CN115564320B - Multi-intelligent-algorithm-oriented scheduling management method, device and medium - Google Patents

Multi-intelligent-algorithm-oriented scheduling management method, device and medium Download PDF

Info

Publication number
CN115564320B
CN115564320B CN202211553465.0A CN202211553465A CN115564320B CN 115564320 B CN115564320 B CN 115564320B CN 202211553465 A CN202211553465 A CN 202211553465A CN 115564320 B CN115564320 B CN 115564320B
Authority
CN
China
Prior art keywords
contraband
algorithm
intelligent
image recognition
algorithms
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
CN202211553465.0A
Other languages
Chinese (zh)
Other versions
CN115564320A (en
Inventor
姜旭
邓军
拜正斌
刘杰
陈晓涛
黄锐
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.)
Chengdu Zhiyuanhui Information Technology Co Ltd
Original Assignee
Chengdu Zhiyuanhui Information Technology Co Ltd
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 Chengdu Zhiyuanhui Information Technology Co Ltd filed Critical Chengdu Zhiyuanhui Information Technology Co Ltd
Priority to CN202211553465.0A priority Critical patent/CN115564320B/en
Publication of CN115564320A publication Critical patent/CN115564320A/en
Application granted granted Critical
Publication of CN115564320B publication Critical patent/CN115564320B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • 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
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01V—GEOPHYSICS; GRAVITATIONAL MEASUREMENTS; DETECTING MASSES OR OBJECTS; TAGS
    • G01V5/00—Prospecting or detecting by the use of ionising radiation, e.g. of natural or induced radioactivity
    • G01V5/20—Detecting prohibited goods, e.g. weapons, explosives, hazardous substances, contraband or smuggled objects
    • G01V5/22—Active interrogation, i.e. by irradiating objects or goods using external radiation sources, e.g. using gamma rays or cosmic rays
    • 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
    • G06Q10/06314—Calendaring for a resource
    • 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
    • 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
    • G06V10/764—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using classification, e.g. of video objects
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/30—Subject of image; Context of image processing
    • G06T2207/30242—Counting objects in image

Landscapes

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

Abstract

The invention provides a scheduling management method, a device and a medium for a multi-intelligent algorithm, which relate to the technical field of security inspection and comprise the following steps: s1: taking a time period from the current time to a certain time in a certain scene as a current sampling time period, and calling all intelligent image recognition algorithms to detect contraband so as to obtain a detection result; s2: estimating to obtain the true estimated quantity of the contraband in each category; s3: carrying out different calculation and comparison on different intelligent image recognition algorithms, and selecting the most suitable intelligent image recognition algorithm to be marked as an intelligent image recognition algorithm X; s4: and taking the time period when the current sampling time period is over and the next sampling time period does not reach as the current application time period, calling an intelligent graph recognition algorithm X to detect contraband, and repeating the steps. The invention reasonably realizes the scheduling and management of various intelligent image recognition algorithms and solves the problem of how to select the most suitable algorithm from the various intelligent image recognition algorithms under the brand-new scene without knowing the proportion of contraband.

