KR20240083448A - Method, computer device, and computer program to detect polygonal signs based on deep learning - Google Patents

Abstract

A method, computer device, and computer program for detecting a polygonal sign based on deep learning are disclosed. A method for detecting a sign in a polygonal shape based on deep learning includes the steps of generating sign data labeling a sign area included in a learning target image in a polygonal shape; and learning a sign detection model that detects a polygon-shaped sign area using the sign data.

Description

Method, computer device, and computer program for detecting polygonal signs based on deep learning {METHOD, COMPUTER DEVICE, AND COMPUTER PROGRAM TO DETECT POLYGONAL SIGNS BASED ON DEEP LEARNING}

The explanation below relates to technology for detecting store signs that exist in real space.

POI (point of interest) information may refer to information about major places such as shops, government offices, and schools in buildings around roads.

In order to provide accurate information on maps or augmented reality (AR), you must always maintain the latest POI information.

As an example of related technology, Korean Patent Publication No. 10-1183519 (registration date September 11, 2012) generates POI information, which is information on key points, using geographic information panoramas created in panoramic form with photos of actual terrain features. A technology that can do this has been disclosed.

Technology that utilizes signage information included in images is being used to automatically generate POI information.

However, recognition is difficult because not only are the sizes and shapes of the signs diverse, but the fonts, positions, and directions of the letters within the signs are different.

In addition, it is not easy to detect the sign area due to perspective distortion depending on the shooting location in real space.

Provide a method and device for detecting the location, size, and shape of a sign by considering the perspective view of a road view image.

We provide a method and device for detecting a sign area in the form of an M-sided polygon through a deep learning model and obtaining a front sign image from this.

A method performed on a computer device, wherein the computer device includes at least one processor configured to execute computer-readable instructions included in a memory, wherein the method includes, by the at least one processor, an image included in a learning target image. Generating signboard data labeling the signboard area in a polygonal shape; and learning, by the at least one processor, a sign detection model for detecting a polygonal sign area using the sign data.

According to one aspect, the generating step includes: acquiring location information of a vertex on the M side of the signboard area in the learning target image; Obtaining a front sign image for the sign area using the M-side vertex position information; And it may include extracting a sign feature vector, which is a feature vector of a certain dimension, from the front sign image.

According to another aspect, the learning step may use the sign feature vector to learn the sign detection model so that feature vectors are similar between the same sign and feature vectors are different between different signs.

According to another aspect, the step of acquiring the front sign image includes creating an enlarged sign area by enlarging each side of the sign area by a certain number of times based on the center point of the M side; And it may include creating the front sign image by warping the enlarged sign area to a predefined size.

According to another aspect, the learning step may learn the sign detection model so that M-side location information and confidence level for the sign area are output as a sign detection result.

According to another aspect, the method further includes detecting, by the at least one processor, a target sign area in the shape of a polygon in the street view image through the sign detection model when an arbitrary street view image is given. can do.

According to another aspect, the detecting step may include obtaining M-side location information and reliability for each of the target sign areas detected in the street view image.

According to another aspect, the detecting step may include calculating the reliability of the target sign area by performing regression from a reference anchor to the target sign area. You can.

According to another aspect, the detecting step may include calculating an intersection of union (IOU) or intersection of area (IOA) between a reference anchor and the target sign area as reliability for the target sign area. You can.

According to another aspect, the detecting step includes obtaining a front sign image by warping the target sign area; And it may include detecting a POI change based on image similarity between signboards using the front signage image.

A computer program stored in a computer-readable recording medium is provided to execute the method on a computer.

A computer device comprising: at least one processor configured to execute computer-readable instructions included in a memory, wherein the at least one processor generates sign data labeling a sign area included in a learning target image in a polygonal shape. process; and a computer device that processes a process of learning a sign detection model that detects a polygonal sign area using the sign data.

According to embodiments of the present invention, by detecting the sign area in the form of an M-sided polygon through a deep learning model and obtaining a front sign image from this, it is possible to provide a sign detection environment that is robust to changes depending on the perspective and to solve the problem of perspective distortion. By solving this, the accuracy of sign detection can be improved.

