Security check quality detection method and system and computer program product

Abstract

The embodiment of the invention provides a security check quality detection method, a security check quality detection system and a computer program product. The AI device obtains a security check image from the security check device, carries out contraband identification on the security check image to obtain an intelligent image discrimination result, identifies an image discrimination mark in the security check image to obtain an artificial image discrimination result, and sends the intelligent image discrimination result and the artificial image discrimination result to the server; and when the server judges that the intelligent graph judgment result is inconsistent with the manual graph judgment result, the server sends the security check image to a quality inspection graph judgment end, so that a quality inspector can re-judge the security check image. Since the artificial image discrimination result of the security inspector is identified from the security inspection image through the AI equipment, the efficiency of acquiring two image discrimination results by the server can be improved, and the quality inspection efficiency is further improved.

Description

Technical Field

[0001] The present application relates to the field of security inspection technology, and more specifically, to a security inspection quality detection method, system and computer program product. Background Art

[0002] In order to ensure the safety of passengers, security inspection equipment is used to conduct security inspections on passenger bags at airports, train stations and other places to determine whether there are prohibited items in the passenger bags. The security inspection equipment can collect images of the inspected passenger bags to obtain security inspection images, and then the security inspectors will determine the contraband on the collected security inspection images. Since security inspectors may miss inspections and make misjudgments, quality inspectors are required to monitor the security inspection quality of security inspectors to ensure the quality of security inspections. The current quality inspection method is usually that quality inspectors make a secondary judgment on the security inspectors' judgment results to verify the accuracy of the security inspectors' judgment results. However, due to the large number of security inspection images, most quality inspectors can only conduct verification by random inspections, which has low quality inspection efficiency, and the quality inspection cannot cover all security inspection images. There is still a high risk of missed inspections and misjudgments. Summary of the invention

[0003] In view of this, the present application provides a security inspection quality detection method, system and computer program product.

[0004] According to a first aspect of the present application, a security inspection quality detection system is provided, the system comprising a service end, an AI device, and a quality inspection image judgment end; The AI ​​device is used to obtain a security inspection image from a security inspection device, identify contraband on the security inspection image, obtain an intelligent image judgment result, and identify an image judgment mark in the security inspection image to obtain a manual image judgment result, and send the intelligent image judgment result and the manual image judgment result to the server; wherein the image judgment mark is information marked by the security inspector in the security inspection image during the image judgment process, and the manual image judgment result and the intelligent image judgment result are used to indicate whether there are contraband in the security inspection image; The server is used to determine whether the intelligent image judgment result is consistent with the manual image judgment result, and if they are inconsistent, send the security inspection image to the quality inspection image judgment end; The quality inspection image judging terminal is used to display the security inspection image so that the quality inspector can re-judge the security inspection image.

[0005] According to a second aspect of the present application, a security inspection quality detection method is provided, the method comprising: Acquire a security inspection image from a security inspection device, wherein the security inspection image includes an image judgment mark, which is information marked in the security inspection image by a security inspector during the image judgment process; the image judgment mark at least includes a manual image judgment result marked in the security inspection image by the security inspector, and the manual image judgment result is used to indicate whether there are contraband items in the security inspection image; The security inspection image is subjected to contraband identification to obtain an intelligent image judgment result, and the manual image judgment result is identified from the image judgment mark in the security inspection image, so as to determine whether to re-judge the security inspection image based on the matching situation between the intelligent image judgment result and the manual image judgment result.

[0006] According to a third aspect of the present application, a computer program product is provided, the computer program product comprising a computer program, and the computer program implements the method of the first aspect and / or the second aspect when executed by a processor.

[0007] According to a fourth aspect of the present application, an electronic device is provided, comprising a processor, a memory, and a computer program stored in the memory and executable by the processor, wherein the processor implements the method mentioned in the first aspect and / or the second aspect when executing the computer program.

[0008] According to a fifth aspect of the present application, a computer-readable storage medium is provided, on which a computer program is stored, and when the computer program is executed, the method mentioned in the first aspect is implemented.