Description

Multi-intelligent-algorithm-oriented scheduling management method, device and medium
Technical Field
The invention relates to the technical field of security inspection, in particular to a scheduling management method, a device and a medium for a multi-intelligent algorithm.
Background
When the X-ray machine detects the contraband, an intelligent image recognizing algorithm can be adopted to detect the contraband. Because the types of contraband articles are various, dozens of types or even hundreds of types, and the detection rate and the false detection rate of different types of contraband articles are different by different intelligent image recognition algorithms.
If in a specific scene, when the X-ray machine only adopts one intelligent attempt algorithm to detect the contraband, the detection result may be that the quantity of the drinking cup liquid contraband is greater than that of the pistol contraband, and when another intelligent attempt algorithm is adopted in the same specific scene, the detection result may be that the quantity of the pistol contraband is greater than that of the drinking cup liquid contraband.
In addition, in the contraband detection of the actual scene, the actual values of the composition and the quantity of the contraband are unknown, and due to the consumption of the algorithm on hardware resources, in the formal operation, the X-ray machine is difficult to simultaneously call a plurality of algorithms to identify the contraband.
Then, the operation of detecting the contraband by the X-ray machine may need to select different intelligent image recognition algorithms according to different scenes. Therefore, it is a problem how to select a most suitable intelligent graph recognition algorithm to detect contraband in a specific scenario.
Disclosure of Invention
The invention aims to provide a scheduling management method, a device and a medium for multiple intelligent algorithms, wherein a time slot from current time to certain time in a certain scene is taken as a current sampling time slot, all intelligent graph recognition algorithms are called in the current sampling time slot to detect contraband, an intelligent graph recognition algorithm which is most suitable for the scene is selected according to calculation and comparison, then a time slot which is finished by the current sampling time slot and is not obtained by the next sampling time slot is taken as a current application time slot, the most suitable intelligent graph recognition algorithm in the scene is called in the current application time slot to detect the contraband, the current sampling time slot and the current application time slot are taken as a cycle time slot, the process of detecting the contraband is carried out according to the cycle, the scheduling and management of the multiple intelligent graph algorithms are reasonably realized by calling all the intelligent graph algorithms in the sampling time slot and selecting the most suitable intelligent graph recognition algorithm from the most suitable intelligent graph recognition algorithms in the current application time slot, and how to select the most suitable intelligent graph recognition algorithm from multiple brand-new intelligent graph recognition algorithms in the scene of which the contraband occupation ratio is not known reasonably.
In order to solve the technical problem, the invention adopts the following scheme:
a scheduling management method for a multi-intelligent algorithm specifically comprises the following steps:
s1: taking a time period from the current time to a certain time in a certain scene as a current sampling time period, and calling all intelligent image recognition algorithms to detect contraband in the current sampling time period to obtain the detection result of each type of contraband under different intelligent image recognition algorithms;
s2: estimating to obtain the real estimated quantity of each type of contraband according to the detection result of each type of contraband under different intelligent identification map algorithms;
s3: carrying out proportion selection calculation or proportion estimation calculation on different intelligent image recognition algorithms according to the estimated real estimated quantity of each type of contraband, comparing calculation results, and selecting the most suitable intelligent image recognition algorithm from all intelligent image recognition algorithms to be marked as an intelligent image recognition algorithm X;
s4: and taking the time period in which the current sampling time period is finished and the next sampling time period does not yet arrive as the current application time period, calling an intelligent graph recognition algorithm X in the current application time period to detect contraband, and repeating the steps.
Further, the detection results are a detection rate, a false detection rate and the number of contraband articles detected by the algorithm,
the number of contraband detected by the algorithm includes the number of real contraband and the number of non-contraband,
the detection rate refers to the percentage of the number of genuine contraband detected by the algorithm to the actual number of contraband,
the false positive rate refers to the percentage of the number of non-contraband items detected by the algorithm to the number of contraband items detected by the algorithm,
the actual amount of contraband comprises an algorithm detected amount of contraband.
Further, the step S2 specifically includes the following steps:
s21: obtaining the actual quantity of each type of contraband under different intelligent identification map algorithms by using an actual estimation formula according to the detection result of each type of contraband under different intelligent identification map algorithms;
s22: and averaging the actual quantity of the same type of contraband under different intelligent image recognition algorithms, and taking the average value as the real estimated quantity of the type of contraband.
Further, the actual estimation formula is used for calculating the actual amount of the contraband of the single kind, and the actual estimation formula is as follows:
the actual number of contraband articles detection rate + the number of contraband articles detected by the algorithm detection rate = the number of contraband articles detected by the algorithm.
Further, the selection and calculation of the ratio in S3 specifically includes:
respectively calculating the percentage of the real estimated quantity of the single type of contraband to the real estimated quantity of all types of contraband according to the estimated real estimated quantity of each type of contraband to obtain the proportion of each type of contraband in the scene,
