KR102548129B1 - Defect detection method using feature map selection includes spatial importance of image. - Google Patents

Abstract

Disclosed according to an embodiment of the present disclosure is a method for predicting whether an article is defective, performed by one or more processors of a computing device. The method comprises the steps of: acquiring one or more images; inputting the acquired images into a neural network model and generating a plurality of feature maps; extracting a first feature vector, based on the plurality of feature maps; extracting a second feature vector for identifying a feature map related to whether there is a defect, among the plurality of feature maps, based on the extracted first feature vector; extracting a third feature vector for identifying a feature map region related to whether there is a defect, based on the plurality of feature maps; and predicting whether there is a defect, based on the first feature vector, the second feature vector, and the third feature vector, by using the neural network model. Therefore, the method can use 2D image information when inspecting a defect in the article, thereby inspecting the defect in real time.

Description

Defect detection method using feature map screening including spatial importance of images {DEFECT DETECTION METHOD USING FEATURE MAP SELECTION INCLUDES SPATIAL IMPORTANCE OF IMAGE.}

The present disclosure relates to a method for inspecting a defect in an article using 2D image information, and more particularly, to a method for inspecting a defect in real time using 2D image information when inspecting a defect in an article.

In the case of inspecting defects of an article using a conventional 3D camera, it is difficult to use it to inspect defects in real time due to high price and long processing time.

In addition, in the past, human workers have been used to inspect quality when inspecting defects of an article, but it is difficult to perform a high-level precision inspection process during the inspection time due to eye strain, and it is difficult to continue performing accurate work. In order to ensure quality of inspection, each worker had to have sufficient experience in analyzing items. Therefore, there has been a problem that production efficiency is lowered when quality is inspected using human workers.

Therefore, considering the high price, long processing time, and consistency of quality inspection, which were problems in the existing product defect inspection, a method to ensure consistency of quality inspection while reducing the processing time and cost of defect inspection is needed. And a new technology that can solve these problems or disadvantages is required.

On the other hand, the present disclosure has been derived at least based on the technical background salpin above, but the technical problem or object of the present disclosure is not limited to solving the above salpin problems or disadvantages. That is, the present disclosure may cover various technical issues related to the content to be described below, in addition to the technical issues discussed above.

An object of the present disclosure is to inspect defects in real time by utilizing 2D image information when inspecting defects of an article.

On the other hand, the technical problem to be achieved by the present disclosure is not limited to the above-mentioned technical problem, and may include various technical problems within a range apparent to those skilled in the art from the contents to be described below.

A method performed by a computing device according to an embodiment of the present disclosure for realizing the above object is disclosed. The method includes acquiring one or more images; inputting the obtained image to a neural network model and generating a plurality of feature maps; extracting a first feature vector based on the plurality of feature maps; extracting a second feature vector for identifying a feature map related to whether or not there is a defect among the plurality of feature maps, based on the extracted first feature vector; extracting a third feature vector for identifying a feature map region related to a defect, based on the plurality of feature maps; and predicting whether there is a defect based on the first feature vector, the second feature vector, and the third feature vector by using the neural network model.

Alternatively, the one or more images include one or more 2D images, and the acquiring of the one or more images may include a front image, a back image, a right image, a left image, a top image, or a bottom image of the article. It may include acquiring one or more images.

Alternatively, the step of inputting the acquired image to the neural network model and generating a plurality of feature maps may include: performing gamma correction on the acquired image; and inputting the gamma-corrected image to the neural network model.

Alternatively, the step of inputting the acquired image to the neural network model and generating a plurality of feature maps may include a plurality of first convolution layers and a plurality of first pooling layers included in the neural network model ( and extracting features of the image using a pooling layer) and generating a plurality of feature maps based on the extracted features.

Alternatively, extracting a first feature vector based on the plurality of feature maps includes extracting the first feature vector by performing additional pooling on the plurality of feature maps. can do.

Alternatively, the step of extracting a second feature vector for identifying a feature map related to whether or not there is a defect among the plurality of feature maps, based on the extracted first feature vector, may include the first feature vector as one or more The method may include extracting a second feature vector for inputting to a first fully-connected layer and identifying a feature map related to whether or not there is a defect among the plurality of feature maps.

Alternatively, extracting a third feature vector for identifying a feature map region related to whether or not a defect is present based on the plurality of feature maps includes:

inputting the plurality of feature maps to one or more second convolution layers and applying a convolution operation; and performing additional pooling on the feature maps to which the convolution operation is applied, and extracting a third feature vector for identifying a feature map region related to whether or not there is a defect.

Alternatively, the additional pooling may include global average pooling.

Alternatively, predicting whether there is a defect based on the first feature vector, the second feature vector, and the third feature vector using the neural network model may include the second feature vector and the third feature vector. generating a first composite vector by concatenating feature vectors; generating a final composite vector by concatenating the first composite vector and the first feature vector; and predicting whether or not there is a defect based on the final synthesized vector.

Alternatively, predicting whether or not there is a defect based on the final synthesized vector may include inputting the final synthesized vector to one or more second fully-connected layers and predicting whether or not there is a defect. can

A computer program stored in a computer readable storage medium is disclosed according to an embodiment of the present disclosure for realizing the above object. When the computer program is executed on one or more processors, the one or more processors cause the one or more processors to perform operations for predicting whether an article is defective, the operations comprising: obtaining one or more images; inputting the obtained image to a neural network model and generating a plurality of feature maps; extracting a first feature vector based on the plurality of feature maps; extracting a second feature vector for identifying a feature map related to whether or not there is a defect among the plurality of feature maps, based on the extracted first feature vector; extracting a third feature vector for identifying a feature map region related to a defect, based on the plurality of feature maps; and predicting whether there is a defect based on the first feature vector, the second feature vector, and the third feature vector by using the neural network model.

A computing device according to an embodiment of the present disclosure for realizing the above object is disclosed. The device may include at least one processor; and a memory, wherein the processor: acquires one or more images;

inputting the obtained image to a neural network model and generating a plurality of feature maps; extracting a first feature vector based on the plurality of feature maps; extracting a second feature vector for identifying a feature map related to whether or not there is a defect among the plurality of feature maps, based on the extracted first feature vector; based on the plurality of feature maps, extracting a third feature vector for identifying a feature map region related to whether or not there is a defect; And, using the neural network model, it may be configured to predict whether there is a defect based on the first feature vector, the second feature vector, and the third feature vector.

The present disclosure may provide a method of inspecting a defect of an article using 2D image information, and through this, when inspecting a defect of an article, the defect may be inspected in real time by utilizing the 2D image information.

On the other hand, the effects of the present disclosure are not limited to the effects mentioned above, and various effects may be included within a range apparent to those skilled in the art from the contents to be described below.

