Automated trading equipment and programs

Abstract

This system provides a mechanism to prevent bank transfer fraud using automated trading devices. [Solution] The information processing device 200 of the ATM 100, which performs automatic transactions in response to user operations, comprises an image acquisition unit 230 that acquires an image of an approaching user, an action detection unit 240 that analyzes the user's image and detects the user's call-related actions, and a display control unit 250 that changes the intensity level of an alert displayed on a display device according to the detection time of the call-related action.

Description

Technical Field

[0001] The present invention relates to an automatic transaction device and a program.

Background Art

[0002] In recent years, many ATMs (Automatic Teller Machines) operate 24 hours a day and are widely installed in many places other than financial institutions (for example, retail stores such as convenience stores and public facilities). ATMs have evolved to handle transactions other than ordinary deposits and various cashless transactions (services), and in recent years, they are also called automatic transaction devices.

[0003] Thus, while ATMs enable various transactions and have become more convenient, crimes such as so-called "transfer fraud" have become rampant by taking advantage of the fact that ATMs are unmanned, posing a social problem (see, for example, Patent Document 1).

Prior Art Documents

Patent Documents

[0004]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0005] As a typical example of "transfer fraud", there is a case where a malicious third party instructs a victim to operate an ATM using a user terminal such as a mobile phone and causes the victim to transfer money to the third party's account. The malicious third party gives skillful instructions as if transferring a refund or the like to the victim's account, but actually makes the victim perform a transfer procedure from the victim's account to the third party's account. At present, there is still no preventive measure that can effectively prevent such transfer fraud.

[0006] This invention was made in view of the circumstances described above, and aims to provide a mechanism that can prevent bank transfer fraud using automated trading devices. [Means for solving the problem]

[0007] An automated trading device according to one embodiment of the present disclosure is an automated trading device that performs automated trading in response to user operations, and is characterized by comprising: an image acquisition unit that acquires an image of an approaching user; an action detection unit that analyzes the user's image and detects the user's call-related behavior; and a display control unit that changes the intensity level of an alert displayed on a display device according to the detection time of the call-related behavior. [Effects of the Invention]

[0008] According to the present invention, it is possible to prevent bank transfer fraud using automated trading devices. [Brief explanation of the drawing]

[0009] [Figure 1] This figure shows the appearance of ATM100 according to one embodiment of the present invention. [Figure 2] This diagram shows the main hardware configuration of the information processing device 200. [Figure 3] This block diagram shows the functional configuration realized by ATM100. [Figure 4] This diagram illustrates the contents of the user management database DB1. [Figure 5] This is an explanatory diagram of the behavior recognition operation by the behavior detection unit 240. [Figure 6A] This is an explanatory diagram of the behavior recognition operation by the behavior detection unit 240. [Figure 6B] This is an explanatory diagram of the behavior recognition operation by the behavior detection unit 240. [Figure 7] This diagram illustrates how alerts are displayed. [Figure 8A] This diagram illustrates how alerts are displayed. [Figure 8B]A diagram illustrating the display mode of an alert. [Figure 8C] A diagram illustrating the display mode of an alert. [Figure 8D] A diagram illustrating the display mode of an alert. [Figure 9] A flowchart illustrating the prevention process of transfer fraud.

Embodiments for Carrying Out the Invention

[0010] Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings. The same elements are denoted by the same reference numerals, and redundant descriptions are omitted.

[0011] A. This Embodiment FIG. 1 is a diagram showing the appearance of an ATM 100 as an automatic transaction device according to an embodiment of the present invention. The ATM 100 can be installed in various places such as banks, post offices, convenience stores, retail stores, airports, and stations.

[0012] The ATM 100 includes a main display unit 101, a card reading unit 102, a bill deposit / withdrawal unit 103 for performing bill deposit and withdrawal, an operation unit 104 operated by a user, a code reading unit 105, an optical scanning unit 106, a pedestal 107, a sub-display unit 108, a non-contact IC communication unit 109, and a camera 110.

[0013] The ATM 100 according to this embodiment analyzes the actions of the user acquired by the camera 110 to detect whether the user is taking actions that may result in being victimized by transfer fraud (hereinafter, premonitory actions of transfer fraud). When it is detected that the user is taking premonitory actions of transfer fraud, the intensity level of the alert displayed on the transaction screen of the ATM 100 is changed according to the detection time of the detected premonitory actions. Thereby, transfer fraud is prevented effectively and in advance (details will be described later).