comparing the multiple proportions, selecting the contraband type with the largest proportion,
and comparing the detection rates of all intelligent image recognition algorithms corresponding to the types of the contraband with the largest occupation rate, and selecting the intelligent image recognition algorithm with the highest detection rate as the most suitable intelligent image recognition algorithm under the scene, and recording as the intelligent image recognition algorithm X.
Further, the calculation of the ratio estimation in S3 specifically includes:
respectively calculating the score of each intelligent image recognition algorithm by using an algorithm score formula according to the estimated real estimated quantity of each type of contraband,
and selecting the intelligent image recognition algorithm with the highest score as the most suitable intelligent image recognition algorithm in the scene, and recording as an intelligent image recognition algorithm X.
Further, the algorithm score formula is used for calculating the score of the single intelligent graph recognition algorithm, and the algorithm score formula is as follows:
the score of the intelligent recognition graph algorithm = the estimated real estimated quantity of the prohibited articles No. 1 (detection rate + 1-false detection rate) + the estimated real estimated quantity of the prohibited articles No. 2 (detection rate + 1-false detection rate) + …) + the estimated real estimated quantity of the prohibited articles No. n (detection rate + 1-false detection rate),
the contraband No. 1 and the contraband No. 2 … n are represented as different types of contraband.
A scheduling management device facing a multi-intelligent algorithm comprises:
one or more processors;
a storage unit, configured to store one or more programs, which when executed by the one or more processors, enable the one or more processors to implement the method for scheduling management for multiple intelligent algorithms.
A computer-readable storage medium having stored thereon a computer program,
the computer program can realize the scheduling management method facing the multiple intelligent algorithms when being executed by a processor.
The invention has the beneficial effects that:
the invention provides a multi-intelligent-algorithm-oriented scheduling management method, a device and a medium, which can call all intelligent image recognition algorithms to detect contraband, then obtain the corresponding real estimated quantity of each type of contraband for each intelligent image recognition algorithm by using an actual estimation formula, calculate and compare different intelligent image recognition algorithms according to the real estimated quantity of each type of contraband estimated by the intelligent image recognition algorithms, and select a means which is most suitable for the intelligent image recognition algorithms in the scene.
Because the types of the contraband in the actual scene are more and the occupation ratio is unknown, different intelligent image recognizing algorithms are different for different detection rates and false detection rates of the contraband. Therefore, for the prior art, the detection result obtained by adopting the same intelligent graph recognition algorithm in an actual scene is inaccurate, but due to the consumption of hardware resources by the algorithm, a plurality of algorithms are difficult to be used for recognition in the actual scene, and a method for scheduling and managing the intelligent graph recognition algorithm is needed.
Therefore, in the technical means, the time slot from the current time of the scene to a certain time is taken as the current sampling time slot, all intelligent image recognition algorithms are called to detect the contraband, the detection result of each type of contraband under different intelligent image recognition algorithms is obtained, the true estimated quantity of each type of contraband is estimated, the true estimated quantity of each type of contraband under the scene represents the true occupation ratio of the contraband under the scene, and the most appropriate intelligent image recognition algorithm can be selected according to the occupation ratio of the quantity of the contraband.
According to the proportion selection calculation or the proportion estimation calculation, the proportion of the contraband in the scene of the real estimated quantity of each kind of contraband obtained by estimation can be confirmed. And the proportion selection calculation is mainly performed through two comparison selections to select the intelligent image recognition algorithm with the highest detectable rate as the most suitable intelligent image recognition algorithm in the scene. The ratio estimation calculation is mainly to calculate the score of the intelligent image recognition algorithm based on the real estimated quantity, the detection rate and the false detection rate of each type of contraband obtained by the same algorithm, and the intelligent image recognition algorithm with the highest score is used as the most suitable intelligent image recognition algorithm in the scene.
And then taking a time period when the current sampling time period is finished and the next sampling time period does not yet come as an application time period, and calling the most suitable intelligent image recognition algorithm under the scene in the application time period to detect the contraband. The sampling time period and the application time period are a period, and all the intelligent image recognition algorithms are calculated and selected in the sampling time period, so that the intelligent image recognition algorithm called in the application time period is most suitable.
Drawings
FIG. 1 is a schematic flow chart of the present invention.
Fig. 2 is a time period diagram of the present invention.
Fig. 3 is a schematic diagram of detection results of the intelligent map recognition algorithm a and the intelligent map recognition algorithm B.
Fig. 4 is a schematic diagram of detection results of an intelligent map recognition algorithm C and an intelligent map recognition algorithm D of the present invention.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. The following description of at least one exemplary embodiment is merely illustrative in nature and is in no way intended to limit the invention, its application, or uses. All other embodiments, which can be obtained by a person skilled in the art without inventive step based on the embodiments of the present invention, are within the scope of protection of the present invention.
The relative arrangement of the components and steps, the numerical expressions and numerical values set forth in these embodiments do not limit the scope of the present invention unless it is specifically stated otherwise.
