JP7761412B2 - Camera installation support system, camera control device, and camera installation method - Google Patents

Description

The present invention relates to technology that supports the installation of cameras to capture self-service refueling at gas stations.

In conventional self-service gas stations, dangerous behavior at the station is detected visually by a person (employee). Gas stations that support self-service gas stations are equipped with confirmation cameras that allow employees to confirm whether or not to issue permission for refueling. These confirmation cameras are typically installed diagonally above the vehicle's front or rear relative to the weighing machine, and capture footage of the refueling operation viewed diagonally from the front or rear of the vehicle. Figure 1 shows an example of the installation location of conventional cameras at a gas station. In the example of Figure 1, camera 10 is installed diagonally above the vehicle's rear relative to weighing machine 20.

Here, the following are examples of prior art in the technical field related to the present invention. For example, Patent Document 1 discloses an invention in which information about each refueling point is associated with an image of each refueling point displayed on a monitoring monitor. Furthermore, Patent Document 2 discloses an invention in which a monitoring device determines abnormal user behavior and controls vehicle refueling through user operation.

Japanese Patent Application Laid-Open No. 2000-238899 International Publication No. 2021/015256

In recent years, approaches that combine image processing and AI to detect dangerous behavior have been explored. However, when capturing images of a weighing machine from an oblique angle, as in Figure 1, blind spots can occur and the expected detection results may not be obtained. Furthermore, because the layout of each gas station varies, it is difficult to commonly define camera position, height, angle, field of view, etc. As a result, camera installation accuracy is based on the experience of the construction worker, and camera installation positions and angle of view are not uniform. Furthermore, even if there is an ideal angle of view and camera installation is strictly defined, it is extremely difficult for workers to faithfully install cameras on site in accordance with the definitions, given that different workers install cameras each time for different gas stations with different layouts.

The present invention was made in consideration of the above-mentioned conventional circumstances, and aims to make it easier to adjust the camera to a field of view suitable for monitoring self-service refueling at gas stations.

In order to achieve the above object, a camera installation support system according to one aspect of the present invention is configured as follows.
In other words, the camera installation support system of the present invention comprises a camera installed directly above a weighing machine at a gas station or at a position shifted from directly above the weighing machine toward the vehicle parking position, and a camera control device having the function of controlling the pan, tilt, and zoom of the camera, and the camera control device is characterized in that it uses guide information that specifies the position and size of the weighing machine to perform a process of controlling the pan and tilt of the camera so that the center coordinates of the weighing machine included in the image captured by the camera match the center coordinates of the weighing machine included in the guide information, and a process of controlling the zoom of the camera so that the size of the weighing machine included in the image captured by the camera matches the size of the weighing machine included in the guide information.

Here, the camera control device can be configured to execute a process to detect weighing machines from the video captured by the camera using a trained model for detecting weighing machines contained in the video.

The camera control device can also be configured to execute a process to output an image in which an image indicating the position and size of the weighing machine included in the guide information is superimposed on the image captured by the camera.

A camera control device according to one aspect of the present invention is configured as follows.
In other words, the camera control device of the present invention is a camera control device having the function of controlling the pan, tilt, and zoom of a camera installed at a gas station, and the camera is installed at a position directly above a weighing machine at the gas station or at a position shifted from directly above the weighing machine in the direction of the vehicle parking position, and the camera control device is characterized in that it uses guide information that specifies the position and size of the weighing machine to perform a process of controlling the pan and tilt of the camera so that the center coordinates of the weighing machine included in the image captured by the camera match the center coordinates of the weighing machine included in the guide information, and a process of controlling the zoom of the camera so that the size of the weighing machine included in the image captured by the camera matches the size of the weighing machine included in the guide information.