1 is a block diagram of a computing device for inspecting a defect of an article using 2D image information according to an embodiment of the present disclosure.
2 is a schematic diagram illustrating a network function according to an embodiment of the present disclosure.
3 is a flowchart illustrating a method of inspecting a defect of an article using 2D image information according to an embodiment of the present disclosure.
4 is a schematic diagram illustrating a process of acquiring one or more images and performing gamma correction on the obtained images according to an embodiment of the present disclosure.
5 shows an image acquired in a neural network model according to an embodiment of the present disclosure and using a plurality of first convolution layers and a plurality of first pooling layers included in the neural network model. It is a schematic diagram showing a process of extracting features of an image and generating a plurality of feature maps based on the extracted features.
6 is a schematic diagram illustrating a process of extracting a first feature vector, a second feature vector, and a third feature vector based on a plurality of feature maps according to an embodiment of the present disclosure.
7A is a schematic diagram illustrating a process of extracting a second feature vector for identifying a feature map related to whether or not there is a defect among a plurality of feature maps based on a first feature vector extracted according to an embodiment of the present disclosure.
FIG. 7B is a schematic diagram illustrating a process of extracting a third feature vector for identifying a feature map region related to a defect based on a plurality of feature maps according to an embodiment of the present disclosure.
8 is a schematic diagram illustrating a process of predicting defects based on a first feature vector, a second feature vector, and a third feature vector using a neural network model according to an embodiment of the present disclosure.
9 is a simplified and general schematic diagram of an exemplary computing environment in which embodiments of the present disclosure may be implemented.

Various embodiments are now described with reference to the drawings. In this specification, various descriptions are presented to provide an understanding of the present disclosure. However, it is apparent that these embodiments may be practiced without these specific details.

The terms “component,” “module,” “system,” and the like, as used herein, refer to a computer-related entity, hardware, firmware, software, a combination of software and hardware, or an execution of software. For example, a component may be, but is not limited to, a procedure, processor, object, thread of execution, program, and/or computer running on a processor. For example, both an application running on a computing device and a computing device may be components. One or more components may reside within a processor and/or thread of execution. A component can be localized within a single computer. A component may be distributed between two or more computers. Also, these components can execute from various computer readable media having various data structures stored thereon. Components may be connected, for example, via signals with one or more packets of data (e.g., data and/or signals from one component interacting with another component in a local system, distributed system) to other systems and over a network such as the Internet. data being transmitted) may communicate via local and/or remote processes.

In addition, the term “or” is intended to mean an inclusive “or” rather than an exclusive “or”. That is, unless otherwise specified or clear from the context, “X employs A or B” is intended to mean one of the natural inclusive substitutions. That is, X uses A; X uses B; Or, if X uses both A and B, "X uses either A or B" may apply to either of these cases. Also, the term "and/or" as used herein should be understood to refer to and include all possible combinations of one or more of the listed related items.

Also, the terms "comprises" and/or "comprising" should be understood to mean that the features and/or components are present. However, it should be understood that the terms "comprises" and/or "comprising" do not exclude the presence or addition of one or more other features, elements, and/or groups thereof. Also, unless otherwise specified or where the context clearly indicates that a singular form is indicated, the singular in this specification and claims should generally be construed to mean "one or more".

In addition, the term “at least one of A or B” should be interpreted as meaning “when only A is included”, “when only B is included”, and “when A and B are combined”.

Skilled artisans will further understand that the various illustrative logical blocks, components, modules, circuits, means, logics, and algorithm steps described in connection with the embodiments disclosed herein may be implemented in electronic hardware, computer software, or both. It should be recognized that it can be implemented with combinations of To clearly illustrate the interchangeability of hardware and software, various illustrative components, blocks, configurations, means, logics, modules, circuits, and steps have been described above generally in terms of their functionality. Whether such functionality is implemented in hardware or as software depends on the particular application and design constraints imposed on the overall system. Skilled artisans may implement the described functionality in varying ways for each particular application. However, such implementation decisions should not be interpreted as causing a departure from the scope of this disclosure.

The description of the presented embodiments is provided to enable any person skilled in the art to use or practice the present invention. Various modifications to these embodiments will be apparent to those skilled in the art of this disclosure. The general principles defined herein may be applied to other embodiments without departing from the scope of this disclosure. Thus, the present invention is not limited to the embodiments presented herein. The present invention is to be accorded the widest scope consistent with the principles and novel features set forth herein.

In the present disclosure, network functions, artificial neural networks, and neural networks may be used interchangeably.

1 is a block diagram of a computing device for inspecting a defect of an article using 2D image information according to an embodiment of the present disclosure.

The configuration of the computing device 100 shown in FIG. 1 is only a simplified example. In one embodiment of the present disclosure, the computing device 100 may include other components for performing a computing environment of the computing device 100, and only some of the disclosed components may constitute the computing device 100.

The computing device 100 may include a processor 110 , a memory 130 , and a network unit 150 .

The processor 110 may include one or more cores, and includes a central processing unit (CPU), a general purpose graphics processing unit (GPGPU), and a tensor processing unit (TPU) of a computing device. unit), data analysis, and processors for deep learning. The processor 110 may read a computer program stored in the memory 130 and process data for machine learning according to an embodiment of the present disclosure. According to an embodiment of the present disclosure, the processor 110 may perform an operation for learning a neural network model. The processor 110 processes input data for learning in deep learning (DL), extracts features from input data, calculates an error, and learns a neural network model such as weight update of the neural network model using backpropagation. Calculations can be performed for At least one of the CPU, GPGPU, and TPU of the processor 110 may process learning of the neural network model. For example, the CPU and GPGPU can process learning of neural network models and data classification using neural network models. In addition, in an embodiment of the present disclosure, the neural network model learning and data classification using the neural network model may be processed by using processors of a plurality of computing devices together. In addition, a computer program executed in a computing device according to an embodiment of the present disclosure may be a CPU, GPGPU or TPU executable program.

According to an embodiment of the present disclosure, the memory 130 may store any type of information generated or determined by the processor 110 and any type of information received by the network unit 150 .

According to an embodiment of the present disclosure, the memory 130 is a flash memory type, a hard disk type, a multimedia card micro type, or a card type memory (eg, SD or XD memory, etc.), RAM (Random Access Memory, RAM), SRAM (Static Random Access Memory), ROM (Read-Only Memory, ROM), EEPROM (Electrically Erasable Programmable Read-Only Memory), PROM (Programmable Read-Only Memory) -Only Memory), a magnetic memory, a magnetic disk, and an optical disk may include at least one type of storage medium. The computing device 100 may operate in relation to a web storage that performs a storage function of the memory 130 on the Internet. The above description of the memory is only an example, and the present disclosure is not limited thereto.

The network unit 150 according to an embodiment of the present disclosure includes a Public Switched Telephone Network (PSTN), x Digital Subscriber Line (xDSL), Rate Adaptive DSL (RADSL), Multi Rate DSL (MDSL), and VDSL ( Various wired communication systems such as Very High Speed DSL), Universal Asymmetric DSL (UADSL), High Bit Rate DSL (HDSL), and Local Area Network (LAN) may be used.

