CN116416538A - Method, device, equipment and medium for discriminating images in security inspection - Google Patents

Abstract

The disclosure provides a method for discriminating images in security inspection, and relates to the field of artificial intelligence. The method comprises the following steps: acquiring a first discrimination result obtained by discriminating a target object in a security inspection image by an artificial intelligent model; acquiring a second discrimination result obtained by discriminating the target object in the security inspection image by a target security inspection person; when the first discrimination result is inconsistent with the second discrimination result, comparing the discrimination level of the artificial intelligent model on the similar articles of the target article with the discrimination level of the target security inspection personnel on the similar articles of the target article to obtain a comparison result; and determining a reliable discrimination result from the first discrimination result and the second discrimination result based on preset conditions satisfied by the comparison result. The present disclosure also provides an apparatus, a device, a storage medium, and a program product for discriminating an image in security inspection.

Description

Method, device, equipment and medium for discriminating images in security inspection

Technical Field

The present disclosure relates to the field of artificial intelligence, and more particularly to a method, apparatus, device, medium and program product for discriminating images in security inspection.

Background

Security inspection is an indispensable link in public transportation to ensure public safety. In the current security inspection process, X-ray security inspection equipment is mostly adopted, and after security inspection images are acquired through the X-ray security inspection equipment, the security inspection images are sent to a graph judging station to be manually judged by security inspection personnel so as to identify dangerous goods. However, at present, the identification of dangerous goods is mainly performed on the images acquired by the security inspection equipment, the manual identification mode needs security inspection personnel to focus on a screen in a highly concentrated mode, the working strength is very high, the experience and the working state of the images are seriously judged by the security inspection personnel, and the missed inspection is easy to generate.

Disclosure of Invention

In view of the foregoing, the present disclosure provides a method, apparatus, device, medium, and program product for discriminating an image in security inspection, which can improve the accuracy of discriminating a security inspection image.

According to a first aspect of the present disclosure, a method for discriminating an image in security inspection is provided. The method comprises the following steps: acquiring a first discrimination result obtained by discriminating a target object in a security inspection image by an artificial intelligent model; acquiring a second discrimination result obtained by discriminating the target object in the security inspection image by a target security inspection person; when the first discrimination result is inconsistent with the second discrimination result, comparing the discrimination level of the artificial intelligent model on the similar articles of the target article with the discrimination level of the target security inspection personnel on the similar articles of the target article to obtain a comparison result; and determining a reliable discrimination result from the first discrimination result and the second discrimination result based on preset conditions satisfied by the comparison result.

According to an embodiment of the present disclosure, the discrimination level includes at least one of an identification rate, a false positive rate, or a false negative rate.

According to an embodiment of the present disclosure, the discrimination level of the artificial intelligence model is higher than a preset reference level.

According to an embodiment of the disclosure, the selecting a trusted discrimination result from the first discrimination result and the second discrimination result based on the preset condition satisfied by the comparison result includes: and when the comparison result shows that the discrimination level of the target security inspection personnel on the similar objects of the target object is higher than that of the artificial intelligent model, determining that the second discrimination result is a credible discrimination result.

According to an embodiment of the disclosure, the selecting a trusted discrimination result from the first discrimination result and the second discrimination result based on the preset condition satisfied by the comparison result includes: and when the comparison result shows that the discrimination level of the target security inspection personnel on the similar articles of the target articles is lower than the discrimination level of the artificial intelligent model and the difference is greater than or equal to a preset threshold value, determining that the first discrimination result is a reliable discrimination result.

According to an embodiment of the present disclosure, the method further comprises setting the threshold value based on historical security performance data of the target security personnel. The historical security inspection performance data comprises at least one or more of the discrimination level of the target security inspection personnel on the similar objects of the target objects, the duration of the target security inspection personnel in the security inspection work or the experience rating data of the target security inspection personnel in the security inspection work.

According to an embodiment of the present disclosure, after setting the threshold value, the method further includes adjusting the threshold value based on the field security check condition data at the time of obtaining the second discrimination result. The on-site security inspection status data comprise duration of continuous security inspection judgment by the target security inspection personnel at the moment of obtaining the second judgment result or congestion degree data of articles subjected to on-site security inspection judgment.

According to an embodiment of the disclosure, after the determining that the first discrimination result is a trusted discrimination result, the method further includes sending notification alert information to a target terminal to alert a user of the target terminal to re-discriminate the target item.

According to an embodiment of the present disclosure, the method further comprises: acquiring historical judgment chart data obtained by judging various articles by at least one security personnel in the process of judging the historical security images; and obtaining the discrimination level of each security personnel in the at least one security personnel on the various objects according to the historical graph discrimination data. The at least one security personnel comprises the target security personnel, and the various articles comprise the same type of articles as the target articles.

In a second aspect of embodiments of the present disclosure, an apparatus for discriminating an image in security inspection is provided. The device comprises a first acquisition module, a comparison module and a determination module. The first acquisition module is used for acquiring a first discrimination result obtained by discriminating the target object in the security inspection image by the artificial intelligence model and acquiring a second discrimination result obtained by discriminating the target object in the security inspection image by a target security inspection person. And the comparison module is used for comparing the discrimination level of the artificial intelligent model on the similar articles of the target article with the discrimination level of the target security inspection personnel on the similar articles of the target article when the first discrimination result and the second discrimination result are inconsistent, so as to obtain a comparison result. The determining module is used for determining a reliable judging result from the first judging result and the second judging result based on preset conditions met by the comparing result.