A camera installation method according to one aspect of the present invention is configured as follows.
In other words, the camera installation method of the present invention is characterized by having an installation step of installing a camera at a position directly above a weighing machine at a gas station or at a position shifted from directly above the weighing machine toward the vehicle stopping position, and an adjustment step of adjusting the field of view of the camera using an adjustment image including a guide that defines the position and size of the weighing machine within the image.

Here, the adjustment process may be configured to include a process of detecting weighing machines from camera footage using a trained model for detecting weighing machines from camera footage.

The adjustment process can also be configured to include a process of controlling the pan and tilt of the camera so that the center coordinates of the weighing device included in the image captured by the camera match the center coordinates of the guide in the adjustment image, and a process of controlling the zoom of the camera so that the size of the weighing device included in the image captured by the camera matches the size of the guide in the adjustment image.

The adjustment process may also be configured to include a process of superimposing an adjustment image on the image captured by the camera.

This invention makes it easy to adjust the camera to a field of view suitable for monitoring self-service refueling at gas stations.

FIG. 1 is a front view showing an example of the installation position of a conventional camera. FIG. 10 is a front view showing an example of a camera installation position suitable for monitoring self-service refueling stations. FIG. 10 is a plan view showing an example of a camera installation position suitable for monitoring self-service gas stations. FIG. 10 is a side view showing an example of a camera installation position suitable for monitoring self-service refueling stations. 5A to 5C are diagrams showing examples of the field of view of the cameras installed at the positions shown in FIGS. 2 to 4. 1 is a diagram illustrating an example of the configuration of a camera installation support system according to an embodiment of the present invention. 7A and 7B are diagrams showing examples of adjustment images used in the camera installation support system of FIG. 6; 7A and 7B are diagrams showing examples of adjustment images used in the camera installation support system of FIG. 6; FIG. 7 is a diagram showing an example of a flowchart relating to automatic adjustment of the field of view range of the camera performed by the camera installation support system of FIG. 6 .

Hereinafter, an embodiment of the present invention will be described with reference to the drawings.
First, we will explain the installation location of a camera suitable for monitoring self-service refueling at a gas station. Camera position is an important factor when detecting dangerous behavior using image processing AI. However, since the purpose of conventional confirmation cameras at gas stations is for employees to roughly check the situation visually, camera position and field of view are not particularly important. As a result, camera positions and field of view are often inappropriate for detecting dangerous behavior using image processing AI. Furthermore, to maximize the effectiveness of image processing AI, precise adjustment of camera position, height, field of view, etc. is required. However, because the layout of gas stations varies from store to store, camera position, height, and field of view are not commonly defined. For this reason, camera installation based on the experience of construction workers will not achieve uniform detection accuracy with image processing AI. Therefore, to determine the optimal camera installation location, we analyzed three factors: the vehicle stopping position, the posture of the refueler, and the behavior of the refueler during self-service refueling.

(1) Vehicle Parking Position Although it depends on the personal judgment of the person who visits the gas station to fill up the tank, vehicles are naturally drawn to the rectangular vehicle parking positions marked on the road surface of the gas station, so vehicles are generally parked so that the fuel filler opening is directly in front of the metering machine. In addition, the distance between the metering machine and the vehicle's fuel filler opening is generally kept long enough for the fuel hose to reach, and long enough so that there is no interference with the refueling operation and people can enter and exit smoothly.

(2) Posture of the Fueler The fueler generally holds the fuel nozzle with one hand and stands next to the vehicle while filling the tank. This position is generally maintained until the fuel is filled. In conventional camera positions that capture the refueling operation from a diagonal downward view from the front or rear of the vehicle, the fuel nozzle may be hidden by the hand or arm holding the nozzle, making the fuel nozzle invisible. In rare cases, the fueler may hold the fuel nozzle with both hands, but the situation is similar.