Meanwhile, it should be understood that the sizes of the respective portions shown in the drawings are not drawn in an actual proportional relationship for the convenience of description.
In addition, descriptions of well-known structures, functions, and configurations may be omitted for clarity and conciseness. Those of ordinary skill in the art will recognize that various changes and modifications of the examples described herein can be made without departing from the spirit and scope of the disclosure.
Techniques, methods, and apparatus known to those of ordinary skill in the relevant art may not be discussed in detail but are intended to be part of the specification where appropriate.
In all examples shown and discussed herein, any particular value should be construed as exemplary only and not as limiting. Thus, other examples of the exemplary embodiments may have different values.
The invention is explained in detail below with reference to the figures and with reference to embodiments:
example 1
As shown in fig. 1 and fig. 2, a scheduling management method facing multiple intelligent algorithms is applied to a specific scene, which may be a brand-new scene without knowing the proportion of contraband, the scheduling management method periodically performs contraband detection on packages by an X-ray machine, one periodic time period is a sampling time period and an application time period, all intelligent recognition graph algorithms are called in the sampling time period to perform contraband detection, detection results of each type of contraband under different intelligent recognition graph algorithms are obtained, and a true estimated quantity of each type of contraband is estimated, since the estimated true estimated quantity of each type of contraband under the scene represents the true proportion of the contraband under the scene, and the most suitable intelligent recognition graph algorithm can be selected according to the proportion of the quantity of the contraband.
And calling the most suitable intelligent graph recognition algorithm in the application time period to detect the contraband. By calculating and selecting all the intelligent image recognition algorithms in the sampling time period, the intelligent image recognition algorithm called in the application time period can be the most suitable.
The scheduling management method specifically comprises the following steps:
s1: taking a time period from the current time of the scene to a certain time as a current sampling time period, and calling all intelligent image recognition algorithms to detect contraband in the current sampling time period to obtain the detection result of each type of contraband under different intelligent image recognition algorithms;
s2: estimating and obtaining the real estimated quantity of each type of contraband according to the detection result of each type of contraband under different intelligent identification map algorithms,
s21: obtaining the actual quantity of each type of contraband under different intelligent identification map algorithms by using an actual estimation formula according to the detection result of each type of contraband under different intelligent identification map algorithms;
s22: averaging the actual quantity of the same type of contraband under different intelligent image recognition algorithms, and taking the average value as the real estimated quantity of the type of contraband;
s3: carrying out proportion selection calculation or proportion estimation calculation on different intelligent image recognition algorithms according to the estimated real estimated quantity of each type of contraband, comparing calculation results, and selecting the most suitable intelligent image recognition algorithm from all intelligent image recognition algorithms to be marked as an intelligent image recognition algorithm X;
s4: and taking the time period in which the current sampling time period is finished and the next sampling time period does not yet arrive as the current application time period, calling an intelligent graph recognition algorithm X in the current application time period to detect contraband, and repeating the steps.
Preferably, the detection results are a detection rate, a false detection rate and the number of contraband articles detected by the algorithm,
the number of contraband detected by the algorithm includes the number of real contraband and the number of non-contraband,
the detection rate refers to the percentage of the number of real contraband detected by the algorithm to the actual number of contraband,
the false detection rate refers to the percentage of the number of non-contraband detected by the algorithm to the number of contraband detected by the algorithm,
the actual amount of contraband comprises an algorithm detected amount of contraband.
Preferably, the actual estimation formula is used for calculating the actual amount of the contraband of a single kind, and the actual estimation formula is:
the actual number of contraband articles detection rate + the number of contraband articles detected by the algorithm detection rate = the number of contraband articles detected by the algorithm.
For example: in a specific scenario, the actual number of contraband is 100, and the smart graph algorithm detects 25 total contraband articles, wherein 10 of the detected 25 articles are real contraband articles, and the other 15 articles are detection errors. Then 100 × 10% + 25 × 60% = 25.
The actual estimation formula has 4 total parameters, and the 4 th parameter can be calculated as long as any 3 parameters are known.
However, since the real detection rate and the false detection rate fluctuate in each actual detection, only the detection rate and the false detection rate inherent in the algorithm can be used, and the real estimated quantity of each type of contraband can be roughly estimated through the formula by the maximum likelihood estimation method.
Further, the ratio selection calculation in S3 specifically includes:
respectively calculating the percentage of the real estimated quantity of the single type of contraband to the real estimated quantity of all types of contraband according to the estimated real estimated quantity of each type of contraband to obtain the proportion of each type of contraband in the scene,
comparing the multiple proportions, selecting the contraband type with the largest proportion,
and comparing the detection rates of all the intelligent image recognition algorithms corresponding to the types of the contraband with the largest occupation rate, and selecting the intelligent image recognition algorithm with the highest detection rate as the most suitable intelligent image recognition algorithm under the scene, and marking as an intelligent image recognition algorithm X.
Further, the calculation of the ratio estimation in S3 specifically includes:
respectively calculating the score of each intelligent image recognition algorithm by using an algorithm score formula according to the estimated real estimated quantity of each type of contraband,
and selecting the intelligent image recognition algorithm with the highest score as the most suitable intelligent image recognition algorithm in the scene, and recording as an intelligent image recognition algorithm X.
The algorithm score formula is used for calculating the score of the single intelligent graph recognition algorithm, and the algorithm score formula is as follows:
the score of the intelligent recognition algorithm = number 1 estimated real quantity of contraband (detection rate + 1-false detection rate) obtained by estimation, number 2 estimated real quantity of contraband (detection rate + 1-false detection rate) + … + estimated number n estimated quantity of contraband (detection rate + 1-false detection rate),
the contraband No. 1 and the contraband No. 2 … n are represented as different types of contraband.
In the duty ratio selection calculation process, the detection rates of all intelligent recognition graph algorithms corresponding to the types of the contraband with the largest duty ratios are compared, and the intelligent recognition graph algorithm with the highest detection rate is selected as the most suitable intelligent recognition graph algorithm in the scene.
The algorithm score formula of the proportion estimation calculation can also be seen visually, in this scene, the larger the quantity of a certain type of contraband is, the higher the detection rate of the certain algorithm to the type of contraband is, and the lower the false detection rate is, the higher the algorithm score is, so that the intelligent graph recognition algorithm with the highest score is selected as the most suitable intelligent graph recognition algorithm in the scene. The larger the number of the types of contraband in the scene, the largest the proportion of the types of contraband in the scene.
It can be seen that the processes of the ratio estimation calculation and the ratio selection calculation are different, but the ratio estimation calculation and the ratio selection calculation are all selected according to the highest detection rate of the types of contraband with the largest ratio in a certain intelligent recognition graph algorithm under the scene.
As shown in fig. 3, if a specific scenario is detected, the intelligent recognition graph algorithm a and the intelligent recognition graph algorithm B are used to detect contraband in the scenario.
The detection rate of the intelligent image recognition algorithm A to the pistols is 50%, the false detection rate of the intelligent image recognition algorithm A is 50%, the number of the pistols detected by the algorithm is 100, the detection rate of the intelligent image recognition algorithm A to the nunchakus is 20%, the false detection rate of the intelligent image recognition algorithm A is 40%, and the number of the nunchakus detected by the algorithm is 10. The detection rate of the intelligent recognition algorithm B to the pistols is 10%, the false detection rate is 10%, the number of the pistols detected by the algorithm is 50, the detection rate to the nunchakus is 0%, the false detection rate is 0%, and the number of the nunchakus detected by the algorithm is 0.
For a pistol, the actual quantity of contraband articles of the pistol can be estimated to be 100 in the intelligent image recognition algorithm A according to the detection result of the intelligent image recognition algorithm A by using an actual estimation formula, and the specific formula is as follows: the actual number of contraband +100 + 50% =100.
In the intelligent map recognizing algorithm B, the actual number of the pistol contraband articles obtained by estimation is 450 according to the detection result of the intelligent map recognizing algorithm B by using an actual estimation formula, and the specific formula is as follows: the actual number of contraband 10% +50 10% =50.
Therefore, under the same specific scene, the actual quantity of the contraband with larger difference can be obtained according to different intelligent image recognition algorithms, so that the most suitable intelligent image recognition algorithm needs to be selected according to different scenes to detect the contraband, and the obtained actual quantity of the contraband is more accurate.
Here, in order to make the estimated actual estimated number of pistol contraband articles closer to the actual value, an arithmetic mean is used as an approximate actual value, so in this specific scenario, the estimated actual estimated number of pistol contraband articles is 275, and this value is closer to the actual value, namely, (100 + 450)/2 =275.
As described above, the actual number of the nunchakus estimated by the intelligent image recognition algorithm a is 30, and the actual number of the nunchakus estimated by the intelligent image recognition algorithm B is 0. Therefore, in this specific scenario, the estimated real estimated number of the nunchakus is 20, i.e., (30 + 0)/2 =15.
When the proportion selection calculation is adopted, when the percentage of the real estimated quantity of the contraband of a single type to the real estimated quantity of the contraband of all types needs to be calculated respectively, the real proportion of the gun contraband in the scene is about 95%, and the real proportion of the nunchakus is about 5%.
And comparing the proportion of the gun contraband with the proportion of the nunchakus contraband, and selecting a larger proportion of the gun contraband. And comparing the detection rates of the intelligent recognition graph algorithm A and the intelligent recognition graph algorithm B corresponding to the gun contraband, wherein the detection rate of the intelligent recognition graph algorithm A on the gun contraband is 50%, the detection rate of the intelligent recognition graph algorithm B on the gun contraband is 10%, and the intelligent recognition graph algorithm A with the highest detection rate is selected as the most suitable intelligent recognition graph algorithm under the scene.
In addition, when the proportion estimation calculation is adopted, the score of each intelligent graph recognition algorithm needs to be calculated by using an algorithm score formula according to the estimated real estimated quantity of each type of contraband, and the intelligent graph recognition algorithm corresponding to the highest algorithm score is selected as the most suitable intelligent graph recognition algorithm in the scene, specifically:
the score of smart graph algorithm a is 290, i.e. 275 x (50% + 1-50%) +15 x (20% + 1-20%) =290,
the score of the smart mapping algorithm B is 240, i.e. 275 × (20% + 1-40%) +20 × (0% + 1-0%) =240,