According to an embodiment of the disclosure, the apparatus further comprises a second acquisition module and an evaluation module. The second acquisition module is used for acquiring historical judgment chart data obtained by judging various articles in the process of judging the historical security image by at least one security personnel. The evaluation module is used for obtaining the discrimination level of each security personnel in the at least one security personnel on the various objects according to the historical graph discrimination data, wherein the at least one security personnel comprises the target security personnel, and the various objects comprise the similar objects of the target object.

According to an embodiment of the disclosure, the apparatus further comprises a threshold setting module. The threshold setting module is used for setting the threshold based on the historical security inspection performance data of the target security inspection personnel. The historical security inspection performance data comprises at least one or more of the discrimination level of the target security inspection personnel on the similar objects of the target objects, the duration of the target security inspection personnel in the security inspection work or the experience rating data of the target security inspection personnel in the security inspection work.

According to an embodiment of the disclosure, the apparatus further comprises a threshold adjustment module. The threshold value adjusting module is used for adjusting the threshold value based on-site security inspection status data when the second judging result is obtained, wherein the on-site security inspection status data comprises duration of continuous security inspection judgment by the target security inspection personnel or congestion degree data of articles subjected to on-site security inspection judgment when the second judging result is obtained.

According to an embodiment of the disclosure, the apparatus further comprises a message alert module. The message reminding module is used for sending notification reminding information to a target terminal after the first judging result is determined to be a reliable judging result so as to remind a user of the target terminal to judge the target object again.

In a third aspect of the present disclosure, an electronic device is provided. The electronic device includes one or more processors and memory. The memory is used for storing one or more programs. Wherein the one or more programs, when executed by the one or more processors, cause the one or more processors to perform the above-described method for discriminating images in security.

In a fourth aspect of the present disclosure, a computer-readable storage medium is provided, on which executable instructions are stored, which instructions, when executed by a processor, cause the processor to perform the above-described method for discriminating images in security inspection.

In a fifth aspect of the present disclosure, a computer program product is provided, comprising a computer program which, when executed by a processor, implements the above-described method for discriminating images in security inspection.

Drawings

The foregoing and other objects, features and advantages of the disclosure will be more apparent from the following description of embodiments of the disclosure with reference to the accompanying drawings, in which:

FIG. 1 schematically illustrates application scenarios of methods, apparatuses, devices, media and program products for discriminating images in security inspection according to embodiments of the present disclosure;

FIG. 2 schematically illustrates a flow chart of a method for discriminating images in security inspection according to an embodiment of the disclosure;

FIG. 3 schematically illustrates a flow chart of a method for discriminating images in security inspection according to another embodiment of the disclosure;

FIG. 4 schematically illustrates a flow chart of a method for discriminating images in security inspection according to yet another embodiment of the disclosure;

fig. 5 schematically illustrates an application concept of a method for discriminating an image in security inspection according to an embodiment of the present disclosure;

FIG. 6 schematically illustrates a block diagram of an apparatus for discriminating images in security inspection according to an embodiment of the disclosure; and

fig. 7 schematically illustrates a block diagram of an electronic device adapted to implement a method for discriminating images in security inspection according to an embodiment of the disclosure.

Detailed Description

Hereinafter, embodiments of the present disclosure will be described with reference to the accompanying drawings. It should be understood that the description is only exemplary and is not intended to limit the scope of the present disclosure. In the following detailed description, for purposes of explanation, numerous specific details are set forth in order to provide a thorough understanding of the embodiments of the present disclosure. It may be evident, however, that one or more embodiments may be practiced without these specific details. In addition, in the following description, descriptions of well-known structures and techniques are omitted so as not to unnecessarily obscure the concepts of the present disclosure.

The terminology used herein is for the purpose of describing particular embodiments only and is not intended to be limiting of the disclosure. The terms "comprises," "comprising," and/or the like, as used herein, specify the presence of stated features, steps, operations, and/or components, but do not preclude the presence or addition of one or more other features, steps, operations, or components.

All terms (including technical and scientific terms) used herein have the same meaning as commonly understood by one of ordinary skill in the art unless otherwise defined. It should be noted that the terms used herein should be construed to have meanings consistent with the context of the present specification and should not be construed in an idealized or overly formal manner.

Where expressions like at least one of "A, B and C, etc. are used, the expressions should generally be interpreted in accordance with the meaning as commonly understood by those skilled in the art (e.g.," a system having at least one of A, B and C "shall include, but not be limited to, a system having a alone, B alone, C alone, a and B together, a and C together, B and C together, and/or A, B, C together, etc.).

In the technical scheme of the disclosure, the processes of acquiring, collecting, storing, using, processing, transmitting, providing, disclosing, applying and the like of the data all conform to the regulations of related laws and regulations, necessary security measures are adopted, and the public order harmony is not violated.

In order to reduce the dependence on the experience and working state of the security personnel in the security image discrimination process and improve the security image discrimination accuracy, the artificial intelligent model can be utilized to assist the security personnel in discriminating the security image or recheck the discrimination result of the security personnel. However, the identification rates of the artificial intelligence model on different kinds of objects are different, and great differences exist. When the discrimination result of the artificial intelligent model is excessively acquired, a great amount of unnecessary re-inspection work can be caused by misjudgment easily caused by the artificial intelligent model on some types of articles.

In this regard, the embodiments of the present disclosure provide a method, apparatus, device, medium, and program product for discriminating an image in security inspection, which can analyze and determine a discrimination result to be adopted in comparison with a discrimination level of a security inspection person for a similar article and a discrimination level of an artificial intelligent model for a similar article in a case where the discrimination result of the security inspection person is inconsistent with the discrimination result of the artificial intelligent model. In this way, the experience of manual discrimination of the security personnel is brought into play, the accuracy of discrimination of the security image is improved by the aid of the artificial intelligent model, and the manual discrimination skill is improved by the aid of the motivation security personnel.