The person filling up the tank is often standing at the front or rear of the vehicle relative to the fuel nozzle. For example, if the camera is positioned to film from the rear of the vehicle and the person filling up the tank is standing at the rear of the vehicle, the fuel filler opening and fuel nozzle will be hidden by the person filling up the tank, making it impossible to see the filling action at all. Also, for example, there are cases where the vehicle is filled up with the door open, and if the camera is positioned at the front of the vehicle, the filling action will be hidden by the door and will not be visible at all.

(3) Behavior of the Fueler When refueling, the fueler must hold the fuel nozzle in his/her hand, so he/she must maintain a standing posture, and as a result, the fueler rarely moves significantly while refueling. However, when refilling fuel, the fueler may look into the fuel filler neck while refueling. In this case, the fueler's head or upper body may cover the fuel filler neck or fuel filler nozzle, hiding them and making them impossible to capture on the camera.

The vehicle's fuel filler cap is located on the side of the vehicle, and the fuel nozzle is inserted perpendicular to the vehicle. Therefore, when adding fuel, you must pull the fuel nozzle back slightly and look into the fuel filler cap from either the side or the front. Due to the vehicle's structure, it is not possible to see the inside of the fuel tank from directly above the fuel filler cap, so it is difficult to look into the fuel filler cap from directly above.

The above analysis results show that although there are various detailed refueling actions taken by refueling drivers, the layout of gas stations and the shape of vehicles are patterned, so the camera's field of view can be fixed to a certain extent. Therefore, by patterning the camera position, height, and angle for detecting refueling actions using image processing AI, it becomes possible to standardize them for various gas station layouts, making it possible to detect dangerous behavior with high accuracy.

Next, we will analyze the number of cameras. The pattern of camera positioning is also influenced by the number of cameras. For example, when taking pictures at an angle looking down from the front or rear of the vehicle, as is done conventionally, one possible solution is to install two cameras per lane and take pictures from two directions as a countermeasure against the fuel pumper hiding behind the vehicle or behind a door. In this case, the number of cameras installed increases, which increases the costs of equipment and construction. Furthermore, since it is necessary to capture images from both cameras and have the image processing AI perform inference processing, the image processing AI becomes cumbersome and complex. Furthermore, since it is necessary to process images from two cameras simultaneously, resource consumption is approximately twice that of a single camera. As a result, a high-performance model must be used as the computer to perform the AI inference processing, which also increases costs.

On the other hand, confirmation cameras already installed at gas stations sometimes capture two lanes with a single camera. In this case, each lane is captured on the left and right sides of the camera's field of view. When these two lanes are captured simultaneously, the number of effective pixels per lane is at least half of the original number of pixels, halving the amount of information. Furthermore, when using image processing AI to monitor refueling operations, the camera's field of view includes a significant amount of parts that are not related to refueling operations (parts that are not necessary for monitoring) because the two lanes are captured simultaneously. As a result, the camera's resolution cannot be used effectively, and the camera is not used efficiently. At the same time, this can also lead to false alarms.

These analysis results show that the number of cameras should be limited to one per lane, and that the field of view should be set to include as many useful areas as possible, while excluding areas that are unnecessary for the image processing AI. Furthermore, when simultaneously capturing images of two lanes with one camera, the original concerns may be alleviated by using a high-resolution camera, such as a 4K camera. In practice, either type of operation is acceptable, as long as the images can be captured with consistent and consistent accuracy.

Next, we will analyze fixing the camera installation position. From the analysis so far, the appropriate camera installation position when monitoring refueling operations using image processing AI can be narrowed down to a certain extent. In this embodiment, the standard is a position that allows continuous monitoring of the entire refueling operation, regardless of the movements of the person refueling, as much as possible.

(1) Reference field of view The target reference field of view is centered on the weighing machine. For example, the field of view should be such that the weighing machine is positioned at the top center of the captured image. Analysis of the vehicle stopping position reveals that the location where the refueling operation takes place is the center of the weighing machine, so the position should be such that this part can be captured as large as possible. This improves the detection accuracy of the image processing AI. It also reduces the possibility that the refueling operation will be cut off from the field of view.