In addition, the network unit 150 presented in the present disclosure includes Code Division Multi Access (CDMA), Time Division Multi Access (TDMA), Frequency Division Multi Access (FDMA), Orthogonal Frequency Division Multi Access (OFDMA), SC-FDMA ( Single Carrier-FDMA) and other systems.

In the present disclosure, the network unit 150 may be configured regardless of its communication mode, such as wired and wireless, and may be configured with various communication networks such as a personal area network (PAN) and a wide area network (WAN). can In addition, the network may be the known World Wide Web (WWW), or may use a wireless transmission technology used for short-range communication, such as Infrared Data Association (IrDA) or Bluetooth. The techniques described in this disclosure may also be used in other networks mentioned above.

2 is a schematic diagram illustrating a network function according to an embodiment of the present disclosure.

Throughout this specification, computational model, neural network, network function, and neural network may be used interchangeably. A neural network may consist of a set of interconnected computational units, which may generally be referred to as nodes. These nodes may also be referred to as neurons. A neural network includes one or more nodes. Nodes (or neurons) constituting neural networks may be interconnected by one or more links.

In a neural network, one or more nodes connected through a link may form a relative relationship of an input node and an output node. The concept of an input node and an output node is relative, and any node in an output node relationship with one node may have an input node relationship with another node, and vice versa. As described above, an input node to output node relationship may be created around a link. More than one output node can be connected to one input node through a link, and vice versa.

In a relationship between an input node and an output node connected through one link, the value of data of the output node may be determined based on data input to the input node. Here, a link interconnecting an input node and an output node may have a weight. The weight may be variable, and may be changed by a user or an algorithm in order to perform a function desired by the neural network. For example, when one or more input nodes are interconnected by respective links to one output node, the output node is set to a link corresponding to values input to input nodes connected to the output node and respective input nodes. An output node value may be determined based on the weight.

As described above, in the neural network, one or more nodes are interconnected through one or more links to form an input node and output node relationship in the neural network. Characteristics of the neural network may be determined according to the number of nodes and links in the neural network, an association between the nodes and links, and a weight value assigned to each link. For example, when there are two neural networks having the same number of nodes and links and different weight values of the links, the two neural networks may be recognized as different from each other.

A neural network may be composed of a set of one or more nodes. A subset of nodes constituting a neural network may constitute a layer. Some of the nodes constituting the neural network may form one layer based on distances from the first input node. For example, a set of nodes having a distance of n from the first input node may constitute n layers. The distance from the first input node may be defined by the minimum number of links that must be passed through to reach the corresponding node from the first input node. However, the definition of such a layer is arbitrary for explanation, and the order of a layer in a neural network may be defined in a method different from the above. For example, a layer of nodes may be defined by a distance from a final output node.

An initial input node may refer to one or more nodes to which data is directly input without going through a link in relation to other nodes among nodes in the neural network. Alternatively, in a relationship between nodes based on a link in a neural network, it may mean nodes that do not have other input nodes connected by a link. Similarly, the final output node may refer to one or more nodes that do not have an output node in relation to other nodes among nodes in the neural network. Also, the hidden node may refer to nodes constituting the neural network other than the first input node and the last output node.

In the neural network according to an embodiment of the present disclosure, the number of nodes in the input layer may be the same as the number of nodes in the output layer, and the number of nodes decreases and then increases again as the number of nodes progresses from the input layer to the hidden layer. can In addition, the neural network according to another embodiment of the present disclosure may be a neural network in which the number of nodes of the input layer may be less than the number of nodes of the output layer and the number of nodes decreases as the number of nodes increases from the input layer to the hidden layer. there is. In addition, the neural network according to another embodiment of the present disclosure is a neural network in which the number of nodes in the input layer may be greater than the number of nodes in the output layer, and the number of nodes increases as the number of nodes increases from the input layer to the hidden layer. can A neural network according to another embodiment of the present disclosure may be a neural network in the form of a combination of the aforementioned neural networks.

A deep neural network (DNN) may refer to a neural network including a plurality of hidden layers in addition to an input layer and an output layer. Deep neural networks can reveal latent structures in data. In other words, it can identify the latent structure of a photo, text, video, sound, or music (e.g., what objects are in the photo, what the content and emotion of the text are, what the content and emotion of the audio are, etc.). . Deep neural networks include convolutional neural networks (CNNs), recurrent neural networks (RNNs), auto encoders, generative adversarial networks (GANs), and restricted boltzmann machines (RBMs). machine), a deep belief network (DBN), a Q network, a U network, a Siamese network, a Generative Adversarial Network (GAN), and the like. The description of the deep neural network described above is only an example, and the present disclosure is not limited thereto.

In one embodiment of the present disclosure, the network function may include an autoencoder. An autoencoder may be a type of artificial neural network for outputting output data similar to input data. An auto-encoder may include at least one hidden layer, and an odd number of hidden layers may be disposed between input and output layers. The number of nodes of each layer may be reduced from the number of nodes of the input layer to an intermediate layer called the bottleneck layer (encoding), and then expanded symmetrically with the reduction from the bottleneck layer to the output layer (symmetrical to the input layer). Autoencoders can perform non-linear dimensionality reduction. The number of input layers and output layers may correspond to dimensions after preprocessing of input data. In the auto-encoder structure, the number of hidden layer nodes included in the encoder may decrease as the distance from the input layer increases. If the number of nodes in the bottleneck layer (the layer with the fewest nodes located between the encoder and decoder) is too small, a sufficient amount of information may not be conveyed, so more than a certain number (e.g., more than half of the input layer, etc.) ) may be maintained.

The neural network may be trained using at least one of supervised learning, unsupervised learning, semi-supervised learning, and reinforcement learning. Learning of the neural network may be a process of applying knowledge for the neural network to perform a specific operation to the neural network.

A neural network can be trained in a way that minimizes output errors. In the learning of the neural network, the learning data is repeatedly input into the neural network, the output of the neural network for the training data and the error of the target are calculated, and the error of the neural network is transferred from the output layer of the neural network to the input layer in the direction of reducing the error. It is a process of updating the weight of each node of the neural network by backpropagating in the same direction. In the case of teacher learning, the learning data in which the correct answer is labeled is used for each learning data (ie, the labeled learning data), and in the case of comparative teacher learning, the correct answer may not be labeled in each learning data. That is, for example, learning data in the case of teacher learning regarding data classification may be data in which each learning data is labeled with a category. Labeled training data is input to a neural network, and an error may be calculated by comparing an output (category) of the neural network and a label of the training data. As another example, in the case of comparative history learning for data classification, an error may be calculated by comparing input learning data with a neural network output. The calculated error is back-propagated in a reverse direction (ie, from the output layer to the input layer) in the neural network, and the connection weight of each node of each layer of the neural network may be updated according to the back-propagation. The amount of change in the connection weight of each updated node may be determined according to a learning rate. The neural network's computation of input data and backpropagation of errors can constitute a learning cycle (epoch). The learning rate may be applied differently according to the number of iterations of the learning cycle of the neural network. For example, a high learning rate may be used in the early stage of neural network training to increase efficiency by allowing the neural network to quickly obtain a certain level of performance, and a low learning rate may be used in the late stage to increase accuracy.