after the scores of the intelligent map recognition algorithm A and the intelligent map recognition algorithm B are compared, the score of the intelligent map recognition algorithm A is higher than that of the intelligent map recognition algorithm B, so that the intelligent map recognition algorithm A is selected as the most appropriate algorithm in the specific scene.
In embodiment 1, according to two different calculation processes, it can be obtained that the intelligent recognition graph algorithm a is the most suitable algorithm in the specific scene, and it is proved that the most suitable intelligent recognition graph algorithm in the specific scene can be actually obtained in the two different calculation processes.
Example 2
If the detection rate of the intelligent recognition graph algorithm C on the pistol is 100%, the false detection rate is 0%, the detection rate of the intelligent recognition graph algorithm C on the nunchakus is 0%, the false detection rate is 100%, the detection rate of the intelligent recognition graph algorithm D on the pistol is 0%, the false detection rate is 100%, the detection rate of the nunchakus is 100%, and the false detection rate is 0%.
Then, under the scene that the actual number of the pistol contraband is 175 and the actual number of the nunchakus contraband is 10, it can be visually seen that the intelligent image recognizing algorithm C is the most suitable intelligent image recognizing algorithm under the scene.
However, in an actual scene, the actual values of the types and the quantities of the contraband are unknown, so that a monitoring value needs to be obtained through an intelligent graph recognition algorithm, the monitoring value is different from the actual value in size, but the fluctuation of the monitoring value changes along with the fluctuation of the actual value.
The monitoring value can be a real estimated quantity of the contraband of each category obtained through algorithm estimation, and the distribution ratio of the real estimated quantity represents the real ratio of the contraband categories in the scene.
As shown in fig. 3 and 4, if a specific scene is detected, the intelligent recognition algorithm a, the intelligent recognition algorithm B, the intelligent recognition algorithm C, and the intelligent recognition algorithm D are used to detect contraband in the scene.
Wherein, the intelligent map-identifying algorithm a and the intelligent map-identifying algorithm B are the intelligent map-identifying algorithm a and the intelligent map-identifying algorithm B described in the above embodiment 2, the detection rate of the intelligent map-identifying algorithm C to the pistol is 100%, the false detection rate is 0%, the number of pistols detected by the algorithm is 150, the detection rate to the nunchakus is 0%, the false detection rate is 100% and the number of nunchakus detected by the algorithm is 0, the detection rate of the intelligent map-identifying algorithm D to the pistol is 0%, the false detection rate is 100% and the number of pistols detected by the algorithm is 0, and the detection rate to the nunchakus is 100%, the false detection rate is 0% and the number of nunchakus detected by the algorithm is 10.
For the pistol, the actual number of pistol contraband is estimated to be 150 in the smart map algorithm C and 0 in the smart map algorithm D.
For the nunchakus, the actual number of the pistol contraband is estimated to be 0 in the intelligent recognition algorithm C, and 10 in the intelligent recognition algorithm D.
Therefore, by using the intelligent recognition algorithm a and the intelligent recognition algorithm B in example 1 to estimate the actual quantities of the pistol contraband and the nunchakus contraband, the true estimated quantity of the pistol contraband is 175, and the true estimated quantity of the nunchakus contraband is 10.
Then, in the case that the estimated actual number of the pistol contraband is 175 and the estimated actual number of the nunchakus contraband is 10, it can be visually seen that the intelligent recognition graph algorithm C is the most suitable intelligent recognition graph algorithm in the scene.
In order to confirm the correctness of the idea, the scores of the intelligent image recognition algorithm can be respectively calculated according to the proportion estimation calculation,
the score of the intelligent recognition algorithm a is 185, i.e. 175 x (50% + 1-50%) +10 x (20% + 1-20%) =185,
the score of the smart graph algorithm B is 150, i.e. 175 x (20% + 1-40%) +10 x (0% + 1-0%) =150,
the score of the smart graph algorithm C is 350, i.e. 175 x (100% + 1-0%) +10 x (0% + 1-100%) =350,
the score of the smart graph algorithm D is 20, i.e. 175 x (0% + 1-100%) +10 x (100% + 1-0%) =20,
after the scores of the intelligent image recognition algorithms are compared, the score of the intelligent image recognition algorithm C is higher than the scores of the other three intelligent image recognition algorithms, so that the intelligent image recognition algorithm C is selected as the most appropriate algorithm in the specific scene.
In embodiment 2, in an actual situation, it can be visually seen that the intelligent map recognizing algorithm C is the most suitable intelligent map recognizing algorithm in the scene. However, the intelligent graph recognition algorithm C can be obtained according to the proportion selection calculation or the proportion estimation calculation, and is the most suitable intelligent graph recognition algorithm in the scene, and the proportion selection calculation or the proportion estimation calculation is proved to be the most suitable intelligent graph recognition algorithm selected from all the intelligent graph recognition algorithms, so that the scheduling and the management of the intelligent graph recognition algorithm are realized.
Example 3
A scheduling management device facing a multi-intelligent algorithm comprises:
one or more processors;
the storage unit is used for storing one or more programs, and when the one or more programs are executed by the one or more processors, the one or more processors can realize the scheduling management method facing the multi-intelligent algorithm.
A computer-readable storage medium having stored thereon a computer program,
the computer program can realize the scheduling management method facing the multiple intelligent algorithms when being executed by a processor.
The foregoing is only a preferred embodiment of the present invention, and the present invention is not limited thereto in any way, and any simple modification, equivalent replacement and improvement made to the above embodiment within the spirit and principle of the present invention still fall within the protection scope of the present invention.