(2) Camera Installation Position In order to center the field of view on the weighing machine, the camera installation position is based on the positions shown in Figures 2 to 5. Figure 2 is a front view showing an example of the camera installation position. Figure 3 is a plan view showing an example of the camera installation position. Figure 4 is a side view showing an example of the camera installation position.

As shown in these figures, the camera 10 is ideally installed along axis X, which passes through the center of the weighing machine 20 and extends in the direction of the vehicle stopping position 40. Because gas stations always have a canopy (roof), the camera is generally mounted on the ceiling. In the example of Figure 3, two vehicle stopping positions 40A, 40B are set on either side of the weighing machine 20, and two cameras 10A, 10B are installed corresponding to these positions.

If the camera is installed directly above the weighing machine, there is a possibility that the fuel filler cap will be obscured by the person filling the tank. On the other hand, if the camera is installed too far away from directly above the weighing machine, the fuel filler cap will be obscured by vehicles. Therefore, as shown in Figures 3 and 4, it is preferable to position the camera a certain distance (for example, about 1.5 meters) from directly above the weighing machine 20 toward the vehicle stopping position 40. However, if one camera is to simultaneously photograph two lanes, the camera may be installed directly above the weighing machine 20, as shown by the dashed line 10' in Figure 3.

As described above, the camera is installed directly above the weighing machine (when one camera is used to capture two lanes) or at a position shifted from directly above the weighing machine toward the vehicle's stopping position (when one camera is used to capture one lane). Here, "directly above the weighing machine" in this specification does not only mean vertically above the center of the weighing machine, but also includes the area above the weighing machine that corresponds to the installation range of the weighing machine when viewed from above. Therefore, "a position directly above the weighing machine or a position shifted from directly above the weighing machine toward the vehicle's stopping position" includes the area above the weighing machine that corresponds to the width W of the weighing machine as seen from the vehicle (see Figure 3). However, it goes without saying that it is preferable to install the camera closer to the axis X (see Figure 3) that passes through the center of the weighing machine.

(3) Camera Installation Height The height of the canopy on which the camera is mounted varies depending on the gas station, generally ranging from 4 to 6 meters. Therefore, differences in camera installation height are accommodated by the zoom function of the camera lens. For example, the camera specifications are as follows: focal length = 2.8 to 8.5 mm, horizontal angle of view = 38 to 110°, vertical angle of view = 21 to 62°, resolution = Full HD (1920 x 1080), and aspect ratio = 16:9. It is desirable to select a camera that can be set to a certain degree of telephoto and wide-angle. For example, if the camera is installed at a height of 4 meters, it should be set to wide-angle, and if the camera is installed at a height of 6 meters, it should be set to telephoto. Furthermore, it is preferable for the camera to be capable of automatic zoom adjustment, and it is even more preferable for it to be capable of automatic pan and tilt adjustment.

(4) Camera Angle If the optical axis of the camera is pointed directly downward, the weighing machine body will not be visible, and the operation in front of the weighing machine will not be visible. Also, part of the refueling operation may be hidden in the shadow of the vehicle. Therefore, it is ideal to angle the camera so that the optical axis is slightly tilted toward the weighing machine. Therefore, as shown in Figure 4, it is desirable to tilt the body of camera 10 slightly toward the weighing machine 20. This makes it possible to capture an image of the weighing machine looking down from the front, as shown in Figure 5.

In the example of Figure 5, video 50 is captured that shows the space between the weighing machine 20 and the vehicle 30, i.e., the overall state of the refueling operation. Therefore, video can be captured that shows the area around the vehicle's fuel filler opening and the state of the inserted fuel filler nozzle without being obscured by the person filling the fuel, making it possible to check the situation in front of the weighing machine without any obstructions.