Fig. 1 schematically illustrates an application scenario 100 of a method, apparatus, device, medium and program product for discriminating images in security inspection according to an embodiment of the disclosure.

As shown in fig. 1, an application scenario 100 according to this embodiment may include at least one X-ray security inspection device (two X-ray security inspection devices 101 and 102 are shown in the figure), a server 103, a graph-determining terminal 104, and an artificial intelligence graph-determining device 105.

The X-ray security devices 101, 102 may scan items conveyed through their interiors to obtain security images.

The server 103 may send the security inspection image to the image judging terminal 104 and the artificial intelligence image judging device 105, respectively.

The graph judging terminal 104 can display the security inspection image to the security inspection personnel 10 and receive the operation of judging whether the security inspection personnel 10 contain forbidden articles or not in the security inspection image. In one embodiment, when the security personnel 10 considers that the items in the security image do not contain forbidden items, the security personnel 10 may not perform any operation, and the graph judging terminal 104 may default to judge that the items in the security image do not contain forbidden items. In one embodiment, when the security personnel 10 consider that the items in the security image contain forbidden items, the security personnel 10 can select the security image through the graph judging terminal 104, and further, can also circle the forbidden items in the security image. In some embodiments, the security personnel 10 may also indicate the class of contraband when the contraband is circled.

An artificial intelligence model is deployed in the artificial intelligence image determining apparatus 105, and the received security inspection image may be determined by using the artificial intelligence model to determine whether the security inspection image includes contraband in the article. In some embodiments, the artificial intelligence model may also give a class of contraband while discriminating the contraband. The algorithm for discriminating the image by the artificial intelligent model can be an analysis algorithm generated by an artificial intelligent learning platform through learning an X-ray scanning image, for example, a target detection algorithm (convolutional neural network) based on deep learning, and the detection and identification of the target are completed by extracting the characteristics of contraband data from massive X-ray scanning image data and modeling the characteristics of the contraband.

The method for discriminating an image in security inspection according to the embodiment of the present disclosure may be executed by the server 103 or may be executed by the image discriminating terminal 104. Accordingly, the apparatus, device, medium, and program product for discriminating an image in security inspection of the embodiment of the present disclosure may be provided in the server 103, or may also be provided in the image discriminating terminal 104.

The image determining terminal 104 may be a terminal device for determining a security check image, which is used by a security check person at a security check site, a terminal device for determining a security check image, which is used by a security check image rechecking person, or a terminal device for determining a security check image, which is used by a security check quality control person, which is not limited in this disclosure.

It is to be understood that the application scenario shown in fig. 1 is merely exemplary, and that the disclosed embodiments may be applied in any scenario, architecture, and environment. The number of terminal devices, networks and servers in fig. 1 is merely illustrative. There may be any number of terminal devices, networks, and servers, as desired for implementation.

The method for discriminating an image in security inspection according to the disclosed embodiment will be described in detail with reference to fig. 2 to 5 based on the scene described in fig. 1.

Fig. 2 schematically illustrates a flow chart of a method for discriminating images in security inspection according to an embodiment of the disclosure.

As shown in fig. 2, the method 200 according to this embodiment may include operations S210 to S240.

In operation S210, a first determination result obtained by determining a target object in a security inspection image by an artificial intelligence model is obtained. The first discrimination result may be used to indicate whether the target item belongs to contraband. In some embodiments, the first discrimination result may also be used to indicate a class of contraband to which the target item belongs when the target item belongs to contraband. The artificial intelligence model may be, for example, an algorithmic model of the operation of the artificial intelligence device 105 of fig. 1.

In operation S220, a second determination result obtained by determining the target object in the security inspection image by the target security inspection personnel is obtained. The second discrimination result may be used to indicate whether the target item belongs to contraband. In some embodiments, the second discrimination result may also be used to indicate a class of contraband to which the target item belongs when the target item belongs to contraband.

The term "target security personnel" is used herein to refer to security personnel who judge the same object in the same security image with the artificial intelligent model, and can be security personnel who judge the image in real time on the security site, or can be security personnel who review, or can also be quality control personnel. In the application scenario 100 shown in fig. 1, the target security personnel may be security personnel 10. The graph judging terminal 104 can obtain the judging result of the security personnel 10 by receiving the operation of the security personnel 10. It should be noted that the order of the operations S210 and S220 in fig. 2 is merely exemplary and not limiting, and the two operations may be replaced in parallel or sequentially in practice.

In operation S230, when the first discrimination result and the second discrimination result are inconsistent, the discrimination level of the similar object of the object by the artificial intelligent model and the discrimination level of the similar object of the object by the object security inspection personnel are compared to obtain a comparison result. The similar objects of the target object can be determined according to the type of forbidden objects of the target object identified in the first discrimination result or the second discrimination result; or may be based on the security personnel or artificial intelligence model identifying the category of the target item.

