KR20240057142A - Method, computer device, and computer program to detect and classify objects based on deep learning - Google Patents

Abstract

A method, computer device, and computer program for detecting and classifying objects based on deep learning are disclosed. The object classification method includes detecting an object from an image of a target space through a deep learning model; and classifying the object using planogram (POG) data related to the target space.

Description

Method for detecting and classifying objects based on deep learning, computer device, and computer program {METHOD, COMPUTER DEVICE, AND COMPUTER PROGRAM TO DETECT AND CLASSIFY OBJECTS BASED ON DEEP LEARNING}

The explanation below is about technology for detecting and classifying objects in images.

One of the fields where deep learning is most actively applied is computer vision.

Computer vision technology is a technology that generates useful information by analyzing images of surrounding objects and environmental properties from images input from a computer through visual media such as cameras and scanners. It is used to perform various tasks through image processing and background analysis. Generate the necessary information.

In particular, object detection and classification algorithms that display and label the location of a specific object in a given input image are developing at a uniquely fast pace among computer vision technologies.

As an example of related technology, Korean Patent Publication No. 10-2021-0067498 (published on June 8, 2021) discloses a technology for automatically detecting objects in an image based on deep learning.

Based on deep learning, we provide a method and device for classifying objects detected in images using NLP (natural language processing) technology and POG (planogram) data.

After performing primary classification of an object based on NLP technology, we provide a method and device that can perform detailed classification of the object using the primary classification result and POG data.

In an object classification method executed on a computer device, the computer device includes at least one processor configured to execute computer-readable instructions included in a memory, and the object classification method includes, by the at least one processor, deep Detecting an object from an image of a target space through a learning model; and classifying the object using planogram (POG) data related to the target space, by the at least one processor.

According to one aspect, the detecting step may detect an object area in the image using a region proposal network (RPN) model.

According to another aspect, the RPN model may be a model in which a region estimate (region proposal) for an item included in the target space is learned using public data related to the target space.

According to another aspect, the classifying step includes registering the product name and product image for each product included in the POG data in a gallery for product recognition; And it may include classifying the object by the product name registered in the gallery according to its similarity to the product image registered in the gallery.

According to another aspect, the classifying step includes performing primary classification on the object through a natural language processing (NLP)-based object classification model; And it may include performing detailed classification of the object using the primary classification result and the POG data.

According to another aspect, the NLP-based object classification model may include a transformer-based text encoder.

According to another aspect, performing primary classification on the object may include recognizing a class name corresponding to the object by comparing image features of the object with text features of the class name included in public data. You can.

According to another aspect, performing detailed classification of the object includes registering the product name and product image for each product included in the POG data in a gallery for product recognition; Comparing product images in a range corresponding to the primary classification result among product images registered in the gallery with the object; And it may include classifying the object by the product name registered in the gallery according to the similarity with the product image compared with the object.

According to another aspect, performing detailed classification of the object includes registering the product name and product image for each product included in the POG data in a gallery for product recognition; and classifying the object by a product name registered in the gallery according to its similarity to the product image registered in the gallery, and a weight according to the primary classification result may be reflected in the similarity.

According to another aspect, the object classification method may further include acquiring, by the at least one processor, the image in which the target space is captured at a certain period or in real time through an image sensing module composed of an RGB sensor. You can.

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

A computer device comprising: at least one processor configured to execute computer readable instructions included in a memory, the at least one processor comprising: detecting an object from an image of a target space through a deep learning model; and a computer device that processes the process of classifying the object using POG data related to the target space.

According to embodiments of the present invention, detection and classification of tens of thousands of objects without class restrictions can be performed by classifying objects detected in images based on deep learning using NLP technology and POG data. You can.

According to embodiments of the present invention, detection and classification of new objects can be performed by performing primary classification of an object based on NLP technology and then performing detailed classification of the object using the primary classification result and POG data. You can.

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.
FIG. 3 is a diagram illustrating an example of components that a processor of a computer device according to an embodiment of the present invention may include.
Figure 4 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 5 shows an example of a process for acquiring an image of a target space in one embodiment of the present invention.
Figure 6 shows an example of a process for detecting an object in an image according to an embodiment of the present invention.
Figure 7 shows an example of an NLP-based object classification process in one embodiment of the present invention.
Figure 8 shows an example of POG data related to the target space in one embodiment of the present invention.
Figure 9 shows an example of a detailed object classification process using NLP-based object classification results and POG data in an 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 automatically detecting and classifying objects in images.