[0014] The main display unit 101 has, for example, display means such as a liquid crystal display, and input means such as a touch panel for receiving input from a user. The display means displays a screen or input keys for guiding the user's operation, and by the user pressing the input keys from above the touch panel, information corresponding to the input keys can be input. Instead of the input means displayed on the main display unit 101, the user can also input the same information by operating the input means (such as a numeric keypad) of the operation unit 104.

[0015] The card reading unit 102 can read card information recorded on a cash card (including a debit card) or the like. The cash card may be of either an IC card type or a magnetic card type. The code reading unit 105 is, for example, a code reader, and can read a two-dimensional code or the like displayed on a user terminal held directly below the code reading unit 105. The card information and the code information include an identification ID for identifying the user and account identification information (described later) for identifying the transaction account.

[0016] The optical scanning unit 106 is, for example, an image scanner, and can scan a photo or characters attached to the face of a cash card, an IC card, or the like placed on the mounting table 107 as an image. The mounting table 107 has a surface that is substantially horizontal with respect to the ground, and the optical scanning unit 106 can obtain an image of a certain quality without being affected by hand shaking or the like.

[0017] The sub-display unit 108 has, for example, display means such as a liquid crystal display, and is installed on the surface of the mounting table 107. The sub-display unit 108 may have only the display means, or may have input means in addition to the display means, similar to the main display unit 101. For example, the sub-display unit 108 may display a screen for guiding the position where a cash card, an IC card, or the like read by the optical scanning unit 106 is placed. Thereby, even a user who is not familiar with the operation of the ATM 100 can smoothly perform the operation without getting lost.

[0018] The contactless IC communication unit 109 is, for example, a contactless IC reader / writer and is installed inside the base 107. The contactless IC communication unit 109 can communicate with a contactless IC chip mounted on an IC card or the like, and can read and write information stored on the contactless IC chip. The contactless IC communication unit 109 can communicate with the contactless IC chip when an IC card or the like is placed on the base 107.

[0019] Camera 110 is a high-definition camera device capable of capturing video or still images. Camera 110 is installed to capture images of the user's face and actions when operating the ATM 100. More specifically, the camera is installed in a position that allows for reliable confirmation of whether the user is making a transfer (transaction) while talking on a mobile device. The installation location and number of cameras 110 can be set arbitrarily. Camera 110 starts recording when a predetermined operation is performed by the user using the ATM 100.

[0020] For example, the camera 110 may start taking pictures when a cash card is inserted into the card reader 102, or when a two-dimensional code is read by the code reader 105. Furthermore, if the ATM 100 is equipped with a human presence sensor (not shown), the camera 110 may start taking pictures when the human presence sensor detects a user. The conditions under which photography begins can be arbitrarily set and changed.

[0021] Figure 2 is a diagram showing the main hardware configuration of the information processing unit (computer) 200 installed in the ATM 100, and Figure 3 is a block diagram showing the functional configuration realized by the ATM 100. As shown in Figure 2, the information processing device 200 is composed of a control device 201 equipped with a CPU, etc., a storage device 202 equipped with semiconductor memory, etc., an input device 203 consisting of a touch panel, etc., and a communication device 204 equipped with various communication interfaces, etc., similar to the hardware of a typical computer.

[0022] The information processing device 200 realizes functions such as the storage unit 210, transaction detection unit 220, image acquisition unit 230, behavior detection unit 240, and display control unit 250 shown in Figure 3, through the cooperation of hardware resources such as the control device 201 with various software stored in the storage device 202.

[0023] The memory unit 210 stores user images acquired by the camera 110, various application programs, databases, and the like.

[0024] Figure 4 is an example of the contents of the user management database DB1 stored in the storage unit 210. Various types of information registered in the user management database DB1 may be obtained from, for example, a management server (not shown) that maintains and manages the ATM 100.

[0025] The user management database DB1 stores account identification information and personal information for each user, linked together. Account identification information includes, for example, some or all of the following: financial institution name, branch name, deposit type, account number, account holder name, PIN, etc. Personal information includes, for example, attribute information such as the user's age, gender, and date of birth, as well as basic information such as name, address, telephone number, email address, and SNS account name, and also some or all of the history information showing the user's past use of ATM 100 (for example, the name of the branch where ATM 100 is installed, location information, time information, transaction details, etc.). Personal information also includes user images used when each user conducts transactions (for example, a front view or side view of the user, etc.). The control device 201 of the information processing device 200 performs new registration and updates (including correction and deletion) of personal information in the user management database DB1.