[0009] By applying the solution provided in the present application, the AI ​​device obtains the security inspection image from the security inspection device, identifies contraband in the security inspection image, obtains the intelligent image judgment result, and identifies the image judgment mark in the security inspection image to obtain the manual image judgment result, and sends the intelligent image judgment result and the manual image judgment result to the server; when the server determines that the intelligent image judgment result and the manual image judgment result are inconsistent, it sends the security inspection image to the quality inspection image judgment end so that the quality inspector can re-judge the security inspection image.

[0010] The security inspection quality detection system provided in the embodiment of the present application obtains security inspection images from the security inspection equipment through an AI device installed near the security inspection equipment, and identifies the manual image judgment results of the security inspector from the security inspection images. There is no need for the server to obtain the image judgment results of the security inspector from the security inspection equipment. This can improve the efficiency of the server in obtaining the two image judgment results, thereby improving the quality inspection efficiency. At the same time, for security inspection equipment that does not provide a network access interface to external devices, security inspection images and manual image judgment results of the security inspector can also be obtained from such security inspection equipment, so that the above-mentioned AI technology-assisted quality inspection solution can be applied to all types of security inspection equipment. Moreover, the AI ​​technology-assisted quality inspection solution can greatly improve the quality inspection efficiency and reduce the risk of misjudgment and missed judgment.

[0011] It is to be understood that the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the present specification. BRIEF DESCRIPTION OF THE DRAWINGS

[0012] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative labor.

[0013] Figure 1 It is a schematic diagram of a security quality inspection system according to an embodiment of the present application.

[0014] Figure 2 It is a schematic diagram of a manual image judgment result of a security inspection image marked with a security inspector according to an embodiment of the present application.

[0015] Figure 3 This is a schematic diagram of an AI device acquiring a security inspection image from a security inspection device in an embodiment of the present application.

[0016] Figure 4 This is a schematic diagram of another embodiment of the present application in which an AI device obtains a security inspection image from a security inspection device.

[0017] Figure 5 It is a schematic diagram of a security inspection image marked with an item identification frame according to another embodiment of the present application.

[0018] Figure 6 It is a schematic diagram of a security inspection quality detection method according to another embodiment of the present application.

[0019] Figure 7 It is a schematic diagram of the logical structure of an electronic device according to an embodiment of the present application. DETAILED DESCRIPTION

[0020] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0021] In order to ensure the safety of passengers, security inspection equipment is used to conduct security inspections on passenger packages at airports, train stations and other places to determine whether there are prohibited items in the passenger packages. The security inspection equipment can collect images of the inspected passenger packages to obtain security inspection images, and then the security inspectors will determine the prohibited items on the collected security inspection images. At present, most of the security inspection images collected by the security inspection equipment are interpreted in the form of on-site or remote centralized image judgment. The security inspectors rely on their knowledge and experience to judge whether the package contains prohibited items. The experience and working status of the security inspectors will affect the accuracy of the image judgment results. Therefore, there is a risk of misjudgment and omission of prohibited items in the package to a certain extent. Once missed, it will have a great impact on the safety of passengers.

[0022] At present, the quality of security inspectors' image judgment can be monitored by quality inspectors. For example, quality inspectors can make a second judgment on the security inspectors' image judgment results to verify the accuracy of the security inspectors' image judgment results. However, due to the large number of security inspection images, most quality inspectors can only conduct verification through random inspections, and the quality inspection efficiency is low. In addition, the quality inspection cannot cover all security inspection images, and there is still a high risk of missed detection and misjudgment.

[0023] Taking into account the wide application of AI technology in the field of image recognition, the present application conceives that after the security inspector manually judges the security inspection image, the security inspection image can be intelligently judged based on AI technology. When the judgment results of the two are inconsistent, the quality inspector will conduct quality inspection on the security inspection image again. In this way, since the quality inspector only needs to review the security inspection images where the manual judgment results and the AI ​​judgment results are inconsistent, the quality inspection efficiency can be improved and the probability of missed detection or misjudgment can be greatly reduced.