Embodiments including those specifically disclosed in this specification can classify objects detected in an image using NLP technology and POG data based on deep learning.

The object classification system according to embodiments of the present invention may be implemented by at least one computer device, and the object classification method according to embodiments of the present invention is performed through at least one computer device included in the object classification 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 object classification method according to the embodiment 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 object classification 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, PDAs (Personal Digital Assistants), and PMPs (Portable Multimedia Players). ), 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 may be a system that provides services (eg, object management services, etc.) to a plurality of electronic devices 110, 120, 130, and 140 connected through the network 170.

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, 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.

Hereinafter, specific embodiments of a method and device that can detect and classify objects based on deep learning will be described.

In a target space where products of numerous classes are displayed, such as a supermarket or various stores, there may be products with small class differences, and for the same product, there may be significant differences depending on the viewing direction or the shape in which it is displayed.

The way products are displayed is different for each target space, and the display location may change frequently depending on new product launches, discount events, etc.

These embodiments can automatically detect and classify objects in images taken of the target space based on deep learning. The object detection and classification technology according to the present invention can be used in various fields such as searching for the location of items in space, managing inventory, providing product descriptions, and checking display status.

Figure 3 is a block diagram showing an example of components that the processor of a computer device according to an embodiment of the present invention may include, and Figure 4 is an object that the computer device can perform according to an embodiment of the present invention. This is a flowchart showing an example of a classification method.

The computer device 200 according to this embodiment can provide object management services through a dedicated application or access to a web/mobile site. The computer device 200 may be configured with a computer-implemented object classification system. For example, an object classification 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 so that it can operate on the specific application.

The processor 220 of the computer device 200 is a component for performing an object classification method to be described later, and as shown in FIG. 3, the processor 220 includes an image sensing module 301, an object detection module 302, and an object classification module. It may include (303). 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 its components may control the computer device 200 to perform steps included in the object classification method to be described later. 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. For example, the image sensing module 301 may be used as a functional representation of the processor 220 that controls the computer device 200 according to the above-described instructions so that the computer device 200 acquires an image.

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 a command for controlling the processor 220 to execute an object classification method that will be described later.

Steps included in the object classification 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 object classification 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. .

Referring to FIG. 4, in step S410, the image sensing module 301 may acquire an image of the target space for object detection and classification. For example, the image sensing module 301 may be configured with an RGB sensor, and can obtain images captured at a certain period or in real time for a space where numerous products are displayed.

In step S420, the object detection module 302 can detect the object area by distinguishing all possible objects of interest from the image from the background through a deep learning-based object detection technique. For example, the object detection module 302 can extract all possible object regions (i.e., region proposals) from an image using a region proposal network (RPN) model, which is one of the deep learning models.

In step S430, the object classification module 303 may classify the object detected in the image using NLP technology and POG data related to the target space. The object classification module 303 may perform primary classification of an object based on NLP technology and then perform detailed classification of the object using the primary classification result and POG data.

Figure 5 shows an example of a process for acquiring an image of a target space in one embodiment of the present invention.

Referring to Figure 5, the image sensing module 301 may be composed of an RGB sensor, a CCTV 501 fixed in the space 50, a cart 501 movable in the space 50, and a user hand-held ( It may be included in a hand-held camera 503, a self-driving robot 504, etc. The image sensing module 301 can collect images taken of the space 50 where numerous items are displayed at a certain period or in real time.

The image sensing module 301 may include means for interfacing with other components of the computer device 200, through which an image of the space 50 may be transmitted to the processor 220, the object detection module 302, etc. It can be delivered.

Figure 6 shows an example of a process for detecting an object in an image according to an embodiment of the present invention.

Referring to FIG. 6, the object detection module 302 is a network that performs an object area extraction task and may include an RPN model. For example, the RPN model can use an anchor box, which is a bounding box of various sizes, for region proposal as an object area. As an object detection model including target items, the RPN model can be trained to detect regions for all salient objects in an image. At this time, candidate areas for all possible objects can be learned using key public data and their data combinations.