In neural network learning, generally, training data can be a subset of real data (ie, data to be processed using the trained neural network), and therefore, errors for training data are reduced, but errors for real data are reduced. There may be incremental learning cycles. Overfitting is a phenomenon in which errors for actual data increase due to excessive learning on training data. For example, a phenomenon in which a neural network that has learned a cat by showing a yellow cat does not recognize that it is a cat when it sees a cat other than yellow may be a type of overfitting. Overfitting can act as a cause of increasing the error of machine learning algorithms. Various optimization methods can be used to prevent such overfitting. To prevent overfitting, methods such as increasing the training data, regularization, inactivating some nodes in the network during learning, and using a batch normalization layer should be applied. can

3 is a flowchart illustrating a method of inspecting a defect of an article using 2D image information according to an embodiment of the present disclosure.

The computing device 100 according to an embodiment of the present disclosure may directly acquire “image information for inspecting a defect of an article” or may receive it from an external system. The external system may be a server or database that stores and manages image information. The computing device 100 may use image information obtained directly or received from an external system as “input data for inspecting defects of an article”.

The computing device 100 may acquire one or more images (S110). For example, one or more images may include an image of an item to inspect for defects. Specifically, the image of the article to be inspected for defects may include images of fabric, bottle, vehicle body surface, metal, and the like. Also, the one or more images may include images of a three-dimensional (3D) mechanical product such as a ball joint socket, and may include one or more 2D images. Meanwhile, according to an embodiment of the present disclosure, the computing device 100 may acquire one or more images of a front image, a back image, a right image, a left image, a top image, or a bottom image of an article. A specific process of acquiring one or more images will be described later with reference to FIG. 4 .

The computing device 100 may input the image obtained through step S110 to the neural network model and generate a plurality of feature maps (S120). For example, the computing device 100 may perform gamma correction on the obtained image and input the gamma-corrected image to the neural network model. At this time, when the acquired image is an image of a metal material, since the light reflection value of the surface may be different, the light intensity of the obtained image is reduced through gamma correction to obtain a clear image of the article without light reflection. can do. A detailed process of performing gamma correction on the image obtained through step S110 and obtaining a corrected image will be described later with reference to FIG. 4 .

Also, the computing device 100 may input the gamma-corrected image to the neural network model and generate a plurality of feature maps. For example, the computing device 100 extracts the features of the image using a plurality of first convolution layers and a plurality of first pooling layers included in the neural network model, and the extracted A plurality of feature maps may be generated based on the features. Specifically, the computing device 100 inputs the gamma-corrected image to a structure in which a plurality of first convolution layers and a plurality of first pooling layers are alternately connected, and the characteristics of the image may be extracted, and a plurality of feature maps may be generated based on the extracted features. In this case, the plurality of first convolutional layers may be used to extract image features required for classification from an input image. For example, a convolution operation is performed while passing an image through a plurality of first convolution layers, and a plurality of first pooling layers alternately connected to the plurality of first convolution layers for the passed image. Max pooling may be performed to extract features of the input image, and a plurality of feature maps may be generated based on the features of the extracted image. At this time, the reason for performing max pooling is that ① the largest feature value has the greatest influence on the calculation, so it has the greatest effect on the output value and ② represents the feature the best. Extracting features of an image using a plurality of first convolution layers and a plurality of first pooling layers included in a neural network model, and generating a plurality of feature maps based on the extracted features A specific process will be described later with reference to FIG. 5 .

The computing device 100 may extract a first feature vector based on the plurality of feature maps generated through step S120 (S130-1). For example, the computing device 100 may extract the first feature vector by performing additional pooling on the plurality of feature maps. Specifically, the computing device 100 may perform global average pooling on the plurality of feature maps and extract the first feature vector. At this time, by performing global average pooling for each channel on the plurality of feature maps, one feature map of the size of HxW can be converged to one value. The plurality of feature maps may be extracted as one first feature vector. The extracted first feature vector may be used in a process of extracting a second feature vector for identifying a feature map related to a defect among a plurality of feature maps, and a process of predicting a defect by using a neural network model It can be used in, and a detailed description thereof will be described later through FIGS. 6, 7a, and 8.

The computing device 100 may extract a second feature vector for identifying a feature map related to whether or not there is a defect among a plurality of feature maps based on the first feature vector extracted through step S130-1 (S130-1). 2). Specifically, the computing device 100 inputs the first feature vector to one or more first fully-connected layers, and identifies a feature map related to whether there is a defect among the plurality of feature maps. 2 feature vectors can be extracted. For example, global average pooling is performed on the plurality of feature maps to remove spatial information, and the first feature vector with only channel information remaining is converted into two first fully-connected layers. , the importance level for each channel is learned, and a second feature vector for identifying a feature map most related to whether or not there is a defect among the plurality of feature maps can be extracted. The extracted second feature vector may be used in a process of predicting defects using a neural network model, and a detailed description thereof will be described later with reference to FIGS. 6, 7A, and 8 .

The computing device 100 may extract a third feature vector for identifying a feature map region related to a defect, based on the plurality of feature maps generated in step S120 (S130-3). Specifically, the computing device 100 inputs the plurality of feature maps to one or more second convolution layers, applies a convolution operation, and performs additional pooling on the feature maps to which the convolution operation is applied. and a third feature vector for identifying a feature map region related to whether or not there is a defect may be extracted. For example, the computing device 100 performs global average pooling on the feature maps to which the convolution operation is applied, and extracts a third feature vector for identifying a feature map region related to whether or not there is a defect. can At this time, the computing device 100 inputs the plurality of feature maps to one or more second convolution layers, and deletes channel information by applying a convolution operation to obtain a feature map having spatial importance. A third feature vector for identifying a feature map region related to whether or not there is a defect may be extracted by performing global average pooling on the generated feature map having spatial importance. The extracted third feature vector may be used in a process of predicting defects using a neural network model, and a detailed description thereof will be described later with reference to FIGS. 6, 7B, and 8 .