[0024] At present, the security inspection equipment on the market generally includes X-ray security inspection machines and supporting display devices. The security inspection images collected by the X-ray security inspection machine can be displayed on the display device so that the security inspector can judge the security inspection images and determine whether there are contraband in the security inspection images. Considering that the security inspection equipment on the market may come from different manufacturers, some manufacturers' security inspection equipment may provide a network interface for external devices to access, that is, the external device can remotely access the security inspection equipment through the network interface and obtain the data in the security inspection equipment (for example, the security inspection images collected by the security inspection equipment, the security inspector's judgment results, etc.). However, some manufacturers' security inspection equipment will not provide a network interface for external devices to access, that is, external devices cannot remotely access the security inspection equipment and obtain the data in the security inspection equipment.

[0025] In order to apply the above-mentioned AI-based image judgment solution to security inspection equipment produced by various manufacturers, the present application proposes that an AI device can be deployed near the security inspection equipment, and an AI model for realizing the image judgment function can be deployed in the AI ​​device. Considering that a video transmission interface is usually configured in the security inspection equipment to output the collected security inspection images, therefore, for security inspection equipment that does not provide a network interface for external device access, the AI ​​device can be connected to the video transmission interface of the security inspection equipment to obtain the security inspection equipment from the video transmission interface. In addition, considering that currently when allocating security inspection images to the quality inspection image judgment end of the quality inspector, it is usually allocated by the server (for example, a cloud server). For security inspection equipment that does not provide an access interface to external devices, the server is usually unable to obtain the security inspector's image judgment results from the security inspection equipment and compare them with the image judgment results of the AI ​​device. In order to solve this problem, the present application has thought of using the AI ​​device to identify the image judgment results marked by the security inspector during the image judgment process from the security inspection image, and the AI ​​device sends the security inspector's image judgment results to the server, so that the server can compare the image judgment results of the AI ​​device with the image judgment results of the security inspector. If the two are inconsistent, they are allocated to the quality inspection image judgment end, and the quality inspector reviews the security inspection image. Through the above method, for all security inspection equipment on the market, a solution that uses AI technology to assist quality inspection can be deployed, and the efficiency of quality inspection can be improved.

[0026] Based on the above inventive concept, the embodiment of the present application provides a security inspection quality detection system, such as Figure 1 As shown, the system includes a server, an AI device, and a quality inspection and image judgment terminal. The server may be a cloud server or a server cluster, and the quality inspection and image judgment terminal may be a client deployed on a computer, a mobile phone or other device for realizing the image judgment function. For example, it may be an APP with an image judgment function. The server may be connected to one or more quality inspection and image judgment terminals in communication. The AI ​​device may be a device deployed near the security inspection device that can automatically determine whether there are contraband in the security inspection image. Usually, a pre-trained AI model may be deployed in the AI ​​device, and the AI ​​model may be used to identify contraband in the security inspection image, and to identify the information marked by the security inspector in the security inspection image.

[0027] In the process of determining whether there are prohibited items in the security inspection images captured by the security inspection equipment, the security inspector can mark relevant image judgment information in the security inspection images, such as the security inspector's image judgment results for the security inspection images, the locations of the prohibited items marked in the security inspection images, the types of prohibited items, etc. Therefore, the AI ​​device can obtain the security inspection images from the security inspection equipment, identify the contraband in the security inspection images, and obtain image judgment results, hereinafter referred to as intelligent image judgment results. At the same time, the AI ​​device can also identify the image judgment marks in the security inspection images to obtain the security inspector's image judgment results for the security inspection images, hereinafter referred to as manual image judgment results. Among them, the image judgment mark is the information marked by the security inspector in the security inspection image during the image judgment process. For example, the image judgment mark can be the image judgment result marked by the security inspector on the security inspection image, such as Figure 2 As shown, the image judgment result can be text information on the security inspection image, such as "safe" or "suspicious". Of course, the image judgment mark can also be the prohibited items selected by the security inspector in the image, and the types of prohibited items marked near the prohibited items, etc. Among them, the image judgment mark can be any information related to the image judgment marked by the security inspector in the security inspection image, and the embodiment of the present application is not limited.

[0028] After the AI ​​device obtains the intelligent image judgment result and the manual image judgment result from the security inspection image, it can send both to the server at the same time. Among them, the intelligent image judgment result and the manual image judgment result can be used to indicate whether there are contraband in the security inspection image. Of course, in some embodiments, the intelligent image judgment result and the manual image judgment result can also be used to indicate the type of contraband in the security inspection image.