1 is a diagram illustrating an example of a network environment according to an embodiment of the present invention.
Figure 2 is a block diagram showing an example of a computer device according to an embodiment of the present invention.
Figure 3 is a flowchart showing an example of a method that can be performed by a computer device according to an embodiment of the present invention.
Figure 4 shows an example of a deep learning-based sign detection model in one embodiment of the present invention.
Figure 5 is a flowchart showing the process of generating training data for a sign detection model in one embodiment of the present invention.
Figures 6 and 7 are exemplary diagrams for explaining the sign image warping process in one embodiment of the present invention.
Figures 8 and 9 are exemplary diagrams for explaining the sign detection reliability calculation process in one embodiment of the present invention.
Figure 10 shows an example of a sign detection result in the form of an M-side polygon in one embodiment of the present invention.

Hereinafter, embodiments of the present invention will be described in detail with reference to the attached drawings.

Embodiments of the present invention relate to technology for detecting store signs existing in real space.

Embodiments including those specifically disclosed in this specification detect the sign area in the form of an M-side polygon through a deep learning model and obtain a front sign image from this, thereby minimizing misrecognition due to perspective distortion and accurately detecting the sign area. You can.

The sign detection system according to embodiments of the present invention may be implemented by at least one computer device, and the sign detection method according to embodiments of the present invention is performed through at least one computer device included in the sign detection system. It can be. At this time, the computer program according to an embodiment of the present invention may be installed and driven in the computer device, and the computer device may perform the sign detection method according to the embodiments of the present invention under the control of the driven computer program. . The above-described computer program can be combined with a computer device and stored in a computer-readable recording medium to execute the sign detection method on the computer.

1 is a diagram illustrating an example of a network environment according to an embodiment of the present invention. The network environment in FIG. 1 shows an example including a plurality of electronic devices 110, 120, 130, and 140, a plurality of servers 150 and 160, and a network 170. Figure 1 is an example for explaining the invention, and the number of electronic devices or servers is not limited as in Figure 1. In addition, the network environment in FIG. 1 only explains one example of environments applicable to the present embodiments, and the environment applicable to the present embodiments is not limited to the network environment in FIG. 1.

The plurality of electronic devices 110, 120, 130, and 140 may be fixed terminals or mobile terminals implemented as computer devices. Examples of the plurality of electronic devices 110, 120, 130, and 140 include smart phones, mobile phones, navigation devices, computers, laptops, digital broadcasting terminals, Personal Digital Assistants (PDAs), and Portable Multimedia Players (PMPs). ), tablet PC, etc. For example, in FIG. 1, the shape of a smartphone is shown as an example of the electronic device 110. However, in embodiments of the present invention, the electronic device 110 actually communicates with other devices through the network 170 using a wireless or wired communication method. It may refer to one of various physical computer devices capable of communicating with electronic devices 120, 130, 140 and/or servers 150, 160.

The communication method is not limited, and may include not only a communication method utilizing a communication network that the network 170 may include (for example, a mobile communication network, wired Internet, wireless Internet, and a broadcast network), but also short-range wireless communication between devices. For example, the network 170 may include a personal area network (PAN), a local area network (LAN), a campus area network (CAN), a metropolitan area network (MAN), a wide area network (WAN), and a broadband network (BBN). , may include one or more arbitrary networks such as the Internet. Additionally, the network 170 may include any one or more of network topologies including a bus network, star network, ring network, mesh network, star-bus network, tree or hierarchical network, etc. Not limited.

Each of the servers 150 and 160 is a computer device or a plurality of computers that communicate with a plurality of electronic devices 110, 120, 130, 140 and a network 170 to provide commands, codes, files, content, services, etc. It can be implemented with devices. For example, the server 150 is a system that provides services (e.g., map services, augmented reality services, etc.) to a plurality of electronic devices 110, 120, 130, and 140 connected through the network 170. You can.

Figure 2 is a block diagram showing an example of a computer device according to an embodiment of the present invention. Each of the plurality of electronic devices 110, 120, 130, and 140 described above or each of the servers 150 and 160 may be implemented by the computer device 200 shown in FIG. 2.

As shown in FIG. 2, this computer device 200 may include a memory 210, a processor 220, a communication interface 230, and an input/output interface 240. The memory 210 is a computer-readable recording medium and may include a non-permanent mass storage device such as random access memory (RAM), read only memory (ROM), and a disk drive. Here, non-perishable large-capacity recording devices such as ROM and disk drives may be included in the computer device 200 as a separate permanent storage device that is distinct from the memory 210. Additionally, an operating system and at least one program code may be stored in the memory 210. These software components may be loaded into the memory 210 from a computer-readable recording medium separate from the memory 210. Such separate computer-readable recording media may include computer-readable recording media such as floppy drives, disks, tapes, DVD/CD-ROM drives, and memory cards. In another embodiment, software components may be loaded into the memory 210 through the communication interface 230 rather than a computer-readable recording medium. For example, software components may be loaded into memory 210 of computer device 200 based on computer programs installed by files received over network 170.