Claims (9)

1. A scheduling management method for a multi-intelligent algorithm is characterized by comprising the following steps:
s1: taking a time period from the current time to a certain time in a certain scene as a current sampling time period, and calling all intelligent image recognition algorithms to detect contraband in the current sampling time period to obtain the detection result of each type of contraband under different intelligent image recognition algorithms;
s2: estimating the real estimated quantity of each type of contraband according to the detection result of each type of contraband under different intelligent identification map algorithms;
s3: carrying out proportion selection calculation or proportion estimation calculation on different intelligent image recognition algorithms according to the estimated real estimated quantity of each type of contraband, comparing calculation results, and selecting the most suitable intelligent image recognition algorithm from all intelligent image recognition algorithms to be marked as an intelligent image recognition algorithm X;
s4: and taking the time period in which the current sampling time period is finished and the next sampling time period does not yet arrive as the current application time period, calling an intelligent graph recognition algorithm X in the current application time period to detect contraband, and repeating the steps.
2. The scheduling management method for multiple intelligent algorithms according to claim 1, wherein the detection results are a detection rate, a false detection rate and the number of contraband detected by the algorithm,
the number of contraband detected by the algorithm includes the number of real contraband and the number of non-contraband,
the detection rate refers to the percentage of the number of genuine contraband detected by the algorithm to the actual number of contraband,
the false detection rate refers to the percentage of the number of non-contraband detected by the algorithm to the number of contraband detected by the algorithm,
the actual amount of contraband comprises an algorithm detected amount of contraband.
3. The scheduling management method for multiple intelligent algorithms according to claim 2, wherein the S2 specifically includes the following steps:
s21: obtaining the actual quantity of each type of contraband under different intelligent identification map algorithms by using an actual estimation formula according to the detection result of each type of contraband under different intelligent identification map algorithms;
s22: and averaging the actual quantity of the same type of contraband under different intelligent image recognition algorithms, and taking the average value as the real estimated quantity of the type of contraband.
4. The multi-intelligence algorithm-oriented scheduling management method according to claim 3, wherein the actual estimation formula is used for calculating the actual amount of the single type of contraband, and the actual estimation formula is as follows:
the actual number of contraband articles, the detection rate + the number of contraband articles detected by the algorithm, the false detection rate = the number of contraband articles detected by the algorithm.
5. The scheduling management method for multiple intelligent algorithms according to claim 2, wherein the selection and calculation of the ratio in S3 specifically comprises:
respectively calculating the percentage of the real estimated quantity of the single type of contraband to the real estimated quantity of all types of contraband according to the estimated real estimated quantity of each type of contraband to obtain the proportion of each type of contraband in the scene,
comparing the multiple proportions, selecting the contraband type with the largest proportion,
and comparing the detection rates of all intelligent image recognition algorithms corresponding to the types of the contraband with the largest occupation rate, and selecting the intelligent image recognition algorithm with the highest detection rate as the most suitable intelligent image recognition algorithm under the scene, and recording as the intelligent image recognition algorithm X.
6. The scheduling management method oriented to multiple intelligent algorithms according to claim 2, wherein the S3 duty ratio estimation calculation specifically comprises:
respectively calculating the score of each intelligent image recognition algorithm by using an algorithm score formula according to the estimated real estimated quantity of each type of contraband,
and selecting the intelligent image recognition algorithm with the highest score as the most suitable intelligent image recognition algorithm in the scene, and marking as an intelligent image recognition algorithm X.
7. The scheduling management method for multiple intelligent algorithms according to claim 6, wherein the algorithm score formula is used for calculating the score of a single intelligent recognition algorithm, and the algorithm score formula is as follows:
the score of the intelligent recognition algorithm = number 1 estimated real quantity of contraband (detection rate + 1-false detection rate) obtained by estimation, number 2 estimated real quantity of contraband (detection rate + 1-false detection rate) + … + estimated number n estimated quantity of contraband (detection rate + 1-false detection rate),
the contraband No. 1 and the contraband No. 2 … n are represented as different types of contraband.
8. A scheduling management device for multiple intelligent algorithms is characterized by comprising the following components:
one or more processors;
a storage unit configured to store one or more programs which, when executed by the one or more processors, enable the one or more processors to implement a method for scheduling management for multiple intelligent algorithms according to any one of claims 1 to 7.
9. A computer-readable storage medium having stored thereon a computer program, characterized in that,
the computer program, when executed by a processor, is capable of implementing a method for scheduling management for a multiple intelligence algorithm according to any of claims 1 to 7.
CN202211553465.0A 2022-12-06 2022-12-06 Multi-intelligent-algorithm-oriented scheduling management method, device and medium Active CN115564320B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202211553465.0A CN115564320B (en) 2022-12-06 2022-12-06 Multi-intelligent-algorithm-oriented scheduling management method, device and medium