After installing the camera in the position determined as described above, the camera's field of view must be adjusted so that it can capture images suitable for the image processing AI. Figure 6 shows an example configuration of a camera installation support system used to adjust the camera's field of view according to one embodiment of the present invention. The camera installation support system shown in the figure comprises a camera 10 and a camera control device 60. The camera control device 60 is implemented, for example, by a computer equipped with hardware resources such as a processor and memory, and is configured so that the processor executes programs to realize each function according to the present invention.

The camera control device 60 in Figure 6 has a memory 61, a control unit 62, a display unit 63, and an operation unit 64, and is connected to the camera 10 for wired or wireless communication. The operation of the camera control device 60 will be explained below using three examples: manual adjustment of the field of view of the camera 10, semi-automatic adjustment, and fully automatic adjustment.

(1) Manual adjustment of the field of view After fixing the camera in a predetermined position, the field of view is determined by adjusting the angle and zoom of the camera body. When using a bread-type camera or a box-type camera that does not have the function to remotely control pan and tilt, the angle must be adjusted manually. Also, when using a varifocal lens for zoom adjustment, manual adjustment is required. Because camera installation is a manual task at a high altitude, accurate adjustments cannot be made based on instructions based on experience, and strict precision cannot be achieved because it is done manually.

To solve this problem, an adjustment image 70 including a guide 72 defining the position and size of the weighing machine in the image, as shown in an example in Figure 7A, is prepared in advance and stored in the memory 61 of the camera control device 60. Under the control of the control unit 62, the camera control device 60 superimposes the adjustment image 70 on the image captured by the camera 10 in real time and displays it on the display unit 63. The worker directly operates the camera 10 to adjust the field of view of the camera 10 so that the position and size of the weighing machine in the image displayed on the display unit 63 of the camera control device 60 matches the guide 72 in the adjustment image 70. This improves the accuracy of manual field of view adjustment, allowing anyone to adjust the field of view uniformly.

As one example, the camera control device 60 is realized as a terminal (e.g., a tablet) carried by a worker manually adjusting the camera 10. In this case, the worker can check the captured image with the adjustment image superimposed on it on the spot. As another example, the camera control device 60 is realized as a monitoring device installed inside a gas station building. In this case, personnel inside the building can give instructions to the worker while checking the captured image with the adjustment image superimposed on it. As yet another example, the camera control device 60 may be realized as a remote monitoring device installed at a separate location connected to the gas station equipment via a network line.

Figure 7A shows an adjustment image 70 used when one camera is installed per lane, and is intended to capture an image in which the weighing machine is visible in the upper center. When capturing images of two lanes with one camera, as shown in another example in Figure 7B, vehicles will be parked on both sides of the weighing machine, so an adjustment image 74 can be used that includes a guide 76 in the center that defines the position and size of the weighing machine in the image.

(2) Semi-Automatic Adjustment of Field of View When a camera 10 capable of remotely controlling PTZ (pan, tilt, and zoom) is used, the field of view adjustment process can be automated to some extent. That is, under the control of the control unit 62, the camera control device 60 superimposes the adjustment image 70 (or adjustment image 74) on the image captured by the camera 10 in real time and displays it on the display unit 63. The control unit 62 then remotely controls the pan, tilt, and zoom of the camera 10 in accordance with operations received by the operation unit 64. The operator operates the operation unit 64 while viewing the display on the display unit 63 of the camera control device 60 to adjust the orientation of the camera 10 in the vertical and horizontal directions by remotely controlling the pan and tilt. Similarly, the field of view of the camera 10 in the vertical direction can be adjusted by remotely controlling the zoom.

In this way, if the field of view of the camera 10 can be adjusted by remote control, manual adjustment of the field of view at high altitudes will no longer be necessary after the camera is installed. The camera control device 60 may be implemented as a terminal carried by an operator (e.g., a tablet), as a monitoring device installed inside the gas station building, or as a remote monitoring device installed at a separate location connected to the gas station equipment via a network line.