The computing device 100 uses the neural network model to generate the first feature vector generated through step S130-1, the second feature vector generated through step S130-2, and the third feature generated through step S130-3. Based on the vector, it is possible to predict whether or not there is a defect (S140). For example, the computing device 100 generates a first synthesized vector by concatenating the second feature vector and the third feature vector, and concatenates the first synthesized vector and the first feature vector ( concatenate) to generate a final synthesized vector, and based on the final synthesized vector, it is possible to predict whether or not there is a defect. Specifically, the computing device 100 may input the final synthesized vector to one or more second fully-connected layers and predict whether or not there is a defect. A detailed description of a process of predicting defects based on the final synthesized vector will be described later with reference to FIG. 8 .

4 is a schematic diagram illustrating a process of acquiring one or more images and performing gamma correction on the obtained images according to an embodiment of the present disclosure.

Referring to FIG. 4 , a computing device 100 according to an embodiment of the present disclosure may acquire one or more images 12 of an item 10 in order to inspect the item for defects (11). Specifically, the image of the object to be inspected for defects may include images of fabric, bottle, vehicle body surface, metal, and the like. Also, the one or more images may include images of a three-dimensional (3D) mechanical product such as a ball joint socket, and may include one or more 2D images. On the other hand, the defect inspection process of 3D mechanical products such as ball joint sockets has a more difficult problem because defects can occur in any part of the 3D product. However, the cost is much higher than that of 2D camera images, and since various preprocessing processes such as 3D data point processing have to be performed, it takes a lot of work time compared to using 2D camera images, so it is difficult to inspect defects in real time. difficult to apply In contrast, the computing device 100 according to an embodiment of the present disclosure may display images of all parts of a 3D object from various angles, such as a front image, a rear image, a right image, a left image, a top image, or a bottom image of an object. It is possible to take a picture with a 2D camera (11) and acquire one or more images (12). At this time, the processing time of 2D images is shorter than that of 3D camera data, so it can be applied to defect inspection in real time, and the cost is relatively low because the preprocessing process is relatively simple compared to 3D camera data.

The computing device 100 according to an embodiment of the present disclosure may use one or more images 12 directly acquired or received from an external system as input data for inspecting defects of an article. Meanwhile, the one or more images 12 are not limited to a front image, a back image, a right image, a left image, a top image, or a bottom image of an article, and images taken from various angles in addition to these examples may be included. In addition, the article 10 is not limited to a ball joint socket, and other than these examples, various three-dimensional articles to be inspected for defects may be included.

The computing device 100 may perform gamma correction on the one or more images 12 (13). Meanwhile, when an article to be inspected for defects is a metal product, the light reflectance of the article may be different for each part of the surface, and the balance between the light source and the intensity may be a problem. For example, if the intensity of the light source is high, the reflectance of the shiny part of the metal article is high compared to the part where light reflection is low, and if the intensity of the light source is low, it is difficult to capture the part of the metal article where light reflection is small. Therefore, according to an embodiment of the present disclosure, the computing device 100 may first reduce the intensity of light in order to check whether there is no light reflection in the shiny part of the article, and increase the brightness of the one or more images 12 obtained above. By performing gamma correction (13) for this, it is possible to obtain a gamma-corrected image 14 of a clear object without reflection of light. Meanwhile, according to another embodiment of the present disclosure, the computing device 100 may adjust the size of the obtained one or more images 12 . For example, when the front image of the first item and the front image of the second item have different sizes, the computing device 100 may adjust the two images to the same size. At this time, if the difference between the original image and the resized image is excessively large, the defect inspection task may become much more difficult because small defects of the article may hardly be expressed in the resized image. Therefore, the size of the image may need to be adjusted so that the difference between the original image and the resized image does not become excessively large, and the size to be adjusted may be designated as 380x380 (Height x Width). The size to which the image is adjusted for the one or more acquired images 12 may be experimentally selected, and the size of 380x380 (Height x Width) is only an example, and for the one or more acquired images 12 Adjustments of various sizes can be performed. Meanwhile, the gamma-corrected image 14 or the resized image may be input to a neural network model and used in a process of generating a plurality of feature maps, and a detailed process will be described with reference to FIG. 5 below.

5 shows an image acquired in a neural network model according to an embodiment of the present disclosure and using a plurality of first convolution layers and a plurality of first pooling layers included in the neural network model. It is a schematic diagram showing a process of extracting features of an image and generating a plurality of feature maps based on the extracted features.

Referring to FIG. 5 , the computing device 100 inputs the gamma-corrected image 14 to the neural network model 20 and uses a plurality of first convolution layers and a plurality of first convolution layers included in the neural network model. Features of the image 14 may be extracted using a pooling layer 21, and a plurality of feature maps 22 may be generated based on the extracted features. Specifically, the computing device 100 inputs the gamma-corrected image 14 to a structure 21 in which a plurality of first convolution layers and a plurality of first pooling layers are alternately connected. In addition, features of the image 14 may be extracted, and a plurality of feature maps 14 may be generated based on the extracted features. In this case, the plurality of first convolutional layers may be used to extract image features required for classification from an input image. For example, a convolution operation is performed by passing the image 14 through a plurality of first convolution layers, and a plurality of first pooling layers (Pooling Layer) to perform max pooling, thereby extracting features of the input image 14, and generating a plurality of feature maps 22 based on the features of the extracted image. there is. At this time, the reason for performing max pooling is that ① the largest feature value has the greatest influence on the calculation, so it has the greatest influence on the output value, and ② it represents the characteristics of the input data best. However, in one embodiment of the present disclosure, performing max pooling by passing the passed image through a plurality of first pooling layers alternately connected to a plurality of first convolutional layers is only an example, In the process of passing the image through a plurality of first pooling layers alternately connected to a plurality of first convolutional layers, various pooling such as average pooling may be performed in addition to max pooling. On the other hand, as the input data is passed through a plurality of first pooling layers alternately connected to the plurality of first convolution layers, the dimension of the feature map output from the plurality of first convolution layers is lowered. The processing time of the model can be reduced. For example, the computing device 100 may pass the input image 14 through a structure in which seven first convolution layers and seven first pooling layers are alternately connected. At this time, in order to obtain a useful local pattern in the input image 14, 32 filters having a dimension of 3x3 and a stride of 1x1 can be used in the first first convolution layer among the seven, , In the second to fifth first convolutional layers of the seven, 64 filters having a dimension of 3x3 and a stride of 1x1 may be used, and the sixth to last of the seven In one convolutional layer, 128 filters with a dimension of 3x3 and a stride (movement amount of the filter) of 1x1 can be used. On the other hand, all activation functions of the plurality of first convolutional layers may use ReLU (rectified linear unit), the plurality of first pooling layers may be max pooling layers, and the plurality of first pooling layers may each first convolution A max pooling layer having a pool size and a stride size of 2x2 applied next to the layer may be included. For example, when the input tensor shape of the input image 14 is 380x380x3 (height x width x channel), the 7 first convolution layers and the 7 first pooling layers are passed through a structure that is alternately connected. The output tensor shape of the generated feature maps 22 may be in the form of 2x2x128 (height x width x channel). The generated plurality of feature maps 22 include features necessary for defect inspection in the input image 14, but when the defect of the product is very small, the similarity between the image with defects and the image without defects is very high. When additional features are used in the input image 14, the accuracy of the neural network model predicting defects of an article may be increased. Therefore, the computing device 100, based on the plurality of feature maps 22, uses a first feature vector, a second feature vector for identifying a feature map related to whether or not there is a defect among the plurality of feature maps, and whether or not there is a defect. A third feature vector for identifying a feature map region related to may be extracted, and a detailed process will be described below with reference to FIGS. 6 to 7B.