According to some embodiments of the present disclosure, the discrimination level may include at least one of a recognition rate, a false positive rate, or a false negative rate. For example, the discrimination level may be characterized by one of an identification rate, a false positive rate, or a false negative rate. Alternatively, for example, the discrimination level may be characterized by weighting two or all of the recognition rate, false positive rate, or omission rate. In some embodiments, the identification rate of a certain class of contraband may refer to the ratio of the number of times that the class of contraband is correctly identified to the number of times that the class of contraband actually appears in the history discrimination process. Accordingly, the false alarm rate of a certain type of forbidden articles can be the ratio of the times of identifying other articles as the forbidden articles to the times of actually appearing the forbidden articles in the history discrimination process. Similarly, the rate of missing reports of a certain class of contraband may be the ratio of the number of times that the class of contraband is not identified to be in the history of discrimination to the number of times that the class of contraband is actually present. Of course, calculating the discrimination level at the recognition rate, false positive rate, or false negative rate is only one example. In practice, the discrimination level may be calculated by combining data such as the working efficiency and working experience of security personnel, and this may be set according to the actual situation.

In operation S240, a reliable discrimination result is determined from the first discrimination result and the second discrimination result based on the preset condition satisfied by the comparison result.

In one embodiment, when the comparison result shows that the discrimination level of the target security personnel on the similar objects of the target object is higher than that of the artificial intelligence model, the second discrimination result is determined to be a reliable discrimination result. And when the discrimination level of the target security personnel is higher, the manual discrimination conclusion of the target security personnel is determined to be right.

In one embodiment, when the comparison result indicates that the discrimination level of the object security personnel on the similar objects of the object is lower than the discrimination level of the artificial intelligence model, the first discrimination result is determined to be a reliable discrimination result. That is, when the discrimination level of the artificial intelligence model is higher, the discrimination result of the artificial intelligence model is in control.

In another embodiment, the first discrimination result is determined to be a trusted discrimination result only when the comparison result indicates that the discrimination level of the target security personnel for the similar items of the target item is lower than the discrimination level of the artificial intelligence model and the difference is greater than or equal to a preset threshold. And when the difference distance does not meet the preset threshold value, selecting a second judgment result of the acquisition message. Specifically, the identification rates of the artificial intelligent model on different types of articles are distributed at 60-90, the difference is large, and the sensitivity of different security check personnel to different types of suspected articles is also different due to experience and capability, so that the identification accuracy, the reliability and the like of the different security check personnel and the artificial intelligent model for different types of articles are greatly different when the images are identified in security check. In addition, in the security inspection process, the placement mode and the shielding mode of the articles can cause irregular imaging effect of the articles, the form is complex and changeable, and the manual work can be more flexible and sensitive when dealing with the complex situations. According to the embodiment of the disclosure, in the process of utilizing the artificial intelligent model to assist in judging the image or the image review, besides considering the judging level of security personnel and the artificial intelligent model, different thresholds can be set as early warning signals according to the characteristics of different security personnel, so that unnecessary review caused by misjudgment of the artificial intelligent model is reduced.

According to some embodiments of the present disclosure, the discrimination level of the artificial intelligence model may be required to be higher than a predetermined reference level. For example, the recognition rate, false positive rate, and/or omission rate of the artificial intelligence model may be set to the corresponding base values in advance, and the above comparative analysis of operation S230 and operation S240 may be performed only when the correlation value of the artificial intelligence model is higher than the base value. This is because, when the discrimination level of the artificial intelligence model is higher than the reference level, the discrimination result of the artificial intelligence model is used to assist in the analysis of the artificial discrimination result, which has a strong practical value.

According to the embodiment of the disclosure, when the judgment result of the target security inspection personnel is abandoned by taking the judgment result (namely, the first judgment result) of the artificial intelligent model as the reliable judgment result, notification reminding information can be sent to the target terminal to remind the user of the target terminal to re-judge the target object. In one embodiment, the target terminal may be a discrimination terminal used by the target security inspector, for example, to alert the security inspector 10 to re-discriminate in the application scenario 100. In another embodiment, the target terminal may be a terminal used by security personnel at a higher level than the target security personnel, e.g., a terminal device of a quality control personnel, to alert the quality control personnel to the intervention process.

Fig. 3 schematically illustrates a flow chart of a method for discriminating images in security inspection according to another embodiment of the disclosure.

As shown in fig. 3, the method 300 according to this embodiment may include operations S310 to S330, and operations S210 to S240.

In operation S310, historical judgment chart data obtained by judging various articles in the process of judging the historical security image by at least one security personnel is obtained. The at least one security personnel includes a target security personnel. The historical judgment chart data can be, for example, data such as judgment efficiency of each security inspection personnel on each type of forbidden articles in various types of articles, false alarm rate, omission rate and identification rate of each type of forbidden articles, frequency of box opening inspection on each type of forbidden articles, accuracy rate of box opening inspection results and the like.

In operation S320, a discrimination level of each security personnel in the at least one security personnel for each type of article is obtained according to the historical graph data. And calculating according to the calculation mode of the discrimination level by using the historical discrimination map data.

In operation S330, a discrimination level of the artificial intelligence model for various items is acquired. For example, the discrimination level of the artificial intelligence model for each type of article is calculated based on historical discrimination data after the artificial intelligence model is put into use, or based on discrimination data for various types of articles during training or testing of the artificial intelligence model.

It should be noted that, the operation S330 is merely exemplary after the operation S310 and the operation S320, and the operation S330 and the operation S320 have no logical sequence, and may be performed in parallel or the operation S330 may be performed before the operation S310.

And next, through operations S210 to S240, judging whether the target object in the image belongs to the forbidden object through the artificial intelligent model and the target security inspector respectively, and when the judging results of the artificial intelligent model and the target security inspector are inconsistent, determining which judging result is adopted according to the comparison of the judging levels of the similar objects of the target object by the artificial intelligent model and the target security inspector respectively. For details, reference may be made to the descriptions of operations S210 to S240, which are not repeated here.