[0029] The server can determine whether the intelligent image judgment result is consistent with the manual image judgment result. If not, the security inspection image can be sent to the quality inspection image judgment end so that the quality inspector can review the security inspection image. After the server determines that the intelligent image judgment result is consistent with the manual image judgment result, it can use the result as the final image judgment result. In addition, the server can also count the image judgment quality of each security inspector based on the two image judgment results sent by the AI ​​device.

[0030] After receiving the security inspection image sent by the server, the quality inspection image judgment terminal can display the security inspection image through the interactive interface so that the quality inspector can re-judge the security inspection image and obtain the final image judgment result.

[0031] The security inspection quality detection system provided by the embodiment of the present application obtains security inspection images from the security inspection equipment through an AI device set near the security inspection equipment, and identifies the manual image judgment results of the security inspector from the security inspection images. There is no need for the server to obtain the image judgment results of the security inspector from the security inspection equipment. This can improve the efficiency of the server in obtaining the two image judgment results, thereby improving the quality inspection efficiency. At the same time, for security inspection equipment that does not provide a network access interface to external devices, security inspection images and manual image judgment results of security inspectors can also be obtained from such security inspection equipment, so that the above-mentioned AI technology-assisted quality inspection solution can be applied to all types of security inspection equipment. Moreover, the AI ​​technology-assisted quality inspection solution can greatly improve the quality inspection efficiency and reduce the risk of misjudgment and missed judgment.

[0032] Considering that there are two types of security inspection equipment on the market, the first type of security inspection equipment is equipped with a network interface for external devices to access, and external devices can obtain data in the security inspection equipment through this network interface. The second type of security inspection equipment is not equipped with a network interface for external devices to access, that is, external devices cannot obtain data in the security inspection equipment through the network interface. Figure 3 As shown, for the first type of security inspection equipment, when the AI ​​device obtains the security inspection image from the security inspection equipment, it can access the designated storage space of the security inspection equipment based on the network interface to obtain the security inspection image from the storage space. Alternatively, the AI ​​device can also be connected to the security inspection equipment via a data cable and obtain the security inspection image from the security inspection equipment via the video transmission interface in the security inspection equipment. Figure 4 As shown, the AI ​​device can be connected to the security inspection device via a data cable, and obtain the security inspection image from the security inspection device via the video transmission interface in the security inspection device. The video transmission interface can be an HDMI interface, a VGA interface, etc., which is not limited in the embodiment of the present application.

[0033] Usually, when AI devices identify objects in images, they usually include two steps. First, the position of each object in the image can be determined through the target detection model to obtain the target detection frame corresponding to each object. Then, the object in each target detection frame can be identified through the target recognition model to determine the category of the object, for example, whether the object is a knife, liquid or other. Considering that security inspectors can usually mark the objects in the security inspection images during the image judgment process, for example, these objects can be framed in the security inspection images and the categories to which these objects belong can be marked. In order to improve the image judgment efficiency of AI devices, AI devices can reuse the information marked by security inspectors in security inspection images. For example, AI devices can directly identify the object identification frame marked by the security inspectors to determine the location of each object, thereby eliminating the target detection step and greatly improving the image judgment speed. Based on this, in some embodiments, such as Figure 5As shown, the image recognition mark may also include an item identification frame marked by the security inspector in the security inspection image, wherein each item identification frame includes an item. When the AI ​​device identifies prohibited items in the security inspection image, it may identify the above-mentioned item identification frame from the security inspection image, and then identify the item in each item identification frame to determine whether the item is a prohibited item.

[0034] For example, a contraband identification model can be deployed in the AI ​​device. The contraband identification model can be trained using a large number of sample security inspection images with labels, where the labels are used to indicate whether there are contraband in the sample security inspection images and the types of contraband. The contraband identification model can then be used to identify contraband in the security inspection images to obtain intelligent image discrimination results.