6 is a schematic diagram illustrating a process of extracting a first feature vector, a second feature vector, and a third feature vector based on a plurality of feature maps according to an embodiment of the present disclosure.

Referring to FIG. 6 , according to an embodiment of the present disclosure, the computing device 100 performs (30) global average pooling on the plurality of feature maps 22, and generates a first feature vector ( 31) can be extracted. At this time, by performing global average pooling for each channel on the plurality of feature maps 22, one feature map having a height x width size of 2x2 is obtained. Since can converge to one value, the plurality of feature maps 22 can be extracted as one first feature vector 31 . For example, when the plurality of feature maps 22 correspond to feature maps having a size of 2x2 that exist for each 128 channels, performing global average pooling on a tensor shape of 2x2x128 (Height x Width x Channel) A first feature vector 31 in a one-dimensional form having 128 values may be extracted. The extracted first feature vector 31 may be used in a process 40 of extracting a second feature vector 41 for identifying a feature map related to whether or not there is a defect among a plurality of feature maps 22, Using the neural network model 20, it can be used in the process of predicting whether or not there is a defect. A detailed description of the process 40 of extracting the second feature vector 41 will be described below with reference to FIG. 7A.

Meanwhile, referring to FIG. 6 , the computing device 100 may extract (50) a third feature vector 51 based on the plurality of feature maps 22 . The extracted third feature vector 51 can be used in a process of predicting defects using a neural network model, and a detailed description of the process of extracting the third feature vector 51 will be described later with reference to FIG. 7B. do.

7A is a schematic diagram illustrating a process of extracting a second feature vector for identifying a feature map related to whether or not there is a defect among a plurality of feature maps based on a first feature vector extracted according to an embodiment of the present disclosure.

Referring to FIG. 7A , the computing device 100 inputs the extracted first feature vector 31 to one or more first fully connected layers, and selects a feature map related to defects among a plurality of feature maps 22. A second feature vector 41 for identification may be extracted. For example, global average pooling is performed on the plurality of feature maps 22 to remove spatial information, and the first feature vector 31 with only channel information remaining is converted into two first fully connected layers. (Fully-connected layer) 40-1, the importance level for each channel is learned, and a second feature vector for identifying a feature map most related to the defect among the plurality of feature maps 22 ( 41) can be extracted (40). For example, the second feature vector 41 may be calculated and extracted based on the following equation.

[Equation 1]

In Equation 1, FC1 may mean a first layer among the two first fully connected layers 40-1, and FC2 may mean a second layer among the two first fully connected layers 40-1. The hidden layer of FC1 can be set to 128, and the hidden layer of FC2 can be set to 64. Through this, the size of the second feature vector 41 can be extracted as 64. can Meanwhile, in the process of extracting the second feature vector 41 of FIG. 7A, the interdependency between channels is utilized to focus on required input feature points, and channel information is compressed using Global Average Pooling (GAP), Input feature points are restored through the convolution layer, and a channel attention process of reinforcing and restoring the input feature points for the part to be focused within the network may be included.

In addition, through the channel attention process, it is possible to extract a feature map most related to the presence or absence of a defect among the plurality of feature maps 22, and a second feature vector including information on the feature map most related to the presence or absence of a defect Through a process of concatenating (41) with another feature vector (first or third feature vector), the output value of channel attention may be applied in the defect prediction process. A detailed description of this will be described below with reference to FIG. 8 .

FIG. 7B is a schematic diagram illustrating a process of extracting a third feature vector for identifying a feature map region related to a defect based on a plurality of feature maps according to an embodiment of the present disclosure.

Referring to FIG. 7B, the computing device 100 inputs the generated plurality of feature maps 22 to one or more second convolutional layers 50-1, and identifies feature map regions related to defects. A third feature vector 51 for For example, the computing device 100 inputs the plurality of feature maps to three second convolution layers 50-1, and deletes channel information and spatial importance by applying a convolution operation. A feature map having a spatial significance may be generated, and global average pooling (50-2) is performed on the generated feature map having spatial importance to identify a feature map region related to whether or not there is a defect. A feature vector 51 can be extracted. Through this, the extracted third feature vector 51 may include "information of an important region in defect inspection for the feature maps having spatial significance".

For example, the third feature vector 51 may be calculated and extracted based on the following equation.

[Equation 2]

In Equation 2, Z1, Z2, and Z3 may mean the first, second, and third layers of the three second convolutional layers 50-1, respectively, and GAP may mean global average pooling 50-2. can In addition, the Z1 layer and the Z3 layer may include 64 filters having a size of 1x1, and the Z2 layer may include 64 filters having a size of 3x3. The computing device 100 inputs the plurality of feature maps 22 to the three second convolution layers 50-1 of Z1, Z2, and Z3, and applies a convolution operation to channel information. By deleting and generating a feature map with spatial importance, and performing global average pooling (50-2) on the generated feature map with spatial importance, height x width (Height x Since one feature map with spatial importance of which the size of width is 2x2 can be converged to one value, the feature maps with spatial importance can be converted into one third feature. It can be extracted as vector (51). For example, if the feature maps with spatial importance correspond to feature maps with a size of 2x2 that exist for each of 64 channels, performing global average pooling on a tensor shape of 2x2x64 (Height x Width x Channel) A third feature vector 51 in a one-dimensional form having 64 values may be extracted. On the other hand, the process of extracting the third feature vector 41 of FIG. 7B is a spatial attention ( Spatial attention) process may be included.

In addition, the third feature vector 51 extracted through the spatial attention process can extract "information of an important region in the defect inspection for the feature maps having spatial importance", and features related to defects. The output value of spatial attention may be applied in the defect prediction process through a process of concatenating the third feature vector 51 including information on the map area with other feature vectors (first or second feature vectors). A detailed description of this will be described below with reference to FIG. 8 .

8 is a schematic diagram illustrating a process of predicting defects based on a first feature vector, a second feature vector, and a third feature vector using a neural network model according to an embodiment of the present disclosure.