The processor 220 may be configured to process instructions of a computer program by performing basic arithmetic, logic, and input/output operations. Commands may be provided to the processor 220 by the memory 210 or the communication interface 230. For example, processor 220 may be configured to execute received instructions according to program code stored in a recording device such as memory 210.

The communication interface 230 may provide a function for the computer device 200 to communicate with other devices (eg, the storage devices described above) through the network 170. For example, a request, command, data, file, etc. generated by the processor 220 of the computer device 200 according to a program code stored in a recording device such as memory 210 is transmitted to the network ( 170) and can be transmitted to other devices. Conversely, signals, commands, data, files, etc. from other devices may be received by the computer device 200 through the communication interface 230 of the computer device 200 via the network 170. Signals, commands, data, etc. received through the communication interface 230 may be transmitted to the processor 220 or memory 210, and files, etc. may be stored in a storage medium (as described above) that the computer device 200 may further include. It can be stored as a permanent storage device).

The input/output interface 240 may be a means for interfacing with the input/output device 250. For example, input devices may include devices such as a microphone, keyboard, or mouse, and output devices may include devices such as displays and speakers. As another example, the input/output interface 240 may be a means for interfacing with a device that integrates input and output functions into one, such as a touch screen. The input/output device 250 may be configured as a single device with the computer device 200.

Additionally, in other embodiments, computer device 200 may include fewer or more components than those of FIG. 2 . However, there is no need to clearly show most prior art components. For example, the computer device 200 may be implemented to include at least some of the input/output devices 250 described above, or may further include other components such as a transceiver, a database, etc.

Below, specific examples of a method and device that can detect polygonal signboards based on deep learning will be described.

It is important to maintain the latest POI information to provide accurate information about real space in various services such as maps and augmented reality.

To automatically generate POI information, technology is being used to detect sign information from images taken of real space.

However, the signs in the image are often distorted in perspective rather than frontal depending on the shooting angle.

In these embodiments, when using sign information included in a street view image as POI information, it is possible to implement a sign detection function that is robust to perspective distortion depending on the shooting location of the street view image.

The computer device 200 according to this embodiment may be configured with a computer-implemented sign detection system. For example, the sign detection system may be implemented in the form of a program that operates independently, or may be implemented in the form of an in-app of a specific application to enable operation on the specific application.

The processor 220 of the computer device 200 may be implemented as a component for performing the following sign detection method. Depending on the embodiment, components of the processor 220 may be selectively included in or excluded from the processor 220. Additionally, depending on the embodiment, components of the processor 220 may be separated or merged to express the functions of the processor 220.

The processor 220 and the components of the processor 220 can control the computer device 200 to perform the steps included in the sign detection method below. For example, the processor 220 and its components may be implemented to execute instructions according to the code of an operating system included in the memory 210 and the code of at least one program.

Here, the components of the processor 220 may be expressions of different functions performed by the processor 220 according to instructions provided by program codes stored in the computer device 200.

The processor 220 may read necessary instructions from the memory 210 where instructions related to controlling the computer device 200 are loaded. In this case, the read command may include an command for controlling the processor 220 to execute steps that will be described later.

Steps included in the sign detection method to be described later may be performed in an order different from the order shown, and some of the steps may be omitted or additional processes may be included.

The steps included in the sign detection method may be performed in the server 150, and depending on the embodiment, at least some of the steps may also be performed in any one of the electronic devices 110, 120, 130, and 140. .

Figure 3 is a flowchart showing an example of a method that can be performed by a computer device according to an embodiment of the present invention.

Referring to FIG. 3 , in step S310, the processor 220 may generate sign data labeling the sign area included in the learning target image in the form of an M-sided polygon. The processor 220 may collect street view images from various viewpoints including signboards as learning target images. At this time, the signboard area in the learning target image is labeled in the form of an M-sided polygon to collect data for learning a signboard detection model. can do.

In step S320, the processor 220 may learn a deep learning-based sign detection model using sign data generated from the learning target image. A sign detection model can be learned using the sign data generated in step S310 so that the output of the sign detection model is an M-sided polygon.