According to the embodiment of the disclosure, the large data analysis can be performed on the historical judgment chart data of different security check personnel to obtain the judgment level when the different security check personnel judge each type of article, so that the skill characteristics of each security check personnel can be fully mastered, and the large data analysis image of the security check personnel is realized. Therefore, when the judgment results of the security check personnel and the artificial intelligent model are selected, the difference of the security check personnel is considered, so that the security check accuracy is improved, meanwhile, the artificial experience can be effectively exerted, the improvement of the skill improvement of the security check personnel is facilitated, and the personnel enthusiasm is improved.

Fig. 4 schematically illustrates a flow chart of a method for discriminating images in security inspection according to yet another embodiment of the disclosure.

As shown in fig. 4, the method 400 according to the embodiment may include operations S410, S210 to 220, S430, S230, and S440.

First, in operation S410, a threshold is set based on historical security performance data of the target security personnel, wherein the historical security performance data includes at least one or more of a discrimination level of the target security personnel on the similar items of the target item, a duration of a security work performed by the target security personnel, or experience rating data of the target security personnel in the security work.

Next, in operation S210, a first determination result obtained by determining, by the artificial intelligence model, the target object in the security inspection image is obtained. Reference may be made in particular to the description above with respect to operation S210.

In addition, in operation S220, a second determination result is obtained by determining, by the target security personnel, the target object in the security image. Reference is made in particular to the description related above. Reference may be made in particular to the description above with respect to operation S220.

And then, in operation S230, when the first discrimination result and the second discrimination result are inconsistent, comparing the discrimination level of the artificial intelligence model on the similar article of the target article with the discrimination level of the target security inspection personnel on the similar article of the target article to obtain a comparison result.

Next, in operation S440, when the comparison result indicates that the discrimination level of the object security personnel for the similar objects of the object is lower than the discrimination level of the artificial intelligence model and the difference is greater than or equal to the threshold value, it is determined that the first discrimination result is a trusted discrimination result.

In the embodiment, corresponding thresholds can be set for various articles aiming at target security personnel, and the information collection condition of the discrimination results of the target security personnel can be pre-warned and regulated through the thresholds, so that the actual condition that the types of the articles which are good for being identified by the target security personnel are possibly different is fully considered, and the process of selecting the credible discrimination results is more accurate. When the target security personnel judge that the level is higher, the working time is longer, or the experience rating is higher, or the target security personnel are better at processing the judgment of a certain class of objects, the probability of security inspection errors of the target security personnel is lower, the corresponding threshold value can be set to be larger, and accordingly repeated judgment work caused by false inspection of the artificial intelligent model can be effectively reduced.

Furthermore, as the target security personnel can be any security personnel performing image discrimination in the security links, according to the embodiment of the disclosure, the respective corresponding threshold values can be set on the discrimination of various articles for different security personnel, so that the different security personnel can correspondingly have the threshold values on the discrimination of various articles, the process of selecting the reliable discrimination results is realized, the personal characteristics of the different security personnel are taken into consideration, and the obtained reliable discrimination results are more targeted, more accurate and more reliable.

With continued reference to fig. 4, the threshold value preset in operation S410 may also be dynamically adjusted through operation S430 before operation S230 according to an embodiment of the present disclosure.

Specifically, in operation S430, the threshold value is adjusted based on the on-site security inspection condition data when the second discrimination result is obtained, where the on-site security inspection condition data includes a time period during which the target security inspection personnel continuously performs security inspection discrimination by the time when the second discrimination result is obtained, or congestion degree data of the article on-site performing security inspection discrimination. For example, when the target security inspector continuously discriminates images for more than a certain period of time (e.g., 1 hour), the set threshold may be lowered to undershoot the target security inspector with the possible problem of reduced sensitivity to images due to fatigue or the like. For another example, when the congestion degree of the security check site reaches a serious level, the threshold set for the target security check operator can be lowered, so that the possibility of missed check caused by heavy business is reduced.

Therefore, the embodiment of the disclosure can dynamically adjust the preset threshold value of the target security personnel in each type of article discrimination according to the working state of the security inspection site target security personnel or the busyness of the security inspection requirement of the site, so that the selection of the artificial discrimination result and the artificial intelligent model discrimination result is more fit with the site real condition, the phenomena of false inspection and missing inspection are reduced, and the security inspection accuracy and efficiency are improved.

Fig. 5 schematically illustrates an application concept of a method for discriminating an image in security inspection according to an embodiment of the present disclosure.

As shown in fig. 5, according to an embodiment of the present disclosure, when an artificial intelligence model performs image review or real-time auxiliary image discrimination, a forbidden article is discriminated (S501), and is inconsistent with the discrimination conclusion of the image discrimination by a security inspection person (S502), the recognition rate, false alarm rate, omission rate (S503 and S504) of the article by the security inspection person and the artificial intelligence model, respectively, are queried, and the recognition rate, false alarm rate, omission rate of the article are compared with the artificial intelligence model (S505), and then the discrimination conclusion of the security inspection person or the discrimination conclusion of the security inspection artificial intelligence model is determined according to different comparison rules. If the judgment result of the security personnel is in the right (S506), the image judgment is finished. If the judgment result of the artificial intelligence model is positive, the security inspection personnel can be reminded to conduct secondary judgment or remind the security inspection quality control personnel to conduct intervention treatment (S507).