Referring to FIG. 8 , the computing device 100 may generate a first synthesized vector 52 by concatenating the second feature vector 41 and the third feature vector 51 . For example, when the second feature vector 41 is a one-dimensional vector having 64 values and the third feature vector 51 is a one-dimensional vector having 64 values, the two vectors are A first composite vector 52 may be generated by concatenation. In this case, the second feature vector 41 includes information about a feature most closely related to whether there is a defect in the plurality of feature maps 22, and the third feature vector 51 includes a plurality of feature maps 22 ) contains information about the feature map region related to whether or not there is a defect, so the two vectors 41 and 51 are complementary to each other, and the first synthesized vector generated by associating the two vectors 41 and 51 52 may include both information on a feature most closely related to whether or not there is a defect in the plurality of feature maps 22 and information about a feature map area related to whether or not there is a defect.

Additionally, the computing device 100 may concatenate the first synthesized vector 52 and the first feature vector 31 to generate a final synthesized vector 53, and the final synthesized vector 53 Based on this, it is possible to predict whether or not there is a defect. For example, in the plurality of feature maps 22, the first synthesized vector 52 including information on a feature most closely related to whether or not there is a defect and information about a feature map region related to whether or not there is a defect is provided as the first synthesized vector 52 . By associating with the first feature vector 31, which includes information on features necessary for inspecting defects of the plurality of feature maps 22, the final synthesized vector 53 is obtained from defects in the plurality of feature maps 22. The information on the features required for inspection may include information on a feature most closely related to whether or not there is a defect in the plurality of feature maps 22 and information on which information about a feature map area related to whether or not there is a defect is reflected. Therefore, the final synthesized vector 53 is input to one or more second fully connected layers 54, and the neural network model 20 can predict whether or not the product is defective by using the Sotfmax function (55 ). According to an embodiment of the present disclosure, by predicting whether an article is defective based on a 2D image, there is a technical effect of lower cost and faster prediction speed than a method using a 3D camera. In addition, “a process of extracting the second feature vector 41 for identifying a feature map related to whether or not there is a defect among a plurality of feature maps 22” and “a process of extracting a defect” corresponding to the channel attention process and the spatial attention process, respectively. The process of extracting the third feature vector 51 for identifying the feature map area related to whether or not the defect is detected is performed in parallel, and the two complementary vectors 41 and 51 are used together in the defect prediction process, thereby predicting defects. There is a technical effect that can increase the accuracy of The neural network model 20 used to predict defects of an article can be learned through a data set of 80%, which is training data, among data sets consisting of 80% for training and 20% for testing, and included in the data set The image input to the neural network model can be set to a dimension of 380 x 380 x 3 (Height x Width x Channel), and the loss function used to learn the neural network model 20 is CE (Cross Entropy) loss. It can be.

Experimental data related to the technical effects of the neural network model 20 are described with reference to the following table.

<Accuracy comparison table of prediction performance results for each neural network model 20 according to whether the first, second, and third feature vectors are used>

In the comparison table, MODEL-1 is a neural network model that performs defect prediction using only the first feature vector 31, and MODEL-2 is a neural network model that performs defect prediction using the first and second feature vectors. , MODEL-3 is a neural network model that performs defect prediction using the first feature vector and the third feature vector, and MODEL-4 predicts defects using the first feature vector, the second feature vector, and the third feature vector. It is a neural network model (corresponding to the neural network model 20 according to an embodiment of the present disclosure) that performs. Accuracy may mean the accuracy of defect prediction.

Referring to the above table, “the neural network model 20 performing defect prediction according to the embodiment of the present disclosure” is “the neural network model performing defect prediction using only the first feature vector 31”, “the first feature A neural network model that performs defect prediction using the vector 31 and the second feature vector 41”, and a “neural network that performs defect prediction using the first feature vector 31 and the third feature vector 51” It can be seen that the defect prediction is performed more accurately than the model”.

According to an embodiment of the present disclosure, a computer readable medium storing a data structure is disclosed. Data structure can refer to the organization, management, and storage of data that enables efficient access and modification of data. Data structure may refer to the organization of data to solve a specific problem (eg, data retrieval, data storage, data modification in the shortest time). A data structure may be defined as a physical or logical relationship between data elements designed to support a specific data processing function. A logical relationship between data elements may include a connection relationship between user-defined data elements. A physical relationship between data elements may include an actual relationship between data elements physically stored in a computer-readable storage medium (eg, a persistent storage device). The data structure may specifically include a set of data, a relationship between data, and a function or command applicable to the data. Through an effectively designed data structure, a computing device can perform calculations while using minimal resources of the computing device. Specifically, the computing device can increase the efficiency of operation, reading, insertion, deletion, comparison, exchange, and search through an effectively designed data structure.

The data structure can be divided into a linear data structure and a non-linear data structure according to the shape of the data structure. A linear data structure may be a structure in which only one data is connected after one data. Linear data structures may include lists, stacks, queues, and decks. A list may refer to a series of data sets in which order exists internally. The list may include a linked list. A linked list may be a data structure in which data are connected in such a way that each data is connected in a line with a pointer. In a linked list, a pointer can contain information about connection to the next or previous data. A linked list can be expressed as a singly linked list, a doubly linked list, or a circular linked list depending on the form. A stack can be a data enumeration structure that allows limited access to data. A stack can be a linear data structure in which data can be processed (eg, inserted or deleted) at only one end of the data structure. The data stored in the stack may be a LIFO-Last in First Out (Last in First Out) data structure. A queue is a data listing structure that allows limited access to data, and unlike a stack, it can be a data structure (FIFO-First in First Out) in which data stored later comes out later. A deck can be a data structure that can handle data from either end of the data structure.

The nonlinear data structure may be a structure in which a plurality of data are connected after one data. The non-linear data structure may include a graph data structure. A graph data structure can be defined as a vertex and an edge, and an edge can include a line connecting two different vertices. A graph data structure may include a tree data structure. The tree data structure may be a data structure in which one path connects two different vertices among a plurality of vertices included in the tree. That is, it may be a data structure that does not form a loop in a graph data structure.

Throughout this specification, computational model, neural network, network function, and neural network may be used interchangeably. Hereinafter, a neural network is unified and described. The data structure may include a neural network. And the data structure including the neural network may be stored in a computer readable medium. The data structure including the neural network may also include preprocessed data for processing by the neural network, data input to the neural network, weights of the neural network, hyperparameters of the neural network, data obtained from the neural network, activation function associated with each node or layer of the neural network, and neural network It may include a loss function for learning of . A data structure including a neural network may include any of the components described above. That is, the data structure including the neural network includes preprocessed data for processing by the neural network, data input to the neural network, weights of the neural network, hyperparameters of the neural network, data obtained from the neural network, activation function associated with each node or layer of the neural network, and neural network. It may be configured to include all or any combination thereof, such as a loss function for learning of . In addition to the foregoing configurations, the data structure comprising the neural network may include any other information that determines the characteristics of the neural network. In addition, the data structure may include all types of data used or generated in the computational process of the neural network, but is not limited to the above. A computer readable medium may include a computer readable recording medium and/or a computer readable transmission medium. A neural network may consist of a set of interconnected computational units, which may generally be referred to as nodes. These nodes may also be referred to as neurons. A neural network includes one or more nodes.