The object detection module 302 can obtain a feature map by inputting the image delivered from the image sensing module 301 into a pre-trained CNN model. The object detection module 302 may transmit the feature map to the RPN and calculate an area estimate through the RPN. The object detection module 302 can obtain a feature map of a fixed size through ROI pooling using the region estimation results by RPN and the feature map by CNN model. The object detection module 302 may perform classification by inputting a feature map of a fixed size into a classifier.

Figure 7 shows an example of an NLP-based object classification process in one embodiment of the present invention.

Referring to FIG. 7, the object classification module 303 may include an NLP-based object classification module 71. The NLP-based object classification module 71 may be composed of a transformer-based text encoder model. For example, the NLP-based object classification module 71 is a visual model using pre-trained text information such as CLIP ( Contrastive Language-Image Pre-training ) or GLIP (Grounded Language-Image Pre-training). You can utilize the included text encoder.

The RPN and classifier of the image encoder that constitutes the object detection module 302 can be learned by querying the class name (or similar phrase expressing it) of the public dataset. In addition, the text corresponding to the class name (major category/subcategory) of the product that can be displayed in the target space can be used as a query for the text encoder that configures the NLP-based object classification module 71.

The object classification module 303 can detect and recognize the major/small classification of the object detected through the object detection module 302 using a query from a text encoder through the NLP-based object classification module 71. .

The image encoder constituting the object detection module 302 receives an RGB image taken of a target space where items are displayed as input and outputs information on the location and size of the item (object) detected in the image.

The text encoder that constitutes the NLP-based object classification module 71 inputs the class names (major category/subcategory text) of all items included in the public dataset (e.g., snack, drink, toy box, dairy, etc.) and images. The class name (major category/subcategory text) of the detected article (object) is output.

The object detection module 302 can detect all possible object areas from the image transmitted from the image sensing module 301. Image features can be created for each object area detected through an image encoder.

The NLP-based object classification module 71 can generate the class names of all items included in the public dataset as text features through a text encoder. At this time, the NLP-based object classification module 71 may compare the similarity between the image feature of the object and all text features for the object detected in the image and classify it according to the class name of the text feature with the highest similarity.

Furthermore, in this embodiment, POG data related to the target space can be used in object classification. POG data is used to manage goods in supermarkets and various stores.

Referring to FIG. 8, POG data 80 for product management may include product category (major category/subcategory), product name (product code), product location in space, product image, etc., where the product name for each product (81) and product image (82) can be built into a database. The product name 81 and product image 82 constructed in the database can be registered and managed in the gallery for product recognition in the object classification module 303. The category included in the POG data 80 corresponds to the class name of the product and can be used as an input to the NLP-based object classification module 71.

Figure 9 shows an example of a detailed object classification process using NLP-based object classification results and POG data in an embodiment of the present invention.

Referring to FIG. 9, the object classification module 303 compares the similarity between the feature vector of the object area detected in the image through the deep learning-based object detection module 302 and the feature vector of the product image registered in the gallery to determine the actual Detailed object classification, such as product name, can be performed.

For example, the object classification module 303 may limit the range of items to be compared in the POG gallery according to the NLP-based object classification results. For example, if an object detected in an image is classified as a snack through NLP-based object classification, the comparison range of the POG gallery can be limited to items that fall into the snack category.

As another example, the object classification module 303 can improve classification accuracy by reflecting the weight according to the NLP-based object classification result to the similarity. In other words, a weight according to the NLP-based object classification result for the object can be applied to the similarity between the feature vector of the object area detected in the image and the feature vector of the product image registered in the gallery. For example, if an object detected in an image is classified as a snack through NLP-based object classification, a greater weight can be applied to its similarity to a POG image classified in the snack category than to POG images in other categories. The final similarity of the POG image to the object image is It can be calculated as follows.

If the object detected in the image is classified as a snack as a result of NLP-based object classification, the object image can be compared with the POG image belonging to the snack category and the POG product name of the image most similar to the object image can be classified into the detailed class of the object.

While the existing technology performs detection/recognition of a limited class of items, the present embodiments can perform detection/recognition of tens of thousands of types of objects. These embodiments are a new pipeline capable of detecting/recognizing products that do not exist in public data, and can provide detection/recognition of new products using POG used for product management in stores.