[0035] In some embodiments, the manual image judgment result may be text information indicating whether there are prohibited items in the security inspection image. When the AI ​​device identifies the manual image judgment result from the security inspection image, it may identify the above text information based on the optical character recognition algorithm (Optical Character Recognition) to obtain the manual image judgment result. For example, after completing the image judgment, the security inspector may mark the security inspection image with words such as "safe" or "suspicious" to indicate his own image judgment result. The AI ​​device may then recognize the words marked by the security inspector to obtain the manual image judgment result. Alternatively, the security inspector may also mark the type of prohibited items that he determines to exist in the security inspection image in the security inspection image, and mark it in the security inspection image, such as "knife", "lighter", etc. The AI ​​device may recognize these words and determine the manual image judgment result.

[0036] Considering that when a passenger passes through a security check channel, if the package he carries contains prohibited items, he will usually show some specific behavioral characteristics, such as evading security checks, rushing through security checks, nervous facial expressions, etc. Therefore, the AI ​​device can also assist in determining whether there are prohibited items in the security check image in combination with the behavioral characteristics of the passenger passing through the security check channel, so as to judge the security check image from multiple dimensions and obtain more accurate judgment results. Therefore, in some embodiments, a behavior detection model can be deployed in the AI ​​device, and the behavior detection model can be trained based on the passenger's historical behavior data and the corresponding historical judgment results. For example, the passenger's historical behavior data can be obtained from the passenger's historical inspection video as sample data, and the judgment result of the passenger's package image during the inspection can be obtained as the label of the sample data, and the passenger behavior detection model is trained using the above sample data. When the AI ​​device identifies prohibited items in security inspection images and obtains intelligent image judgment results, it can use the behavior detection model to extract the behavior characteristics of the passenger to whom the package in the security inspection image belongs from the target video, and determine the first probability that the security inspection image includes prohibited items based on the extracted behavior characteristics, wherein the target video is obtained by capturing the passenger during the process of the passenger passing through the security inspection channel. Then, based on the recognition result of the prohibited items identification in the security inspection image, the second probability that the security inspection image includes prohibited items can be determined, and the above first probability and second probability are weighted to obtain the target probability, wherein the weights of the two probabilities can be flexibly set based on actual needs, and then the intelligent image judgment result can be obtained based on the target probability.

[0037] AI devices can improve the accuracy of image recognition results by combining information from multiple dimensions to determine whether there are prohibited items in security inspection images.

[0038] In some embodiments, a contraband identification model is deployed in the AI ​​device, and the contraband identification model is used to identify contraband on the security inspection image and obtain an intelligent image judgment result. Among them, it is considered that the categories of contraband are diverse and new types of contraband will continue to appear. In order to improve the generalization ability of the contraband identification model and ensure the accuracy of the recognition results of the contraband identification model, training samples can be continuously collected during the use of the model to continuously train the contraband model. For example, for the scene where the manual image judgment result and the intelligent image judgment result are inconsistent, the quality inspector can review the security inspection image to obtain the final image judgment result. Since the image judgment result is usually more accurate, the image judgment result of the quality inspector can be used as the standard image judgment result of the security inspection image. The server can obtain the above-mentioned standard image judgment result from the quality inspection image judgment end and send it to the AI ​​device. After obtaining the standard image judgment result from the server, the AI ​​device can use the standard image judgment result as the label of the security inspection image and use the security inspection image as a training sample to train the contraband identification model.

[0039] In the process of using AI equipment to judge images, the final accurate judgment results can be continuously used as labels for security inspection images to construct new training samples. The new training samples can be used to continuously train the contraband recognition model, which can ensure that the contraband recognition model has a high degree of accuracy.

[0040] Usually, there are multiple quality inspectors, and thus there are also multiple quality inspection judgment terminals. When the server allocates security inspection images to be reviewed to these multiple quality inspection judgment terminals, it can adopt a reasonable task scheduling strategy to improve the quality inspection efficiency. For example, in some embodiments, the server can determine the allocation weight of each quality inspection judgment terminal based on the judgment data of each quality inspection judgment terminal, and then determine the target quality inspection judgment terminal from multiple quality inspection judgment terminals based on the allocation weight, and send the security inspection image to the target quality inspection judgment terminal. For example, the target quality inspection judgment terminal can be the quality inspection judgment terminal with the largest allocation weight, or the quality inspection judgment terminal with the highest allocation weight.