[0026] The transaction detection unit 220 detects the start of a transfer by a user when a cash card is inserted into the card reader 102 or when a two-dimensional code is read by the code reader 105. When the transaction detection unit 220 detects the start of a transfer by a user, it outputs a detection signal to the image acquisition unit 230. When the card reader 102 or code reader 105 reads the account identification information contained in the card information of the cash card or the account identification information contained in the two-dimensional code, it stores the user images sequentially acquired by the camera 110 in the storage unit 210 in association with the read account identification information.

[0027] The image acquisition unit 230, triggered by a detection signal sent from the transaction detection unit 220, starts taking pictures of the user with the camera 110 and acquires the user's image.

[0028] The behavior detection unit 240 analyzes user images using various AI tools (for example, original AI tools developed in-house) to detect whether a user approaching the ATM 100 is exhibiting behavior that may lead to becoming a victim of wire fraud (i.e., behavior that foreshadows wire fraud). The AI ​​tools installed in the behavior detection unit 240 include a keypoint extraction AI tool 241 that infers the user's key points (in this embodiment, the shape of the face and fingers, and landmark coordinates of the skeleton), and a behavior recognition AI tool 242 that infers the user's behavior using the key points as features.

[0029] Figures 5, 6A, and 6B are explanatory diagrams illustrating the behavior recognition operation by the behavior detection unit 240. As shown in Figure 5, when the action detection unit 240 receives one frame of user image input from the image acquisition unit 230, it uses a keypoint extraction AI tool 241 or the like to estimate the coordinates of three landmarks (i.e., the user's key points): the user's face (expression), hands (fingers), and posture (skeleton).

[0030] Specifically, the system estimates the landmark coordinates of the user's face using a face landmark task M1 to detect the landmark coordinates and facial expressions of the user's face, a hand landmark task M2 to detect the landmark coordinates of the user's hands, and a posture landmark task M3 to detect the landmark coordinates of the user's posture. Next, the behavior detection unit 240 uses a behavior recognition AI tool 242 to infer (detect) the user's actions based on the estimated landmark coordinates of the user. The behavior detection unit 240 repeatedly executes the series of processes described above for a predetermined number of frames. As the behavior recognition AI tool, a trained model for inferring user actions is used. User actions detected by the behavior recognition AI tool include various actions such as the user making a call using a user terminal such as a smartphone (hereinafter referred to as call-related actions) and operating the ATM 100. For the sake of explanation, actions other than call-related actions will be referred to as non-call-related actions below.

[0031] The operation of the behavior detection unit 240 will be explained in detail with reference to Figures 6A and 6B. For example, if the behavior detection unit 240 detects a call-related action by a user using a user terminal such as a smartphone for X consecutive seconds, it will determine that the user's actions are a precursor to wire fraud (see "Detection of Precursor Action to Wire Fraud" in Figure 6A). In this case, the display control unit 250 will display an alert on the transaction screen.

[0032] On the other hand, if the behavior detection unit 240 detects, for example, a non-call-related behavior (such as inputting data into the ATM 100) for Y seconds consecutively, it determines that the user's behavior is not a precursor to wire fraud (see "No Precursor Behavior to Wire Fraud Detected" in Figure 6B). In this case, the display control unit 250 does not display an alert on the transaction screen.

[0033] The display control unit 250 receives the detection result from the action detection unit 240 and displays (outputs) a message based on the received detection result on the main display unit 101. As described above, when the display control unit 250 detects a call-related action, it displays a cautionary alert on the trading screen. In this embodiment, the display control unit 250 changes the intensity level of the alert displayed on the main display unit 101 (i.e., the intensity of the warning) according to the detection time of the call-related action.

[0034] Figures 7 and 8A to 8D illustrate the various ways in which alerts are displayed. When the display control unit 250 detects a call-related action, it displays an alert "At" at the top of the trading screen P1, as shown in Figure 7, to indicate a trading warning. In this embodiment, as shown in Figures 8A to 8C, three types of alerts with different intensity levels are displayed on the main display unit 101, depending on the detection time of the call-related action.