In step S330, when a random street view image is given, the processor 220 uses a deep learning-based sign detection model learned with signboard data labeled in the form of an M-side polygon to select a target in the form of an M-side polygon in the given street view image. Simple areas can be detected.

In this embodiment, in order to detect signboards that are robust to perspective distortion of street view images, a deep learning model learned with signage data labeled in the form of an M-side polygon can provide a signboard detection model that outputs signboard detection results in the form of an M-side polygon. there is.

As the size and shape of signs vary, in many cases it is difficult to define the area of the sign. In the sign detection process, a method of cropping the sign area to the same size or warping it to a certain size can be used.

However, if the sign area is cut to the same size, the effective area is lost or unnecessary areas other than the sign are included, making it difficult to extract correct sign information.

In addition, there are signs with the same business name, although the size is different depending on the location (front or side of the building, etc.) or type (flex sign, protruding sign, etc.), as well as chain stores. In this case, when the sign area is warped to a certain size. There is a problem where matching between identical signs is difficult.

To solve this problem, the present invention learns a sign detection model in the form of an M-side polygon based on deep learning, thereby providing a sign detection environment that is robust to perspective distortion.

The processor 220 can detect a sign area in the form of an M-sided polygon according to the perspective view in the street view image through a deep learning-based sign detection model. Furthermore, the processor 220 can detect POI changes by warping the sign area detected through the sign detection model to a frontal image and measuring image similarity between the signboards based on the frontal image.

Figure 4 shows an example of a deep learning-based sign detection model in one embodiment of the present invention.

Referring to FIG. 4, the sign detection model 40 according to the present invention may use a set of images collected in advance as street view images from various viewpoints as input data. In particular, a street view image set in which the sign area is labeled in the form of an M-point polygon can be used as an input to the sign detection model 40.

The processor 220 learns the sign detection model 40 using a street view image set in which the sign area is labeled in the form of an M-point polygon, and uses a sign detection model to output M-side polygon information as a detection result for the sign area. (40) can be learned.

The output parameters of the sign detection model 40 are the M-side position values (p1, ..., pM) for each sign area (POI #1, ..., POI #N) detected in the street view image and the detection of the corresponding sign area. May include confidence level (conf).

Figure 5 is a flowchart showing the process of generating training data for a sign detection model in one embodiment of the present invention.

Referring to FIG. 5 , in step S501, when a learning target image is given, the processor 220 may detect a polygon-shaped sign area by applying a sign detector to each image. The processor 220 may acquire vertex position information on the M side of the signboard area included in the learning target image.

In step S502, the processor 220 may obtain a front sign image by warping the M-side polygonal sign area to a predefined size.

Figures 6 and 7 are exemplary diagrams for explaining the sign image warping process in one embodiment of the present invention.

First, as shown in FIG. 6, the processor 220 uses the vertex position information of the M side of the sign area 601 detected in the learning target image to divide each side by a certain multiple (e.g., based on the center point of the M side). , 1.5 times, etc.), an enlarged sign area 602 can be created.

Next, as shown in FIG. 7, the processor 220 warps the enlarged signboard area 602 in the form of an M-side polygon into an image of a predefined shape (e.g., rectangle, etc.) and size to obtain surrounding information. A front signage image 703 including can be obtained.

Referring again to FIG. 5, in step S503, the processor 220 generates a feature vector (hereinafter referred to as 'sign feature') of a certain dimension (e.g., 1024 dimensions) from the front sign image 703 acquired in step S502. (referred to as a ‘vector’) can be extracted. The processor 220 may use the sign feature vector extracted from the front sign image 703 as data for learning a deep learning-based sign detection model.

The processor 220 may learn a sign detection model using a sign feature vector extracted from the front sign image 703. The processor 220 may train the sign detection model so that feature vectors between the same signboards are similar to each other and feature vectors between different signboards have large differences.

Figures 8 and 9 are exemplary diagrams for explaining the sign detection reliability calculation process in one embodiment of the present invention.

The processor 220 may perform model learning so that a reliability value is output along with the M-side position value for the sign area detected in the image.

The processor 220 may calculate the reliability of the sign area in the form of an M-sided polygon detected in the learning target image.

For example, referring to FIG. 8, when the processor 220 detects a sign area in the form of a four-sided polygon, regression is performed from the reference square anchor 80 to the sign area 601. Through this method, the IOU (intersection of union), which is a reliability indicator for the corresponding sign area, can be calculated.