[0041] Among them, the image judgment data may include one or more items such as the quality inspection efficiency of the quality inspection image judgment terminal, the number of security inspection images to be inspected by the quality inspection image judgment terminal, and the last allocation time of the quality inspection image judgment terminal. Among them, the image judgment efficiency is the ratio of the total number of image judgments after the quality inspection image judgment terminal is registered online to the image judgment time. The quality inspection image judgment terminal can update the image judgment efficiency in the cache every time it obtains the image judgment result, so as to update its own image judgment efficiency in real time. The number of security inspection images to be inspected by the quality inspection image judgment terminal refers to how many frames of security inspection images are currently left for the quality inspection image judgment terminal to be inspected. The last allocation time refers to the last time the service end allocated a security inspection image to the quality inspection image judgment terminal.

[0042] Among them, the allocation weight is positively correlated with the quality inspection efficiency, that is, the higher the quality inspection efficiency of a certain quality inspection judgment terminal, the greater its allocation weight, that is, the more inclined to allocate the security inspection image to the quality inspection judgment terminal. The allocation weight is negatively correlated with the number of security inspection images to be inspected, that is, the fewer the number of security inspection images to be inspected by a certain quality inspection judgment terminal, the more inclined to allocate the security inspection image to the quality inspection judgment terminal. The allocation weight is negatively correlated with the most recent allocation time, that is, the smaller the most recent allocation time of the quality inspection judgment terminal, that is, the longer the time has passed without allocating a security inspection image to the quality inspection judgment terminal, the more inclined to allocate the security inspection image to the quality inspection judgment terminal.

[0043] In addition, the embodiment of the present application also provides a security inspection quality detection method, which can be executed by a device with an AI image judgment function, for example, the device can be an AI device in the above-mentioned security inspection quality detection system. Of course, in some scenarios, the function of the AI ​​device can also be directly integrated into the server, and the method can also be executed by the server.

[0044] like Figure 6 As shown, the method may include the following steps: S602, obtaining a security inspection image from a security inspection device, wherein the security inspection image includes an image judgment mark, and the image judgment mark is information marked in the security inspection image by a security inspector during the image judgment process; S604: The image judgment mark at least includes a manual image judgment result marked by the security inspector in the security inspection image, and the manual image judgment result is used to indicate whether there are contraband items in the security inspection image; S606: Perform contraband identification on the security inspection image to obtain an intelligent image judgment result, and identify a manual image judgment result from the image judgment mark in the security inspection image, so as to determine whether to review the security inspection image based on the matching situation between the intelligent image judgment result and the manual image judgment result.

[0045] The specific implementation details of each step of the method can be referred to the description in the above system real-time example, which will not be repeated here.

[0046] The method provided in the embodiment of the present application obtains intelligent image judgment results by identifying security inspection images. At the same time, it can identify manual image judgment results marked by security inspectors from security inspection images. Compared with the need to obtain manual image judgment results of security inspectors from security inspection equipment, it can greatly improve quality inspection efficiency and expand the application scenarios of the solution.

[0047] In some embodiments, the image judgment mark also includes an item identification box marked by the security inspector in the security inspection image, wherein each item identification box includes an item, and the prohibited item identification of the security inspection image to obtain an intelligent image judgment result includes: identifying the item identification box from the security inspection image, identifying the item in each item identification box to determine whether the item is a prohibited item.

[0048] In some embodiments, the manual image judgment result includes text information for indicating whether there are contraband items in the security inspection image, and identifying the manual image judgment result from the security inspection image includes: identifying the text information based on an optical character recognition algorithm to obtain the manual image judgment result.

[0049] In some embodiments, the performing contraband identification on the security inspection image to obtain an intelligent image judgment result includes: using the pre-trained behavior detection model to extract the behavior characteristics of the passenger to whom the package in the security inspection image belongs from the target video, and determining a first probability that the security inspection image includes contraband based on the behavior characteristics; wherein the target video is obtained by capturing a video of the passenger in the process of the passenger passing through the security inspection channel, and the behavior detection model is trained based on the passenger's historical behavior data and the corresponding historical image judgment results; determining a second probability that the security inspection image includes contraband based on the recognition result of the contraband identification on the security inspection image; weighting the first probability and the second probability to obtain a target probability, and determining the intelligent image judgment result based on the target probability.

