmission speed. The extraction process of the graph judging task characteristic information specifically comprises the following steps:
inputting the segmented image blocks into a feature extraction module, wherein the extraction mode is distributed parallel processing, and in a first channel, the processing of a note mechanism layer is as follows:
，
，
wherein ,representing the result of the processing of the attention mechanism layer, +.>Representing an activation function->Representing a multi-layer perceptron @, @>Representing global average pooling,/->Representing global maximization,/-pooling>Representing the weight coefficient, ++>Representing average predicted features>Representing the maximum predicted feature->Representing the longitudinal maximum difference of the predicted features, +.>Representing the lateral maximum difference of the predicted features, +.>Representing the pixel value at the j-th column of the i-th row on the image block,/and>representing the pixel value at the j-th column of the i-th row in the prediction feature,/and>，/>。
the first channel passes through the attention mechanism layer and then passes through oneConvolution layers and pooling layers of size. And performing depth separable convolution processing in the second channel, wherein the specific calculation is as follows:
，
wherein ,for the processing result of the depth separable convolution layer, < >>Representing convolution kernel +.>、/>The resolution of the predicted feature map is respectively. The second channel is subjected to a depth-separable convolution process and then to a +.>And the convolution layer and the pooling layer with the sizes are used for carrying out multi-scale feature fusion on the output of the two channels to obtain the final image block features.
Multi-scale feature fusionFirst pass throughThe depth of the separable convolution layer is equal to or greater than that of the input convolution layer, then the convolution layer is subjected to 4 times of upsampling layer, and finally the output is obtained after normalization operation, wherein the normalization operation formula is as follows:
，
wherein ,for normalizing the operation result, <' > for the first time>The results after the convolution processing and the upsampling processing can be separated for the above depth.
The grading module 50 counts the number of the image blocks to be identified in each image judging task, establishes image judging difficulty grades, and grades the image judging tasks with different difficulty grades according to the number of the image blocks to be identified in the image judging task. The task processing module 20 sends the difficulty coefficient of each graph judging task to the grading module 50, the grading module 50 knows the working state of each graph judging seat from the broadcasting module 40, and performs state grading on the working completion efficiency and the idle state of each graph judging seat, wherein the number of the graph judging difficulty grades is the same as that of the state grades of the graph judging seats. The specific formula of the state grade division is as follows:
，
wherein ,representing a rating parameter,/->Indicating the amount of task left->Representing allThe total task amount of the seat of the graph is judged,representing the time parameter required to complete n tasks, < ->Representing the efficiency of the current map judging seat to complete the task with low difficulty level,the efficiency of completing the task with high difficulty level by the current graph judging seat is represented. The low-difficulty and high-difficulty dividing mode is to divide all difficulty levels into high and low categories by adopting a compromise method. The greater the value of the ranking parameter, the lower the state ranking.
The task dispatch module 60 selects the on-line graph judging seats in the state grades of the same grade according to the grade number of the difficulty to dispatch the task randomly, i.e. the grade number of the difficulty of the graph judging task is the same as the grade number of the state of the graph judging seats. In one embodiment, the difficulty level 2 graph task is randomly distributed to any one of the 2-level and online graph seats.
The task distribution module 60 acquires information such as the residual task amount of each graph judging seat, the average efficiency of processing tasks with different difficulty levels, and the like in real time, and can change the distribution strategy of the graph judging task in real time according to the state level calculated by facts before distributing the graph judging task to the graph judging seats.
In summary, the task random dispatch intelligent management system and method based on the centralized judgment chart are completed.
Effect investigation:
the technical scheme of the application can effectively solve the problems that a large number of tasks are not integrated into zero in the prior art, so that the backlog phenomenon of the tasks is caused, the work load of a graph judging seat is heavy, the error rate is easy to increase, and the system or the method is subjected to a series of effect investigation, and finally, the calculation complexity is greatly reduced, the task transmission is accelerated, the work load of the graph judging seat is reduced, and the task processing efficiency is improved.
The present application is described with reference to flowchart illustrations and/or block diagrams of methods, apparatus (systems) and computer program products according to embodiments of the application. It will be understood that each flow and/or block of the flowchart illustrations and/or block diagrams, and combinations of flows and/or blocks in the flowchart illustrations and/or block diagrams, can be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general purpose computer, special purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
While preferred embodiments of the present application have been described, additional variations and modifications in those embodiments may occur to those skilled in the art once they learn of the basic inventive concepts. It is therefore intended that the following claims be interpreted as including the preferred embodiments and all such alterations and modifications as fall within the scope of the application.
It will be apparent to those skilled in the art that various modifications and variations can be made to the present application without departing from the spirit or scope of the application. Thus, it is intended that the present application also include such modifications and alterations insofar as they come within the scope of the appended claims or the equivalents thereof.