According to the embodiment of the disclosure, the data preprocessing can be performed by analyzing and processing the security inspection data such as the unpacking inspection rate of security inspection personnel, the image discrimination historical data, the unpacking inspection rate and the unpacking inspection result of different articles, the artificial intelligent model rechecking and the artificial intelligent model auxiliary discrimination image data, and the like, and by utilizing methods such as an interpolation method, a box graph analysis and the like; performing feature selection and processing by using methods such as correlation analysis, feature normalization, principal component analysis (principal components analysis, PCA) dimension reduction and the like; training and modeling are comprehensively carried out by using K nearest neighbor (K-NearestNeighbor KNN), bayesian, decision tree, random forest, adaBoost and other algorithms, and data such as judging image efficiency, unpacking inspection rate, false alarm rate, missing inspection rate, identification rate and the like of security personnel are calculated, so that the identification rate, false alarm rate and missing inspection rate of various forbidden articles of the security personnel are calculated, and judging level of the security personnel on various articles is evaluated. Further, the embodiment of the disclosure can set corresponding threshold parameters for various articles corresponding to each security inspection personnel, and the threshold parameters are used as a selection basis for selecting a reliable discrimination result between the discrimination of manual discrimination and the discrimination result of the artificial intelligent model, so that unnecessary rechecking caused by misjudgment of the artificial intelligent model on some types of articles is reduced.

Based on the method for distinguishing the image in the security inspection of each embodiment, the disclosure also provides a device for distinguishing the image in the security inspection. The device will be described in detail below in connection with fig. 6.

Fig. 6 schematically illustrates a block diagram of an apparatus 600 for discriminating images in security inspection according to an embodiment of the disclosure.

As shown in fig. 6, the apparatus 600 for discriminating an image in security inspection may include a first acquisition module 610, a comparison module 620, and a determination module 630.

The first obtaining module 610 is configured to obtain a first determination result by determining a target object in the security image by using the artificial intelligence model, and obtain a second determination result by determining a target object in the security image by using a target security personnel. In some embodiments, the first acquisition module 610 may perform operations S210 and S220 described above.

The comparison module 620 is configured to compare the discrimination level of the similar object of the target object by the artificial intelligence model with the discrimination level of the similar object of the target object by the target security inspection personnel when the first discrimination result and the second discrimination result are inconsistent, and obtain a comparison result. In some embodiments, the comparison module 620 may perform operation S230 described above.

The determining module 630 is configured to determine a reliable discrimination result from the first discrimination result and the second discrimination result based on a preset condition satisfied by the comparison result. In some embodiments, the determination module 630 may perform operation S240 described above.

According to further embodiments of the present disclosure, the apparatus 600 may further comprise a second acquisition module 640 and an evaluation module 650.

The second obtaining module 640 is configured to obtain historical judgment chart data obtained by at least one security personnel in judging various objects in the process of judging the historical security image. In some embodiments, the second acquisition module 640 may perform operation S310 described above.

The evaluation module 650 is configured to obtain a discrimination level of each security personnel of the at least one security personnel for each type of article according to the historical graph data, where the at least one security personnel includes a target security personnel, and each type of article includes a similar article of the target article. In some embodiments, the evaluation module 650 may perform operation S320 described above. In other embodiments, the evaluation module 630 may further perform operation S330 described above for obtaining a discrimination level of the artificial intelligence model for various types of articles.

According to further embodiments of the present disclosure, the apparatus 600 may further comprise a threshold setting module 660, a threshold adjustment module 670, and/or a message alert module 680.

The threshold setting module 660 is configured to set a threshold based on historical security performance data of the target security personnel. The historical security inspection performance data comprises at least one or more of the discrimination level of the target security inspection personnel on the similar objects of the target objects, the duration of the target security inspection personnel in the security inspection work or the experience rating data of the target security inspection personnel in the security inspection work. In some embodiments, the threshold setting module 660 may perform operation S410 described above.

The threshold adjustment module 670 is configured to adjust a threshold based on field security check condition data when the second determination result is obtained, where the field security check condition data includes a duration of continuous security check determination by a target security check person at a time when the second determination result is obtained, or congestion degree data of an article subjected to security check determination on the spot. In some embodiments, the threshold adjustment module 670 may perform operation S430 described above.

The message reminding module 680 is configured to send notification reminding information to the target terminal after determining that the first discrimination result is a trusted discrimination result, so as to remind the user of the target terminal to re-discriminate the target object. In some embodiments, the message alert module 680 may perform operation S507 described above.

The apparatus 600 may be used to implement the methods described with reference to fig. 2-5 according to embodiments of the present disclosure, and are specifically described above, and are not described here again.

According to embodiments of the present disclosure, any of the first acquisition module 610, the comparison module 620, the determination module 630, the second acquisition module 640, the evaluation module 650, the threshold setting module 660, the threshold adjustment module 670, or the message alert module 680 may be combined in one module to be implemented, or any of the modules may be split into multiple modules. Alternatively, at least some of the functionality of one or more of the modules may be combined with at least some of the functionality of other modules and implemented in one module. According to embodiments of the present disclosure, at least one of the first acquisition module 610, the comparison module 620, the determination module 630, the second acquisition module 640, the evaluation module 650, the threshold setting module 660, the threshold adjustment module 670, or the message alert module 680 may be implemented at least in part as hardware circuitry, such as a Field Programmable Gate Array (FPGA), a Programmable Logic Array (PLA), a system-on-chip, a system-on-substrate, a system-on-package, an Application Specific Integrated Circuit (ASIC), or may be implemented in hardware or firmware in any other reasonable way of integrating or packaging a circuit, or in any one of or a suitable combination of any of three implementations of software, hardware, and firmware. Alternatively, at least one of the first acquisition module 610, the comparison module 620, the determination module 630, the second acquisition module 640, the evaluation module 650, the threshold setting module 660, the threshold adjustment module 670, or the message alert module 680 may be at least partially implemented as a computer program module that, when executed, may perform the corresponding functions.