[0050] In some embodiments, the contraband identification is performed on the security inspection image to obtain an intelligent image judgment result which is executed through a pre-trained contraband identification model. The method also includes: obtaining a standard image judgment result of a quality inspector for the security inspection equipment; using the standard image judgment result as a label for the security inspection image, and using the security inspection image as a training sample to train the contraband identification model.

[0051] Among them, the specific implementation details of the above-mentioned security inspection quality detection method can be referred to the description in the above-mentioned system real-time example, and will not be repeated here.

[0052] Among them, the solutions of the above-mentioned embodiments can be freely combined to obtain new solutions when there is no conflict. Due to space constraints, they will not be listed one by one here.

[0053] In addition, an embodiment of the present disclosure further provides a computer program product, which includes a computer program. When the computer program is executed by a processor, it implements any method mentioned in the above embodiments.

[0054] In addition, the present disclosure also provides an electronic device, such as Figure 7 As shown, the electronic device 70 includes a processor 71, a memory 72, and a computer program stored in the memory 72 and executable by the processor 71. When the computer program is executed by the processor 71, any method mentioned in the above embodiments is implemented.

[0055] Accordingly, an embodiment of the present specification further provides a computer storage medium, in which a program is stored. When the program is executed by a processor, the method in any of the above embodiments is implemented.

[0056] The embodiments of this specification may take the form of a computer program product implemented on one or more storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing program code. Computer-usable storage media include permanent and non-permanent, removable and non-removable media, and information storage can be implemented by any method or technology. Information can be computer-readable instructions, data structures, modules of programs, or other data. Examples of computer storage media include but are not limited to: phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, read-only compact disk read-only memory (CD-ROM), digital versatile disk (DVD) or other optical storage, magnetic cassettes, disk storage or other magnetic storage devices or any other non-transmission media that can be used to store information that can be accessed by a computing device.

[0057] For the device embodiment, since it basically corresponds to the method embodiment, the relevant parts can refer to the partial description of the method embodiment. The device embodiment described above is only schematic, wherein the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment. Ordinary technicians in this field can understand and implement it without paying creative work.

[0058] The user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of relevant data must comply with the relevant laws, regulations and standards of relevant countries and regions, and provide corresponding operation entrances for users to choose to authorize or refuse.

[0059] It should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. The terms "include", "comprises" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, the elements defined by the statement "comprises a ..." do not exclude the presence of other identical elements in the process, method, article or device including the elements.

[0060] The method and device provided in the embodiments of the present invention are introduced in detail above. Specific examples are used in this article to illustrate the principles and implementation methods of the present invention. The description of the above embodiments is only used to help understand the method of the present invention and its core idea. At the same time, for those skilled in the art, according to the idea of ​​the present invention, there will be changes in the specific implementation methods and application scopes. In summary, the content of this specification should not be understood as limiting the present invention.

Claims

1. A security inspection quality detection system, characterized in that: The system includes a server, an AI device, and a quality inspection and image judgment terminal; The AI ​​device is used to obtain a security inspection image from a security inspection device, identify contraband on the security inspection image, obtain an intelligent image judgment result, and identify an image judgment mark in the security inspection image to obtain a manual image judgment result, and send the intelligent image judgment result and the manual image judgment result to the server; wherein the image judgment mark is information marked by the security inspector in the security inspection image during the image judgment process, and the manual image judgment result and the intelligent image judgment result are used to indicate whether there are contraband in the security inspection image; The server is used to determine whether the intelligent image judgment result is consistent with the manual image judgment result, and if they are inconsistent, send the security inspection image to the quality inspection image judgment end; The quality inspection image judging terminal is used to display the security inspection image so that the quality inspector can re-judge the security inspection image.

2. The system according to claim 1, characterized in that The AI ​​device is used to obtain security inspection images from the security inspection device, including: The security inspection device is configured with a network interface for access by an external device, and the AI ​​device obtains the security inspection image from the security inspection device based on the network interface; and / or The security inspection device includes a video transmission interface, and the AI ​​device obtains the security inspection image from the security inspection device through the video transmission interface.