(3) Fully Automatic Adjustment of Field of View When the camera 10 is installed in a position where a weighing machine is visible somewhere in the captured image, the field of view adjustment process can be fully automated. That is, under the control of the control unit 62, the camera control device 60 detects the weighing machine from the captured image of the camera 10 and determines its position and size. The control unit 62 then remotely controls the pan, tilt, and zoom of the camera 10 so that the weighing machine matches the position and size indicated by the guide 72 in the adjustment image 70 (or the guide 76 in the adjustment image 74). Note that in this embodiment, predetermined numerical information indicating the position and size of the weighing machine may be used instead of the guide on the adjustment image. Hereinafter, information including the position and size of the weighing machine, which serves as the basis for the field of view of the camera 10, will be referred to as guide information.

The fully automatic adjustment of the field of view range will be described below with reference to the flowchart of FIG.
After the camera is installed, the camera control device 60 first detects the weighing machine from the video captured by the camera 10 (step S11). In this example, in order to enable the detection of the weighing machine from the captured video, video of the weighing machine captured by a camera whose field of view has been appropriately adjusted is collected in advance as training data, and a trained model for detecting the weighing machine from the captured video is generated based on this training data. The weighing machine detection AI using this trained model can detect the weighing machine included in the captured video and calculate the position and size of the weighing machine.

Next, the camera control device 60 automatically controls the pan, tilt, and zoom of the camera 10 so that the position and size of the weighing machine included in the captured video match the position and size of the weighing machine included in the guide information. As an example, first, the difference between the center coordinates of the weighing machine detected from the captured video and the center coordinates of the weighing machine in the guide information, i.e., the number of pixels (amount of deviation) in the horizontal (H) and vertical (V) directions, is calculated (step S12). Then, it is determined whether there is any positional deviation of the weighing machine in the captured video (step S13). If the calculated amount of deviation is zero (or is below a predetermined threshold), it is determined that there is no positional deviation; otherwise, it is determined that there is positional deviation.

If it is determined that the weighing machine's position is misaligned in the captured image, the pan/tilt of camera 10 is adjusted so that the center coordinates of the weighing machine included in the captured image approach the center coordinates of the weighing machine in the guide information (i.e., so that the amount of misalignment becomes smaller) (step S14). The processes of steps S12 to S14 are repeated until the positional misalignment is resolved. In other words, each time the pan/tilt of camera 10 is adjusted, the misalignment between the center coordinates of the weighing machine included in the captured image and the center coordinates of the weighing machine in the guide information is calculated, and the pan/tilt is adjusted again to correct that misalignment. In this way, the camera control device 60 gradually corrects the orientation of camera 10 by controlling pan/tilt multiple times.

Next, the camera control device 60 adjusts the zoom of the camera 10 to adjust the field of view in the height direction. Here, when the AI detects a weighing machine in the captured video, it calculates the size of the weighing machine in the captured video (step S15). The size of the weighing machine can be obtained, for example, by determining a rectangle that indicates the boundary of the weighing machine in the captured video and calculating the size of this rectangle. Next, it is determined whether the size of the weighing machine included in the captured video matches the size of the weighing machine in the guide information (step S16).

If a size mismatch is determined, the zoom of camera 10 is adjusted so that the size of the weighing device included in the captured video is approximately the same as the size of the weighing device in the guide information (step S17). That is, if the size of the weighing device included in the captured video is smaller than the size of the weighing device in the guide information, it zooms in; otherwise, it zooms out. The processes of steps S15 to S17 are repeated until the size mismatch is resolved. That is, each time the zoom of camera 10 is adjusted, the size of the weighing device included in the captured video is calculated, and the zoom is adjusted again so that it matches the size of the weighing device in the guide information. In this way, the camera control device 60 gradually corrects the camera's vertical field of view by controlling the zoom multiple times.