As such, according to embodiments of the present invention, detection and classification of tens of thousands of objects or more are performed without class restrictions by classifying objects detected in images based on deep learning using NLP technology and POG data. can do. According to embodiments of the present invention, detection and classification of new objects can be performed by performing primary classification of an object based on NLP technology and then performing detailed classification of the object using the primary classification result and POG data. You can.

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 an object classification method running on a computer device,
The computer device includes at least one processor configured to execute computer readable instructions contained in a memory,
The object classification method is,
Detecting an object from an image of a target space through a deep learning model, by the at least one processor; and
Classifying the object using planogram (POG) data related to the target space, by the at least one processor.
Object classification method including. According to paragraph 1,
The detection step is,
Detecting object regions in the image using a region proposal network (RPN) model
An object classification method characterized by . According to paragraph 2,
The RPN model is a model in which a region estimate (region proposal) for items included in the target space is learned using public data related to the target space.
An object classification method characterized by . According to paragraph 1,
The classification step is,
registering the product name and product image for each product included in the POG data in a gallery for product recognition; and
Classifying the object into a product name registered in the gallery according to its similarity to the product image registered in the gallery.
Object classification method including. According to paragraph 1,
The classification step is,
Performing primary classification on the object through a natural language processing (NLP)-based object classification model; and
Performing detailed classification of the object using the primary classification result and the POG data
Object classification method including. According to clause 5,
The NLP-based object classification model includes a transformer-based text encoder.
An object classification method characterized by . According to clause 5,
The step of performing primary classification for the object is,
Recognizing the class name corresponding to the object by comparing the image feature of the object with the text feature of the class name included in public data
Object classification method including. According to clause 5,
The step of performing detailed classification of the object is,
registering the product name and product image for each product included in the POG data in a gallery for product recognition;
Comparing product images in a range corresponding to the primary classification result among product images registered in the gallery with the object; and
Classifying the object into a product name registered in the gallery according to the similarity with the product image compared with the object.
Object classification method including. According to clause 5,
The step of performing detailed classification of the object is,
registering the product name and product image for each product included in the POG data in a gallery for product recognition; and
Classifying the object into a product name registered in the gallery according to its similarity to the product image registered in the gallery.
Including,
The weight according to the primary classification result is reflected in the similarity.
An object classification method characterized by . According to paragraph 1,
The object classification method is,
Acquiring, by the at least one processor, the image of the target space captured at a certain period or in real time through an image sensing module consisting of an RGB sensor.
An object classification method further comprising: A computer program stored in a computer-readable recording medium for executing the object classification 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,
The process of detecting an object from an image of the target space through a deep learning model; and
The process of classifying the object using POG data related to the target space
A computer device that processes . According to clause 12,
The at least one processor,
Detecting object areas in the image using the RPN model
A computer device characterized by a. According to clause 13,
The RPN model is a model in which area estimation for items included in the target space is learned using public data related to the target space.
A computer device characterized by a. According to clause 12,
The at least one processor,
Register the product name and product image for each product included in the POG data in the gallery for product recognition,
Classifying the object by the product name registered in the gallery according to its similarity to the product image registered in the gallery.
A computer device characterized by a. According to clause 12,
The at least one processor,
Perform primary classification on the object through an NLP-based object classification model,
Performing detailed classification of the object using the primary classification result and the POG data
A computer device characterized by a. According to clause 16,
The NLP-based object classification model includes a transformer-based text encoder.
A computer device characterized by a. According to clause 16,
The at least one processor,
Recognizing the class name corresponding to the object by comparing the image feature of the object with the text feature of the class name included in public data
A computer device characterized by a. According to clause 16,
The at least one processor,
Register the product name and product image for each product included in the POG data in the gallery for product recognition,
Among the product images registered in the gallery, compare product images in a range corresponding to the primary classification result with the object,
Classifying the object by the product name registered in the gallery according to the similarity with the product image compared with the object.
A computer device characterized by a. According to clause 16,
The at least one processor,
Register the product name and product image for each product included in the POG data in the gallery for product recognition,
Classifying the object by the product name registered in the gallery according to its similarity to the product image registered in the gallery,
The weight according to the primary classification result is reflected in the similarity.
A computer device characterized by a.

Images