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202211553465.0A CN115564320B (en) 2022-12-06 2022-12-06 Multi-intelligent-algorithm-oriented scheduling management method, device and medium

Publications (2)

Publication Number Publication Date
CN115564320A CN115564320A (en) 2023-01-03
CN115564320B true CN115564320B (en) 2023-04-07

Family

ID=84770845

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202211553465.0A Active CN115564320B (en) 2022-12-06 2022-12-06 Multi-intelligent-algorithm-oriented scheduling management method, device and medium

Country Status (1)

Country Link
CN (1) CN115564320B (en)

Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109799544A (en) * 2018-12-28 2019-05-24 深圳市华讯方舟太赫兹科技有限公司 Intelligent detecting method, device and storage device applied to millimeter wave safety check instrument
CN110020647A (en) * 2018-01-09 2019-07-16 杭州海康威视数字技术股份有限公司 A kind of contraband object detection method, device and computer equipment
CN114419542A (en) * 2021-12-28 2022-04-29 北京无线电计量测试研究所 Millimeter wave image detection optimal model selection method and device based on deep learning
CN114724002A (en) * 2022-02-25 2022-07-08 广东工业大学 Method for rapidly detecting contraband
CN114862837A (en) * 2022-06-02 2022-08-05 西京学院 Human body security check image detection method and system based on improved YOLOv5s
CN115019114A (en) * 2021-12-07 2022-09-06 昆明理工大学 X-ray contraband rotating frame detection method and system
CN115205547A (en) * 2022-08-01 2022-10-18 北京远鉴信息技术有限公司 Target image detection method and device, electronic equipment and storage medium

Patent Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110020647A (en) * 2018-01-09 2019-07-16 杭州海康威视数字技术股份有限公司 A kind of contraband object detection method, device and computer equipment
CN109799544A (en) * 2018-12-28 2019-05-24 深圳市华讯方舟太赫兹科技有限公司 Intelligent detecting method, device and storage device applied to millimeter wave safety check instrument
CN115019114A (en) * 2021-12-07 2022-09-06 昆明理工大学 X-ray contraband rotating frame detection method and system
CN114419542A (en) * 2021-12-28 2022-04-29 北京无线电计量测试研究所 Millimeter wave image detection optimal model selection method and device based on deep learning
CN114724002A (en) * 2022-02-25 2022-07-08 广东工业大学 Method for rapidly detecting contraband
CN114862837A (en) * 2022-06-02 2022-08-05 西京学院 Human body security check image detection method and system based on improved YOLOv5s
CN115205547A (en) * 2022-08-01 2022-10-18 北京远鉴信息技术有限公司 Target image detection method and device, electronic equipment and storage medium

Also Published As

Publication number Publication date
CN115564320A (en) 2023-01-03

Similar Documents

Publication Publication Date Title
Colas et al. How many random seeds? statistical power analysis in deep reinforcement learning experiments
Zielinski et al. Stopping criteria for differential evolution in constrained single-objective optimization
US8583582B2 (en) Robust automated hierarchical determination for power monitoring systems
CN109649915B (en) Intelligent container cargo identification method and device
JP7346176B2 (en) Systems and methods for binned interquartile range analysis in data series anomaly detection
CN110784355B (en) Fault identification method and device
CN110490516A (en) A kind of unmanned store shelf article detection method and device
CN110992135B (en) Risk identification method and device, electronic equipment and storage medium
CN108022046A (en) A kind of electric power system data method for evaluating quality, storage medium and equipment
JP6200076B2 (en) Method and system for evaluating measurements obtained from a system
CN108629254B (en) Moving target detection method and device
CN109040084B (en) Network flow abnormity detection method, device, equipment and storage medium
CN112669290A (en) Image comparison method and device
CN115564320B (en) Multi-intelligent-algorithm-oriented scheduling management method, device and medium
CN110738415A (en) Electricity stealing user analysis method based on electricity utilization acquisition system and outlier algorithm
US8438128B2 (en) Empirical modeling of confusion matrices
CN110399926B (en) Street lamp fault diagnosis method and device
US9912564B2 (en) Methods and systems to identify bottleneck causes in applications using temporal bottleneck point detection
RU2364926C2 (en) Way of control of multiparameter object state
CN117149746A (en) Data warehouse management system based on cloud primordial and memory calculation separation
CN109462510B (en) CDN node quality evaluation method and device
CN111242546A (en) Goods picking task accounting method and device based on face recognition
CN110443035A (en) The method and apparatus that the system for invading trial for identification is calibrated
CN111353127A (en) Single variable point detection method, system, equipment and storage medium
CN109948626A (en) A kind of object detection method and device

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