Fig. 7 schematically illustrates a block diagram of an electronic device 700 adapted to implement a method for discriminating images in security inspection according to an embodiment of the disclosure.

As shown in fig. 7, an electronic device 700 according to an embodiment of the present disclosure includes a processor 701 that can perform various appropriate actions and processes according to a program stored in a Read Only Memory (ROM) 702 or a program loaded from a storage section 708 into a Random Access Memory (RAM) 703. The processor 701 may include, for example, a general purpose microprocessor (e.g., a CPU), an instruction set processor and/or an associated chipset and/or a special purpose microprocessor (e.g., an Application Specific Integrated Circuit (ASIC)), or the like. The processor 701 may also include on-board memory for caching purposes. The processor 701 may comprise a single processing unit or a plurality of processing units for performing different actions of the method flows according to embodiments of the disclosure.

In the RAM 703, various programs and data necessary for the operation of the electronic apparatus 700 are stored. The processor 701, the ROM 702, and the RAM 703 are connected to each other through a bus 704. The processor 701 performs various operations of the method flow according to the embodiments of the present disclosure by executing programs in the ROM 702 and/or the RAM 703. Note that the program may be stored in one or more memories other than the ROM 702 and the RAM 703. The processor 701 may also perform various operations of the method flow according to embodiments of the present disclosure by executing programs stored in the one or more memories.

According to an embodiment of the present disclosure, the electronic device 700 may further include an input/output (I/O) interface 705, the input/output (I/O) interface 705 also being connected to the bus 704. The electronic device 700 may also include one or more of the following components connected to the I/O interface 705: an input section 706 including a keyboard, a mouse, and the like; an output portion 707 including a Cathode Ray Tube (CRT), a Liquid Crystal Display (LCD), and the like, a speaker, and the like; a storage section 708 including a hard disk or the like; and a communication section 709 including a network interface card such as a LAN card, a modem, or the like. The communication section 709 performs communication processing via a network such as the internet. The drive 710 is also connected to the I/O interface 705 as needed. A removable medium 711 such as a magnetic disk, an optical disk, a magneto-optical disk, a semiconductor memory, or the like is mounted on the drive 710 as necessary, so that a computer program read therefrom is mounted into the storage section 708 as necessary.

The present disclosure also provides a computer-readable storage medium that may be embodied in the apparatus/device/system described in the above embodiments; or may exist alone without being assembled into the apparatus/device/system. The computer-readable storage medium carries one or more programs which, when executed, implement methods in accordance with embodiments of the present disclosure.

According to embodiments of the present disclosure, the computer-readable storage medium may be a non-volatile computer-readable storage medium, which may include, for example, but is not limited to: a portable computer diskette, a hard disk, a Random Access Memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing. In the context of this disclosure, a computer-readable storage medium may be any tangible medium that can contain, or store a program for use by or in connection with an instruction execution system, apparatus, or device. For example, according to embodiments of the present disclosure, the computer-readable storage medium may include ROM 702 and/or RAM 703 and/or one or more memories other than ROM 702 and RAM 703 described above.

Embodiments of the present disclosure also include a computer program product comprising a computer program containing program code for performing the methods shown in the flowcharts. The program code, when executed in a computer system, causes the computer system to perform the methods provided by embodiments of the present disclosure.

The above-described functions defined in the system/apparatus of the embodiments of the present disclosure are performed when the computer program is executed by the processor 701. The systems, apparatus, modules, units, etc. described above may be implemented by computer program modules according to embodiments of the disclosure.

In one embodiment, the computer program may be based on a tangible storage medium such as an optical storage device, a magnetic storage device, or the like. In another embodiment, the computer program may also be transmitted, distributed over a network medium in the form of signals, downloaded and installed via the communication section 709, and/or installed from the removable medium 711. The computer program may include program code that may be transmitted using any appropriate network medium, including but not limited to: wireless, wired, etc., or any suitable combination of the foregoing.

In such an embodiment, the computer program may be downloaded and installed from a network via the communication portion 709, and/or installed from the removable medium 711. The above-described functions defined in the system of the embodiments of the present disclosure are performed when the computer program is executed by the processor 701. The systems, devices, apparatus, modules, units, etc. described above may be implemented by computer program modules according to embodiments of the disclosure.

According to embodiments of the present disclosure, program code for performing computer programs provided by embodiments of the present disclosure may be written in any combination of one or more programming languages, and in particular, such computer programs may be implemented in high-level procedural and/or object-oriented programming languages, and/or assembly/machine languages. Programming languages include, but are not limited to, such as Java, c++, python, "C" or similar programming languages. The program code may execute entirely on the user's computing device, partly on the user's device, partly on a remote computing device, or entirely on the remote computing device or server. In the case of remote computing devices, the remote computing device may be connected to the user computing device through any kind of network, including a Local Area Network (LAN) or a Wide Area Network (WAN), or may be connected to an external computing device (e.g., connected via the Internet using an Internet service provider).

The flowcharts and block diagrams in the figures illustrate the architecture, functionality, and operation of possible implementations of systems, methods and computer program products according to various embodiments of the present disclosure. In this regard, each block in the flowchart or block diagrams may represent a module, segment, or portion of code, which comprises one or more executable instructions for implementing the specified logical function(s). It should also be noted that, in some alternative implementations, the functions noted in the block may occur out of the order noted in the figures. For example, two blocks shown in succession may, in fact, be executed substantially concurrently, or the blocks may sometimes be executed in the reverse order, depending upon the functionality involved. It will also be noted that each block of the block diagrams or flowchart illustration, and combinations of blocks in the block diagrams or flowchart illustration, can be implemented by special purpose hardware-based systems which perform the specified functions or acts, or combinations of special purpose hardware and computer instructions.