Claims (4)

1. The task random dispatch intelligent management system based on the centralized judgment chart is characterized by comprising the following parts:
the system comprises a task generating module, a task processing module, a task simplifying module, a broadcasting module, a grading module, a task distributing module, a task management module and a transmission module;
the task processing module is used for acquiring an image to be interpreted in the graph judging task, carrying out grey-scale processing on the image to be interpreted, dividing the grey-scale image, and connecting the task processing module with the task simplifying module and the grading module in a data transmission mode; the image segmentation algorithm is adopted for segmentation to form a segmented image with a closed contour, and a specific segmentation formula is as follows:
the linear combination of Fourier basis functions is adopted to better represent smooth background areas, connectivity of object pixels and background pixels is enhanced, clear graphic outline is maintained, an image to be interpreted is uniformly divided into R image blocks with equal length and width, and each image block is represented as:
，
wherein ,representing the r-th image block,/->Representing Fourier basis-> and />Background pixels and object pixels representing the r-th image block, respectively, < >>Performing edge matching on R image blocks to construct an image contour model, wherein the model expression is as follows:
，
wherein ,representing an image contour model +.> and />Parameters inside and outside the contour, respectively, +.> and />Respectively representing the start point and the end point of the traversing pixel point in the outline, respectively +.> and />Respectively representing the start point and the end point of the out-of-contour traversal pixel, respectively>Representation->Pixel value of pixel point at position, < >>Representing the weight coefficient, ++> and />Representing pixel center values inside and outside the contour, respectively;
the task simplification module is used for extracting the characteristic information of the graph judging task according to the image contour model, wherein the extracting mode is double-channel distributed parallel processing, the extracted characteristic information is sent to the graph judging seat, and the task simplification module is connected with the transmission module in a data transmission mode; in the first channel of the two-channel distributed parallel processing, the processing of the attention mechanism layer is as follows:
，
，
wherein ,representing the result of the processing of the attention mechanism layer, +.>Representing an activation function->Representing a multi-layer perceptron @, @>Representing global average pooling,/->Representing global maximization,/-pooling>Representing the weight coefficient, ++>Representing the average predicted characteristic of the image,representing the maximum predicted feature->Representing the longitudinal maximum difference of the predicted features, +.>Representing the lateral maximum difference of the predicted features, +.>Representing the pixel value at the j-th column of the i-th row on the image block,/and>representing the pixel value at the j-th column of the i-th row in the prediction feature,，/>；
the first channel passes through the attention mechanism layer and then passes through oneConvolution and pooling layers of size;
the second channel is subjected to a depth-separable convolution process and then to a processThe convolution layer and the pooling layer with the sizes are used for carrying out multi-scale feature fusion on the output of the two channels to obtain final image block features; the multi-scale feature fusion is first subjected to a +.>The depth of the separable convolution layer is equal to that of the input convolution layer, then the convolution layer is subjected to 4 times of up-sampling layer, and finally the output is obtained after normalization operation;
the grading module is used for counting the number of the image blocks to be identified in each image judging task, setting up image judging difficulty grades, and grading the image judging tasks into image judging tasks with different difficulty grades according to the number of the image blocks to be identified in the image judging tasks; the working completion efficiency and the idle state of each graph judging seat are subjected to state grading, and the grading module is connected with the task dispatch module in a data transmission mode;
the task distribution module is used for randomly distributing the task by selecting on-line graph judging seats in the state grades with the same grade according to the grade number of the difficulty, namely the grade number of the graph judging task is the same as the grade number of the state of the graph judging seats, and the task distribution module is connected with the task simplification module in a data transmission mode.
2. The task random dispatch intelligent management method based on the centralized judgment chart is characterized by comprising the following steps of:
s1, acquiring an image to be interpreted in an image judging task, uniformly dividing the image to be interpreted into image blocks with equal length and width, constructing an image contour model, and carrying out image segmentation; the image segmentation algorithm is adopted for segmentation to form a segmented image with a closed contour, and a specific segmentation formula is as follows:
the linear combination of Fourier basis functions is adopted to better represent smooth background areas, connectivity of object pixels and background pixels is enhanced, clear graphic outline is maintained, an image to be interpreted is uniformly divided into R image blocks with equal length and width, and each image block is represented as:
，
wherein ,representing the r-th image block,/->Representing Fourier basis-> and />Background pixels and object pixels representing the r-th image block, respectively, < >>Performing edge matching on R image blocks to construct an image contour model, wherein the model expression is as follows:
，
wherein ,representing an image contour model +.> and />Parameters inside and outside the contour, respectively, +.> and />Respectively representing the start point and the end point of the traversing pixel point in the outline, respectively +.> and />Respectively representing the start point and the end point of the out-of-contour traversal pixel, respectively>Representation->Pixel value of pixel point at position, < >>Representing the weight coefficient, ++> and />Representing pixel center values inside and outside the contour, respectively;
s2, extracting characteristic information of a graph judging task according to an image contour model, sending the extracted characteristic information to a graph judging seat, reducing the workload of the graph judging seat, setting a graph judging difficulty level and a state level of the graph judging seat, and carrying out task random distribution according to the corresponding level; inputting the segmented image blocks into a feature extraction module, wherein the extraction mode is double-channel distributed parallel processing; in the first channel of the two-channel distributed parallel processing, the processing of the attention mechanism layer is as follows:
，
，
wherein ,representing the result of the processing of the attention mechanism layer, +.>Representing an activation function->Representing a multi-layer perceptron @, @>Representing global average pooling,/->Representing global maximization,/-pooling>Representing the weight coefficient, ++>Representing the average predicted characteristic of the image,representing the maximum predicted feature->Representing the longitudinal maximum difference of the predicted features, +.>Representing the lateral maximum difference of the predicted features, +.>Representing the pixel value at the j-th column of the i-th row on the image block,/and>representing the pixel value at the j-th column of the i-th row in the prediction feature,，/>；
the first channel passes through the attention mechanism layer and then passes through oneConvolution and pooling layers of size;
the second channel is subjected to a depth-separable convolution process and then to a processThe convolution layer and the pooling layer with the sizes are used for carrying out multi-scale feature fusion on the output of the two channels to obtain final image block features; the multi-scale feature fusion is first subjected to a +.>The depth of the (c) separable convolution layer, then passing through a 4-time up-sampling layer, and finally obtaining output after normalization operation.
3. The method for intelligently managing task random dispatch based on centralized judgment chart as claimed in claim 2, wherein the step S2 specifically comprises:
carrying out state grade classification on the work completion efficiency and the idle state of each graph judging seat, wherein the graph judging difficulty grade is the same as the number of the state grades of the graph judging seats, and the specific formula of the state grade classification is as follows:
，
wherein ,representing a rating parameter,/->Indicating the amount of task left->Representing the total task amount of all decision diagram seats, < ->Representing the time parameter required to complete n tasks, < ->Representing the efficiency of the current map seat to complete the task with low difficulty level,/for>And the efficiency of completing the task with high difficulty level by the current graph judging seat is represented, and the higher the level dividing parameter value is, the lower the state level is.
4. The method for intelligently managing task random dispatch based on centralized judgment chart as claimed in claim 2, wherein the step S2 specifically comprises:
and selecting the on-line graph judging seats in the state grades with the same grade according to the grade number of the difficulty to carry out task random dispatch, namely the grade number of the difficulty of the graph judging task is the same as the grade number of the state of the graph judging seats.
CN202310671306.9A 2023-06-08 2023-06-08 Task random dispatch intelligent management system and method based on centralized judgment chart Active CN116468990B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202310671306.9A CN116468990B (en) 2023-06-08 2023-06-08 Task random dispatch intelligent management system and method based on centralized judgment chart