As described above, the camera control device 60 adjusts the pan and tilt of the camera 10 so that the center coordinates of the weighing machine included in the image captured by the camera 10 match the center coordinates of the weighing machine in the guide information, and adjusts the zoom of the camera 10 so that the size of the weighing machine included in the image captured by the camera 10 matches the size of the weighing machine in the guide information, thereby making it possible to automatically adjust the field of view of the camera 10. The camera control device 60 may be implemented as a terminal carried by an operator (e.g., a tablet), as a monitoring device installed inside the gas station building, or as a remote monitoring device installed at a separate location connected to the gas station equipment via a network line.

Note that the process described with reference to the flowchart in Figure 8 is merely an example, and the camera's field of view range may be automatically adjusted using other processes. Furthermore, to enable the worker to understand the process of automatically adjusting the field of view range, an adjustment image may be superimposed on the image captured by the camera in real time. The image captured by the camera along with the adjustment image may be displayed on the display unit 63 of the camera control device 60, or on another device connected to the camera control device 60 so that it can communicate with the camera control device 60.

As explained above, based on the analysis of the stopping position of the vehicle performing refueling, the posture of the refueler, and the behavior of the refueler, it is considered effective to install the camera for the image processing AI in a position directly above the weighing machine or in a position shifted from directly above the weighing machine in the direction of the vehicle stopping position. Therefore, in this embodiment, an installation process is first carried out in which the camera is installed in a position directly above the weighing machine installed at the gas station or in a position shifted from directly above the weighing machine in the direction of the vehicle stopping position, and then an adjustment process is carried out in which the field of view of the camera is adjusted using an adjustment image that includes a guide that defines the position and size of the weighing machine in the image.

By installing cameras in this way, it is possible to reduce blind spots when monitoring refueling operations. Therefore, whether an employee visually checks the camera footage or the footage is used to detect dangerous behavior using image processing AI, it is possible to detect dangerous behavior with high accuracy. Furthermore, it is possible to detect not only dangerous behavior but also normal behavior in the same way.

Fuel stations have a variety of layouts, but regardless of the layout, it is possible to capture video that captures the entire refueling process, thereby maintaining consistent accuracy in detecting dangerous behavior. Furthermore, it is now possible to seamlessly trace the refueling process from start to finish, which was not possible with conventional camera angles, contributing to the safe operation of fuel stations even in visual confirmation. Furthermore, the number of cameras can be minimized in system processing, reducing the processing load on the image processing AI and contributing to cost reductions.

In addition, because the field of view of camera footage can be standardized, it is less likely that the accuracy of AI inference will vary significantly from gas station to gas station due to differences in the angle of view. Furthermore, by standardizing the field of view, it becomes possible to build AI that can be used universally, regardless of the gas station. As a result, the field of view will be roughly the same for footage taken at all gas stations, making it possible to standardize the rules for annotation work in generating training data. As a result, the same work can be done regardless of the footage taken at any gas station, making it possible to carry out efficient and highly accurate annotation work, and thereby providing highly accurate AI.

The above explanation does not specifically mention the type of weighing machine, but since there are actually many different types of weighing machines, it is preferable to prepare multiple guide information according to the type of weighing machine. Furthermore, since the same type of weighing machine can vary in size, different guide information may be prepared for each size of weighing machine. Alternatively, guide information may be prepared according to the size of a reference weighing machine, and the size of the weighing machine in the guide information may be adjusted depending on the difference between the actual size of the weighing machine and the size of the reference weighing machine. In this case, there is no need to prepare different guide information for each size of weighing machine. Furthermore, depending on the settings of the image processing AI, it may be necessary to capture an image in which the weighing machine is shown in a different position (for example, the center left side, the center bottom side, or the center right side). In that case, guide information including the required position can be used.

Furthermore, in the above explanation, the field of view range was adjusted using the weighing machine itself in the camera image, but a different reference object may also be used. For example, the reference object may be a square or rectangular mark of a predetermined size that is placed or painted at a predetermined position relative to the position of the weighing machine.