Those skilled in the art will appreciate that the features recited in the various embodiments of the disclosure and/or in the claims may be provided in a variety of combinations and/or combinations, even if such combinations or combinations are not explicitly recited in the disclosure. In particular, the features recited in the various embodiments of the present disclosure and/or the claims may be variously combined and/or combined without departing from the spirit and teachings of the present disclosure. All such combinations and/or combinations fall within the scope of the present disclosure.

The embodiments of the present disclosure are described above. However, these examples are for illustrative purposes only and are not intended to limit the scope of the present disclosure. Although the embodiments are described above separately, this does not mean that the measures in the embodiments cannot be used advantageously in combination. The scope of the disclosure is defined by the appended claims and equivalents thereof. Various alternatives and modifications can be made by those skilled in the art without departing from the scope of the disclosure, and such alternatives and modifications are intended to fall within the scope of the disclosure.

Claims (14)

1. A method for discriminating images in security inspection, comprising:

acquiring a first discrimination result obtained by discriminating a target object in a security inspection image by an artificial intelligent model;

Acquiring a second discrimination result obtained by discriminating the target object in the security inspection image by a target security inspection person;

when the first discrimination result is inconsistent with the second discrimination result, comparing the discrimination level of the artificial intelligent model on the similar articles of the target article with the discrimination level of the target security inspection personnel on the similar articles of the target article to obtain a comparison result; and

and determining a reliable judging result from the first judging result and the second judging result based on preset conditions met by the comparing result.

2. The method of claim 1, wherein the discrimination level comprises at least one of an identification rate, a false positive rate, or a false negative rate.

3. The method of claim 2, wherein the discrimination level of the artificial intelligence model is above a preset reference level.

4. The method of claim 1, wherein the selecting a trusted discrimination result from the first discrimination result and the second discrimination result based on the preset condition satisfied by the comparison result comprises:

and when the comparison result shows that the discrimination level of the target security inspection personnel on the similar objects of the target object is higher than that of the artificial intelligent model, determining that the second discrimination result is a credible discrimination result.

5. The method of claim 1, wherein the selecting a trusted discrimination result from the first discrimination result and the second discrimination result based on the preset condition satisfied by the comparison result comprises:

and when the comparison result shows that the discrimination level of the target security inspection personnel on the similar articles of the target articles is lower than the discrimination level of the artificial intelligent model and the difference is greater than or equal to a preset threshold value, determining that the first discrimination result is a reliable discrimination result.

6. The method of claim 5, wherein the method further comprises:

the threshold is set based on historical security performance data of the target security personnel, wherein the historical security performance data comprises at least one or more of the discrimination level of the target security personnel on the similar objects of the target object, the duration of the target security personnel in working security work or the experience rating data of the target security personnel in working security work.

7. The method of claim 6, wherein after setting the threshold, the method further comprises:

and adjusting the threshold value based on the on-site security inspection condition data when the second judging result is obtained, wherein the on-site security inspection condition data comprises the duration of continuous security inspection judgment by the target security inspection personnel or the congestion degree data of the articles subjected to on-site security inspection judgment at the moment when the second judging result is obtained.

8. The method of claim 5, wherein after the determining that the first discrimination result is a trusted discrimination result, the method further comprises:

and sending notification reminding information to the target terminal so as to remind a user of the target terminal to re-judge the target object.

9. The method according to any one of claims 1-8, wherein the method further comprises:

acquiring historical judgment chart data obtained by judging various articles by at least one security personnel in the process of judging the historical security images; and

obtaining the judging level of each security personnel in the at least one security personnel on the various objects according to the historical judging graph data;

wherein, the liquid crystal display device comprises a liquid crystal display device,

the at least one security personnel comprises the target security personnel, and the various articles comprise the similar articles of the target articles.

10. An apparatus for discriminating an image in security inspection, comprising:

the first acquisition module is used for acquiring a first discrimination result obtained by discriminating the target object in the security inspection image by the artificial intelligence model and acquiring a second discrimination result obtained by discriminating the target object in the security inspection image by a target security inspection person;

The comparison module is used for comparing the discrimination level of the artificial intelligent model on the similar articles of the target article with the discrimination level of the target security inspection personnel on the similar articles of the target article when the first discrimination result and the second discrimination result are inconsistent, so as to obtain a comparison result; and

and the determining module is used for determining a reliable judging result from the first judging result and the second judging result based on preset conditions met by the comparing result.

11. The apparatus of claim 10, wherein the apparatus further comprises:

the second acquisition module is used for acquiring historical judgment chart data obtained by judging various articles in the process of judging the historical security image by at least one security personnel; and

the evaluation module is used for obtaining the judging level of each security personnel in the at least one security personnel on the various objects according to the historical judging graph data;

wherein, the liquid crystal display device comprises a liquid crystal display device,

the at least one security personnel comprises the target security personnel, and the various articles comprise the similar articles of the target articles.

12. An electronic device, comprising:

one or more processors;

A memory for storing one or more programs,

wherein the one or more programs, when executed by the one or more processors, cause the one or more processors to perform the method of any of claims 1-9.

13. A computer readable storage medium having stored thereon executable instructions which, when executed by a processor, cause the processor to perform the method according to any of claims 1 to 9.

14. A computer program product comprising a computer program which, when executed by a processor, implements the method according to any one of claims 1 to 9.

Images