[0035] More specifically, an alert is displayed on the main display unit 101 in which the intensity level of the alert increases as the detection time of call-related activity lengthens. The detection time of call-related activity may be the total time, and for example, the relationship between the intensity level and the detection time can be set as follows. The following relationship is set based on the applicant's research on wire fraud (reference information), but it is merely an example, and the detection time and the number of intensity levels P (≧2) to be set are arbitrary. In this embodiment, an intensity level determination table Tb showing the relationship between the intensity level and the detection time is stored in the storage unit 210. (1) Intensity level "weak" ... for 30 seconds after detection starts (2) Intensity level "medium" ... from 30 seconds after detection starts to 1 minute (3) Intensity level "Strong" ... From 1 minute after detection begins

[0036] <Reference information> Average transaction time for bank transfers: approximately 70 seconds Transaction time for making a bank transfer while on the phone: Approximately 8-10 minutes

[0037] As shown in Figures 8A to 8C, if the alert intensity level is "weak," a general warning message (such as "Caution! 'Call from the ATM' is a scam") may be displayed as Alert At1. If the alert intensity level is "medium," a more strongly warning message ("Scams are rampant! Is that phone call safe?") may be displayed as Alert At2. If the alert intensity level is "strong," an even more strongly warning message ("Scams are rampant! Is this transfer safe?") may be displayed as Alert At3. Note that the messages for each Alert At shown in Figures 8A to 8C are merely examples, and various messages can be used. For example, depending on the alert intensity level, the following messages may be used. • When the alert intensity level is "weak" → "Caution! 'Call me in front of the ATM' is a scam." • When the alert intensity level is "medium" → "Being instructed to operate an ATM over the phone is a scam." • If the alert intensity level is "Strong" → "Have you been instructed to make a bank transfer over the phone? Please hang up and speak to us via intercom."

[0038] Furthermore, to enhance the attention given by the alert display, animations such as the ripple effect Ew may be added during the alert display, for example, as shown in Figure 8D. Also, when displaying each alert, a color scheme that indicates caution (for example, yellow x black) may be used. Alternatively, the color scheme of each alert may be changed for all (or some) intensity levels, or the color scheme of each alert may be the same for all intensity levels.

[0039] Furthermore, the method of highlighting each alert may be arbitrarily set according to the intensity level. For example, the stronger the alert intensity level, the larger the display size or the more prominent the display position may be. In addition to being displayed on the main display unit 101, it may also be displayed on the sub-display unit 108 or other locations. Furthermore, depending on the alert intensity level, it may be output as an audio message from a speaker (not shown). For example, if the alert intensity level is "strong," an audio message may be output along with a display message to draw attention to trading, or a display method may be adopted in which a text box to draw attention to trading pops out.

[0040] As described above, according to the ATM 100 of this embodiment, by analyzing the user's behavior acquired by the camera 110, it is possible to detect whether the user is engaging in behavior that may indicate a wire fraud (in this embodiment, call-related behavior). If it is detected that the user is engaging in such behavior, the intensity level of the alert displayed on the transaction screen of the ATM 100 is changed according to the detection time of the detected behavior. This makes it possible to prevent wire fraud proactively and effectively.

[0041] The following explains the fraud prevention process performed by ATM100, with reference to the diagrams.

[0042] Figure 9 is a flowchart illustrating the process for preventing wire fraud. The user inserts their cash card into the ATM 100, enters their PIN and other information, and then begins the transaction. The transaction detection unit 220 detects the user's commencement of a transaction when the cash card is inserted into the card reader 102 (step S1). Upon detecting the user's commencement of a transaction, the transaction detection unit 220 outputs a detection signal to the image acquisition unit 230. Based on the detection signal sent from the transaction detection unit 220, the image acquisition unit 230 starts taking pictures of the user with the camera 110 and acquires the user's image (step S2).

[0043] When user image acquisition begins, the behavior detection unit 240 starts recognizing (detecting) the user's behavior (step S3). The behavior detection unit 240 uses a keypoint extraction AI tool and a behavior recognition AI tool to detect whether the user's behavior is related to a call or not (step S4). The method by which the behavior detection unit 240 recognizes the user's behavior has already been described in detail, so it will not be described here.

[0044] <If call-related behavior is detected> When the behavior detection unit 240 detects a call-related action by the user, it sends an instruction to the display control unit 250 to display an alert on the trading screen.

[0045] When the display control unit 250 receives instructions from the behavior detection unit 240, it uses a timer (not shown) or the like to determine the detection time of the call-related behavior (step S5), and then refers to the intensity level determination table Tb stored in the memory unit 210 to determine the intensity level of the alert (step S6). The display control unit 250 displays a warning alert on the trading screen according to the determined alert intensity level (step S7).