The above describes embodiments of the present invention, but these embodiments are merely examples and do not limit the technical scope of the present invention. The present invention can take on a variety of other embodiments, and various modifications such as omissions and substitutions can be made without departing from the spirit of the present invention. These embodiments and their modifications are included within the scope and spirit of the invention described in this specification, etc., and are also included in the invention described in the claims and their equivalents.

In addition, the present invention can be provided not only as the devices described above or as systems composed of these devices, but also as methods executed by these devices, programs for implementing the functions of these devices using a processor, and storage media for storing such programs in a computer-readable format.

The present invention can be used in a system that monitors self-service refueling at gas stations.

10, 10A, 10B, 10': camera, 20: weighing machine, 30: vehicle, 40A, 40B: vehicle stopping position, 50: captured image, 60: camera control device, 61: memory, 62: control unit, 63: display unit, 64: operation unit, 70, 74: adjustment image, 72, 76: guide

Claims (11)

A camera installed directly above a metering machine at a gas station or at a position shifted from directly above the metering machine toward a vehicle stopping position;
a camera control device having a function of controlling pan, tilt and zoom of the camera,
The camera control device uses guide information that specifies the position and size of a reference object in the image captured by the camera to perform a process of controlling the pan and tilt of the camera so that the center coordinates of the reference object in the image captured by the camera match the center coordinates of the reference object included in the guide information, and then performs a process of controlling the zoom of the camera so that the size of the reference object in the image captured by the camera matches the size of the reference object included in the guide information. The camera installation support system according to claim 1,
The camera control device is characterized in that it performs a process of detecting a reference object from the image captured by the camera using a trained model for detecting a reference object contained in the image. 3. The camera installation support system according to claim 1,
The camera control device is characterized in that it executes a process of outputting an image in which an image indicating the position and size of a reference object included in the guide information is superimposed on the image captured by the camera. 4. The camera installation support system according to claim 1,
A camera installation support system characterized in that the reference object is a weighing machine. 4. The camera installation support system according to claim 1,
A camera installation support system characterized in that the reference object is a mark installed or painted on the ground. A camera control device having a function of controlling pan, tilt, and zoom of a camera installed at a gas station,
The camera is installed directly above a metering machine at the gas station or at a position shifted from directly above the metering machine toward a vehicle stopping position,
The camera control device uses guide information that specifies the position and size of a reference object in the image captured by the camera to perform a process of controlling the pan and tilt of the camera so that the center coordinates of the reference object in the image captured by the camera match the center coordinates of the reference object included in the guide information, and then performs a process of controlling the zoom of the camera so that the size of the reference object in the image captured by the camera matches the size of the reference object included in the guide information. an installation step of installing a camera at a position directly above a metering machine at a gas station or at a position shifted from directly above the metering machine toward a vehicle stopping position;
an adjustment step of adjusting the field of view of the camera using an adjustment image including a guide that defines the position and size of a reference object in the image captured by the camera ,
A camera installation method characterized in that in the adjustment process, a process is performed to control the pan and tilt of the camera so that the center coordinates of the reference object in the image captured by the camera and the center coordinates of the guide in the adjustment image match, and then a process is performed to control the zoom of the camera so that the size of the reference object in the image captured by the camera and the size of the guide in the adjustment image match . The camera installation method according to claim 7,
A camera installation method characterized in that the adjustment process includes a process of detecting a reference object from the image captured by the camera using a trained model for detecting a reference object from camera image. The camera installation method according to claim 7 or 8 ,
A camera installation method characterized in that the reference object is a weighing machine. The camera installation method according to claim 7 or 8 ,
A camera installation method characterized in that the reference object is a mark installed or painted on the ground. The camera installation method according to any one of claims 7 to 10 ,
The camera installation method, wherein the adjustment step includes a process of displaying the adjustment image superimposed on the image captured by the camera.