IOU is an indicator of how accurate the bounding box detected in the image is. It divides the overlapping intersection area between the detected bounding box and the reference bounding box by the union area of the two bounding boxes. When calculating the IOU between an anchor and a labeled sign image to determine the reliability of the sign area, it is difficult to include all signs in the anchor due to the diversity of signs. In particular, in the case of diagonal sign areas, the IOU is bound to be very low.

Therefore, in the case of a sign 901 in the diagonal direction, as shown in (A) of FIG. 9, the anchor 80 can be activated by calculating the intersection of area (IOA) even if the IOU is low, and as shown in (B) of FIG. 9 In the case of a long sign (901), it can be incorporated into the longest anchor (80) with a low IOU. IOA calculates the ratio of the intersection area of two bounding boxes to the total area of the anchor.

In other words, the processor 220 can calculate the reliability value for the sign area detected in the learning target image by calculating the IOU for the sign area. At this time, the processor 220 evaluates the detection accuracy of the sign area using the IOU. For example, if the IOU is less than a threshold, the processor 220 may evaluate the detection accuracy of the corresponding sign area using the IOA.

Figure 10 shows an example of a sign detection result in the form of an M-side polygon in one embodiment of the present invention.

Referring to FIG. 10, when a random street view image 100 is given, the processor 220 detects sign areas 1001 at various perspective points through a sign detection model 40 learned with sign data in the form of an M-sided polygon. It can be detected.

In other words, the processor 220 may detect the target sign area in the form of an M-sided polygon in the given street view image 100. At this time, the processor 220 may collect M-side polygon coordinate information representing the sign area as a sign detection result, and may also collect reliability information for the corresponding detection area.

The processor 220 can detect/recognize the position, size, and shape of the sign in the street view image in the form of an M-side polygon according to the perspective and perspective through a deep learning-based sign detection model.

The processor 220 may measure image similarity between signs by warping the M-side polygon-shaped sign area detected through the sign detection model into a front sign image including the surrounding area of the sign.

The sign detection technology according to the present invention can be used in the field of technology for automatically generating POI information existing in real space to form a database, and technology for detecting changes in POI existing in real space.

In this way, according to embodiments of the present invention, by detecting the sign area in the form of an M-side polygon through a deep learning model and obtaining a front sign image from this, a sign detection environment that is robust to changes depending on the perspective can be provided and perspective distortion is possible. By solving the problem, sign detection accuracy can be improved.

The device described above may be implemented with hardware components, software components, and/or a combination of hardware components and software components. For example, the devices and components described in the embodiments include a processor, a controller, an arithmetic logic unit (ALU), a digital signal processor, a microcomputer, a field programmable gate array (FPGA), and a programmable logic unit (PLU). It may be implemented using one or more general-purpose or special-purpose computers, such as a logic unit, microprocessor, or any other device capable of executing and responding to instructions. The processing device may execute an operating system (OS) and one or more software applications running on the operating system. Additionally, a processing device may access, store, manipulate, process, and generate data in response to the execution of software. For ease of understanding, a single processing device may be described as being used; however, those skilled in the art will understand that a processing device includes multiple processing elements and/or multiple types of processing elements. It can be seen that it may include. For example, a processing device may include a plurality of processors or one processor and one controller. Additionally, other processing configurations, such as parallel processors, are possible.

Software may include a computer program, code, instructions, or a combination of one or more of these, which may configure a processing unit to operate as desired, or may be processed independently or collectively. You can command the device. The software and/or data may be embodied in any type of machine, component, physical device, computer storage medium or device for the purpose of being interpreted by or providing instructions or data to the processing device. there is. Software may be distributed over networked computer systems and stored or executed in a distributed manner. Software and data may be stored on one or more computer-readable recording media.

The method according to the embodiment may be implemented in the form of program instructions that can be executed through various computer means and recorded on a computer-readable medium. At this time, the medium may continuously store a computer-executable program, or temporarily store it for execution or download. In addition, the medium may be a variety of recording or storage means in the form of a single or several pieces of hardware combined. It is not limited to a medium directly connected to a computer system and may be distributed over a network. Examples of media include magnetic media such as hard disks, floppy disks, and magnetic tapes, optical recording media such as CD-ROMs and DVDs, magneto-optical media such as floptical disks, And there may be something configured to store program instructions, including ROM, RAM, flash memory, etc. Additionally, examples of other media include recording or storage media managed by app stores that distribute applications, sites or servers that supply or distribute various other software, etc.