3. The system according to claim 1, characterized in that The image judgment mark also includes an item identification frame marked by the security inspector in the security inspection image, wherein each item identification frame includes an item. The AI ​​device is used to identify contraband items on the security inspection image. When the intelligent image judgment result is obtained, it is specifically used to: Identifying the object identification frame from the security inspection image; Identify the items in each item identification box to determine whether the item is a prohibited item.

4. The system according to claim 1, characterized in that The manual image judgment result includes text information for indicating whether there are contraband items in the security inspection image, and identifying the manual image judgment result from the security inspection image includes: The text information is recognized based on an optical character recognition algorithm to obtain the manual image judgment result.

5. The system according to claim 1, characterized in that The AI ​​device is deployed with a behavior detection model, and is used to identify contraband items on the security inspection image. When the intelligent image judgment result is obtained, it is specifically used to: The behavior detection model is used to extract the behavior characteristics of the passenger to whom the package in the security inspection image belongs from the target video, and a first probability that the security inspection image includes contraband is determined based on the behavior characteristics; wherein the target video is obtained by capturing a video of the passenger during the passenger passing through the security inspection channel, and the behavior detection model is trained based on the passenger's historical behavior data and the corresponding historical image judgment results; Determining a second probability that the security inspection image includes contraband based on a recognition result of contraband recognition performed on the security inspection image; Performing weighted processing on the first probability and the second probability to obtain a target probability; The intelligent image discrimination result is determined based on the target probability.

6. The system according to claim 1, characterized in that A contraband identification model is deployed in the AI ​​device, and the contraband identification model is used to identify contraband on the security inspection image to obtain the intelligent image judgment result; The server is also used to obtain the image judgment result of the quality inspector for the security inspection device from the quality inspection image judgment terminal as the standard image judgment result of the security inspection image; The AI ​​device is also used to obtain the standard image judgment result from the server, use the standard image judgment result as a label of the security inspection image, and use the security inspection image as a training sample to train the contraband recognition model.

7. The system according to claim 1, characterized in that The quality inspection and judgment end includes a plurality of terminals, and when the service end is used to send the security inspection image to the quality inspection and judgment end, it is specifically used to: Determine the allocation weight of each quality inspection image judging end based on the image judgment data of each quality inspection image judging end, wherein the image judgment data includes one or more of the following: the quality inspection efficiency of the quality inspection image judging end, the number of security inspection images to be inspected by the quality inspection image judging end, and the most recent allocation time of the quality inspection image judging end; wherein the allocation weight is positively correlated with the quality inspection efficiency, negatively correlated with the number of security inspection images to be inspected, and negatively correlated with the most recent allocation time; Determining a target quality inspection judgment image terminal from the plurality of quality inspection judgment image terminals based on the assigned weights; The security inspection image is sent to the target quality inspection image terminal.

8. A security inspection quality detection method, characterized in that: The method comprises: Acquire a security inspection image from a security inspection device, wherein the security inspection image includes an image judgment mark, which is information marked in the security inspection image by a security inspector during the image judgment process; the image judgment mark at least includes a manual image judgment result marked in the security inspection image by the security inspector, and the manual image judgment result is used to indicate whether there are contraband items in the security inspection image; The security inspection image is subjected to contraband identification to obtain an intelligent image judgment result, and the manual image judgment result is identified from the image judgment mark in the security inspection image, so as to determine whether to re-judge the security inspection image based on the matching situation between the intelligent image judgment result and the manual image judgment result.

9. The method according to claim 8, characterized in that The image judgment mark also includes an item identification frame marked by the security inspector in the security inspection image, wherein each item identification frame includes an item. The prohibited items are identified on the security inspection image to obtain an intelligent image judgment result, including: identifying the item identification frame from the security inspection image, identifying the item in each item identification frame to determine whether the item is a prohibited item.

10. A computer program product, characterized in that The computer program product comprises a computer program, and when the computer program is executed by a processor, the method according to any one of claims 8 to 9 is implemented.

Images