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202310671306.9A CN116468990B (en) 2023-06-08 2023-06-08 Task random dispatch intelligent management system and method based on centralized judgment chart

Publications (2)

Publication Number Publication Date
CN116468990A CN116468990A (en) 2023-07-21
CN116468990B true CN116468990B (en) 2023-09-29

Family

ID=87175688

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202310671306.9A Active CN116468990B (en) 2023-06-08 2023-06-08 Task random dispatch intelligent management system and method based on centralized judgment chart

Country Status (1)

Country Link
CN (1) CN116468990B (en)

Families Citing this family (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN117422881A (en) * 2023-11-15 2024-01-19 山东衡昊信息技术有限公司 Intelligent centralized security check graph judging system

Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110929603A (en) * 2019-11-09 2020-03-27 北京工业大学 Weather image identification method based on lightweight convolutional neural network
CN111443100A (en) * 2020-05-12 2020-07-24 北京声迅电子股份有限公司 Centralized image judging system for security inspection based on intelligent identification technology
CN112181629A (en) * 2020-11-30 2021-01-05 同方威视技术股份有限公司 Security inspection centralized image judging system and automatic working mode switching control method thereof
WO2021179634A1 (en) * 2020-09-23 2021-09-16 平安科技（深圳）有限公司 Customer service dispatching method, apparatus and device, and storage medium
CN115546245A (en) * 2022-10-26 2022-12-30 电子科技大学长三角研究院(湖州) Image foreground and background segmentation method and system based on sparse decomposition and graph Laplace regularization
WO2023082061A1 (en) * 2021-11-09 2023-05-19 贵州电网有限责任公司 Smart agent visual dispatching method based on augmented reality image processing

Patent Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110929603A (en) * 2019-11-09 2020-03-27 北京工业大学 Weather image identification method based on lightweight convolutional neural network
CN111443100A (en) * 2020-05-12 2020-07-24 北京声迅电子股份有限公司 Centralized image judging system for security inspection based on intelligent identification technology
WO2021179634A1 (en) * 2020-09-23 2021-09-16 平安科技（深圳）有限公司 Customer service dispatching method, apparatus and device, and storage medium
CN112181629A (en) * 2020-11-30 2021-01-05 同方威视技术股份有限公司 Security inspection centralized image judging system and automatic working mode switching control method thereof
WO2023082061A1 (en) * 2021-11-09 2023-05-19 贵州电网有限责任公司 Smart agent visual dispatching method based on augmented reality image processing
CN115546245A (en) * 2022-10-26 2022-12-30 电子科技大学长三角研究院(湖州) Image foreground and background segmentation method and system based on sparse decomposition and graph Laplace regularization

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
基于轻量级分组注意力模块的图像分类算法;张盼盼;李其申;杨词慧;;计算机应用(03);全文 *

Also Published As

Publication number Publication date
CN116468990A (en) 2023-07-21

Similar Documents

Publication Publication Date Title
CN111431986B (en) Industrial intelligent quality inspection system based on 5G and AI cloud edge cooperation
CN116468990B (en) Task random dispatch intelligent management system and method based on centralized judgment chart
CN112488322A (en) Federal learning model training method based on data feature perception aggregation
CN102044073B (en) Method and system for judging crowd density in image
CN113408087B (en) Substation inspection method based on cloud side system and video intelligent analysis
CN107492103A (en) Gray threshold acquisition methods, image partition method based on APSO algorithm
CN109886937B (en) Insulator defect detection method based on super-pixel segmentation image recognition
CN104992147A (en) License plate identification method of deep learning based on fast and slow combination cloud calculation environment
CN106844636A (en) A kind of unstructured data processing method based on deep learning
CN116485475A (en) Internet of things advertisement system, method and device based on edge calculation
CN109995860A (en) Deep learning task allocation algorithms based on edge calculations in a kind of VANET
CN112419397A (en) Ore particle size grading method and system based on image and deep neural network
CN116166444B (en) Collaborative reasoning method oriented to deep learning hierarchical model
CN117456167A (en) Target detection algorithm based on improved YOLOv8s
CN110213741B (en) Method for detecting authenticity of vehicle sending information in real time based on width learning
CN116524432A (en) Application of small target detection algorithm in traffic monitoring
CN113269339B (en) Method and system for automatically creating and distributing network appointment tasks
CN113793327A (en) High-speed rail foreign matter detection method based on token
CN114048978A (en) Supply and demand scheduling strategy fusion application based on machine learning model
CN117350521B (en) Operation and maintenance duty management system and method based on big data analysis
CN111368732B (en) Method and device for detecting lane lines
CN112200139B (en) User image identification method based on variable-order fractional multilayer convolutional neural network
CN115588090B (en) Aircraft point cloud semantic extraction method for decoupling spatial relationship and characteristic information
CN116935394B (en) Train carriage number positioning method based on PSENT region segmentation
Ya et al. Multi-Scale PointPillars 3D Object Detection Network

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
GR01 Patent grant
GR01 Patent grant
CB03 Change of inventor or designer information

Inventor after: Shao Yaqi

Inventor after: Liu Jun

Inventor after: Zhao Xinyu

Inventor after: Jin Jin

Inventor after: Ruan Bo

Inventor after: Zuo Haibo

Inventor before: Shao Yaqi

Inventor before: Liu Jun

Inventor before: Zhao Xinyu

Inventor before: Jin Jin

Inventor before: Ruan Bo

Inventor before: Zuo Haibo

CB03 Change of inventor or designer information