As described above, although the embodiments have been described with limited examples and drawings, various modifications and variations can be made by those skilled in the art from the above description. For example, the described techniques are performed in a different order than the described method, and/or components of the described system, structure, device, circuit, etc. are combined or combined in a different form than the described method, or other components are used. Alternatively, appropriate results may be achieved even if substituted or substituted by an equivalent.

Therefore, other implementations, other embodiments, and equivalents of the claims also fall within the scope of the claims described below.

Claims (20)

In a method performed on a computer device,
The computer device includes at least one processor configured to execute computer readable instructions contained in a memory,
The method is:
Generating, by the at least one processor, sign data labeling a sign area included in a learning target image in a polygonal shape; and
Learning, by the at least one processor, a sign detection model for detecting a polygonal sign area using the sign data.
How to include . According to paragraph 1,
The generating step is,
Obtaining vertex position information on the M side of the signboard area in the learning target image;
Obtaining a front sign image for the sign area using the M-side vertex position information; and
Extracting a sign feature vector, which is a feature vector of a certain dimension, from the front sign image
How to include . According to paragraph 2,
The learning step is,
Using the signboard feature vectors, learn the signboard detection model so that feature vectors are similar between the same signboards and feature vectors are different between different signboards.
A method characterized by . According to paragraph 2,
The step of acquiring the front signage image is,
Creating an enlarged signboard area by enlarging each side of the signboard area by a certain number of times based on the center point of side M; and
Creating the front sign image by warping the enlarged sign area to a predefined size.
How to include . According to paragraph 1,
The learning step is,
Learning the sign detection model so that M-side location information and confidence level for the sign area are output as a sign detection result.
A method characterized by . According to paragraph 1,
The method is:
Detecting, by the at least one processor, a target sign area in the form of a polygon in the street view image through the sign detection model when a random street view image is given.
How to further include . According to clause 6,
The detecting step is,
Obtaining M-side location information and reliability for each of the target signage areas detected in the street view image.
How to include . According to clause 6,
The detecting step is,
Calculating the reliability of the target sign area by performing regression from a reference anchor to the target sign area.
How to include . According to clause 6,
The detecting step is,
Calculating the intersection of union (IOU) or intersection of area (IOA) between a reference anchor and the target sign area as the reliability for the target sign area.
How to include . According to clause 6,
The detecting step is,
Obtaining a front sign image by warping the target sign area; and
Detecting point of interest (POI) changes based on image similarity between signboards using the front signage image
How to include . A computer program stored in a computer-readable recording medium for executing the method of any one of claims 1 to 10 on a computer. In computer devices,
At least one processor configured to execute computer readable instructions contained in memory
Including,
The at least one processor,
A process of generating sign data that labels the sign area included in the learning target image in a polygonal form; and
The process of learning a sign detection model that detects a polygonal sign area using the sign data
A computer device that processes According to clause 12,
The at least one processor,
Obtaining the location information of the M side vertex of the signboard area in the learning target image,
Obtain a front sign image for the sign area using the M side vertex location information,
Extracting a signboard feature vector, which is a feature vector of a certain dimension, from the front signage image
A computer device characterized by a. According to clause 13,
The at least one processor,
Using the signboard feature vectors, learn the signboard detection model so that feature vectors are similar between the same signboards and feature vectors are different between different signboards.
A computer device characterized by a. According to clause 13,
The at least one processor,
For the sign area, each side is enlarged by a certain number based on the center point of side M to create an enlarged sign area,
Creating the front sign image by warping the enlarged sign area to a predefined size.
A computer device characterized by a. According to clause 12,
The at least one processor,
Learning the sign detection model so that M-side location information and reliability for the sign area are output as a sign detection result.
A computer device characterized by a. According to clause 12,
The at least one processor,
When a random street view image is given, detecting a target sign area in the form of a polygon from the street view image through the sign detection model.
A computer device characterized by a. According to clause 17,
The at least one processor,
Obtaining M-side location information and reliability for each of the target signage areas detected in the street view image.
A computer device characterized by a. According to clause 17,
The detecting step is,
Calculating the IOU or IOA between a reference anchor and the target sign area as the reliability for the target sign area
A computer device characterized by a. According to clause 17,
The at least one processor,
Obtain a front sign image by warping the target sign area,
Detecting POI changes based on image similarity between signs using the front signage image
A computer device characterized by a.