[0046] The display control unit 250 determines whether the transaction has been completed based on the operation information entered by the user (step S8). If it determines that the transaction has not been completed (step S8; NO), it returns to step S4 and continues processing. If the display control unit 250 determines that the transaction operation (in this case, a predetermined transfer) has been completed while processing is continuing (step S8; YES), it terminates the series of processes described above.

[0047] <If no call-related activity is detected> If the behavior detection unit 240 does not detect any call-related behavior in step S4, i.e., if it detects a non-call-related behavior, it sends an instruction to the display control unit 250 to display a normal trading screen (i.e., a trading screen without alerts).

[0048] The display control unit 250, following instructions from the behavior detection unit 240, displays a transaction screen without alerts (step S9), and then determines whether the transaction has been completed based on the operation information entered by the user (step S8). If it determines that the transaction has not been completed (step S8; NO), it returns to step S4 and continues processing. If the display control unit 250 determines that the transaction operation (in this case, a predetermined transfer) has been completed while processing is continuing (step S8; YES), it terminates the series of processes described above.

[0049] B. Others It should be noted that the present invention is not limited to the embodiments described above, and can be implemented in various other forms without departing from the spirit of the invention. For this reason, the above embodiments are merely illustrative in all respects and should not be interpreted restrictively. For example, the order of each processing step described above can be arbitrarily changed or executed in parallel, as long as there is no inconsistency in the processing content. Also, in this embodiment, call-related behavior was given as an example of a precursor to wire fraud, but this is not the only example. For example, the behavior of a user operating the ATM 100 while looking at a memo or the like may be detected as a precursor to wire fraud. Also, in this embodiment, the intensity level of the alert was changed according to the detection time of the call-related behavior, but this is not the only example. For example, depending on the detection time of the call-related behavior, the control unit of the ATM 100 may forcibly stop the transaction with the user (e.g., a transfer to a designated account) or notify a call center (not shown). As an example, if call-related behavior is detected for X seconds or more, the transaction with the user may be forcibly stopped or a call center may be notified. This will make it possible to more reliably and effectively prevent potentially fraudulent activities such as bank transfer scams. [Explanation of Symbols]

[0050] 100…ATM, 200…Information Processing Device, 201…Control Device, 202…Storage Device, 203…Input Device, 204…Communication Device, 210…Storage Unit, 220…Transaction Detection Unit, 230…Image Acquisition Unit, 240…Action Detection Unit, 241…Keypoint Extraction AI Tool, 242…Action Recognition AI Tool, 250…Display Control Unit, DB1…User Management Database, Tb…Intensity Level Determination Table, M1…Facial Landmark Task, M2…Hand Landmark Task, M3…Posture Landmark Task, At, At1, At2, At3, At4…Alerts, P1…Transaction Screen

Claims

1. An automated trading device that performs automated trading in response to user operations, An image acquisition unit that acquires an image of the approaching user, An action detection unit analyzes the user's image, detects the user's call-related behavior, and determines whether the time during which the call-related behavior is detected is continuous for a predetermined period of time or longer. Based on the above determination, a display control unit displays an alert on the display device if the call-related behavior continues for a predetermined period of time or longer, and increases the intensity level of the alert the longer the detection time of the call-related behavior. An automated trading device equipped with the following features.

2. The aforementioned alert includes a message to draw attention to the transaction, The automated trading device according to claim 1, wherein the display control unit changes the content of the message according to the intensity level.

3. The automated trading device according to claim 1 or 2, wherein the display control unit changes at least one of the display position, size, and color of the alert according to the intensity level.

4. The automated trading device according to claim 1 or 2, wherein the detection time is the total time of the detected call-related actions.

5. The automated trading device according to claim 1 or 2, wherein the behavior detection unit estimates the user's landmark coordinates using a first AI tool and detects the user's call-related behavior from the user's landmark coordinates using a second AI tool.

6. A computer that performs automated trading in response to user input, An image acquisition unit that acquires an image of the approaching user, An action detection unit analyzes the user's image, detects the user's call-related behavior, and determines whether the time during which the call-related behavior is detected is continuous for a predetermined period of time or longer. A program to function as a display control unit that, as a result of the above determination, displays an alert on the display device if the call-related behavior continues for a predetermined period of time or longer, and increases the intensity level of the alert the longer the detection time of the call-related behavior.

Images