The data structure may include data input to the neural network. A data structure including data input to the neural network may be stored in a computer readable medium. Data input to the neural network may include training data input during a neural network learning process and/or input data input to a neural network that has been trained. Data input to the neural network may include pre-processed data and/or data subject to pre-processing. Pre-processing may include a data processing process for inputting data to a neural network. Accordingly, the data structure may include data subject to pre-processing and data generated by pre-processing. The foregoing data structure is only an example, and the present disclosure is not limited thereto.

The data structure may include the weights of the neural network. (In this specification, weights and parameters may be used in the same meaning.) Also, a data structure including weights of a neural network may be stored in a computer readable medium. A neural network may include a plurality of weights. The weight may be variable, and may be changed by a user or an algorithm in order to perform a function desired by the neural network. For example, when one or more input nodes are interconnected by respective links to one output node, the output node is set to a link corresponding to values input to input nodes connected to the output node and respective input nodes. A data value output from an output node may be determined based on the weight. The foregoing data structure is only an example, and the present disclosure is not limited thereto.

As a non-limiting example, the weights may include weights that are varied during neural network training and/or weights for which neural network training has been completed. The variable weight in the neural network learning process may include a weight at the time the learning cycle starts and/or a variable weight during the learning cycle. The weights for which neural network learning has been completed may include weights for which learning cycles have been completed. Accordingly, the data structure including the weights of the neural network may include a data structure including weights that are variable during the neural network learning process and/or weights for which neural network learning is completed. Therefore, it is assumed that the above-described weights and/or combinations of weights are included in the data structure including the weights of the neural network. The foregoing data structure is only an example, and the present disclosure is not limited thereto.

The data structure including the weights of the neural network may be stored in a computer readable storage medium (eg, a memory or a hard disk) after going through a serialization process. Serialization can be the process of converting a data structure into a form that can be stored on the same or another computing device and later reconstructed and used. A computing device may serialize data structures to transmit and receive data over a network. The data structure including the weights of the serialized neural network may be reconstructed on the same computing device or another computing device through deserialization. The data structure including the weights of the neural network is not limited to serialization. Furthermore, the data structure including the weights of the neural network is a data structure for increasing the efficiency of operation while minimizing the resource of the computing device (for example, B-Tree, Trie, m-way search tree, AVL tree, Red-Black Tree). The foregoing is only an example, and the present disclosure is not limited thereto.

The data structure may include hyper-parameters of the neural network. Also, the data structure including the hyperparameters of the neural network may be stored in a computer readable medium. A hyperparameter may be a variable variable by a user. Hyperparameters include, for example, learning rate, cost function, number of learning cycle iterations, weight initialization (eg, setting the range of weight values to be targeted for weight initialization), hidden unit number (eg, the number of hidden layers and the number of nodes in the hidden layer). The foregoing data structure is only an example, and the present disclosure is not limited thereto.

9 is a simplified and general schematic diagram of an exemplary computing environment in which embodiments of the present disclosure may be implemented.

Although the present disclosure has been described above as being generally embodied by a computing device, those skilled in the art will understand that the present disclosure may be combined with computer executable instructions and/or other program modules that may be executed on one or more computers and/or with hardware. It will be appreciated that it can be implemented as a combination of software.

Generally, program modules include routines, programs, components, data structures, etc. that perform particular tasks or implement particular abstract data types. In addition, it will be appreciated by those skilled in the art that the methods of the present disclosure may be used in single-processor or multiprocessor computer systems, minicomputers, mainframe computers as well as personal computers, handheld computing devices, microprocessor-based or programmable consumer electronics, and the like. It will be appreciated that other computer system configurations may be implemented, including (each of which may be operative in connection with one or more associated devices).

The described embodiments of the present disclosure may also be practiced in distributed computing environments where certain tasks are performed by remote processing devices that are linked through a communications network. In a distributed computing environment, program modules may be located in both local and remote memory storage devices.

Computers typically include a variety of computer readable media. Computer readable media can be any medium that can be accessed by a computer, including volatile and nonvolatile media, transitory and non-transitory media, removable and non-transitory media. Includes removable media. By way of example, and not limitation, computer readable media may include computer readable storage media and computer readable transmission media. Computer readable storage media are volatile and nonvolatile media, transitory and non-transitory, removable and non-removable media implemented in any method or technology for storage of information such as computer readable instructions, data structures, program modules or other data. includes media Computer readable storage media may include RAM, ROM, EEPROM, flash memory or other memory technology, CD-ROM, digital video disk (DVD) or other optical disk storage device, magnetic cassette, magnetic tape, magnetic disk storage device or other magnetic storage device. device, or any other medium that can be accessed by a computer and used to store desired information.

A computer readable transmission medium typically embodies computer readable instructions, data structures, program modules or other data in a modulated data signal such as a carrier wave or other transport mechanism. Including all information delivery media. The term modulated data signal means a signal that has one or more of its characteristics set or changed so as to encode information within the signal. By way of example, and not limitation, computer readable transmission media includes wired media such as a wired network or direct-wired connection, and wireless media such as acoustic, RF, infrared, and other wireless media. Combinations of any of the above are also intended to be included within the scope of computer readable transmission media.

An exemplary environment 1100 implementing various aspects of the present disclosure is shown including a computer 1102, which includes a processing unit 1104, a system memory 1106, and a system bus 1108. do. System bus 1108 couples system components, including but not limited to system memory 1106 , to processing unit 1104 . Processing unit 1104 may be any of a variety of commercially available processors. Dual processor and other multiprocessor architectures may also be used as the processing unit 1104.

System bus 1108 may be any of several types of bus structures that may additionally be interconnected to a memory bus, a peripheral bus, and a local bus using any of a variety of commercial bus architectures. System memory 1106 includes read only memory (ROM) 1110 and random access memory (RAM) 1112 . A basic input/output system (BIOS) is stored in non-volatile memory 1110, such as ROM, EPROM, or EEPROM, and is a basic set of information that helps transfer information between components within computer 1102, such as during startup. contains routines. RAM 1112 may also include high-speed RAM, such as static RAM, for caching data.

The computer 1102 may also include an internal hard disk drive (HDD) 1114 (eg, EIDE, SATA) - the internal hard disk drive 1114 may also be configured for external use within a suitable chassis (not shown). Yes—a magnetic floppy disk drive (FDD) 1116 (e.g., for reading from or writing to a removable diskette 1118), and an optical disk drive 1120 (e.g., a CD-ROM) for reading disc 1122 or reading from or writing to other high capacity optical media such as DVDs). The hard disk drive 1114, magnetic disk drive 1116, and optical disk drive 1120 are connected to the system bus 1108 by a hard disk drive interface 1124, magnetic disk drive interface 1126, and optical drive interface 1128, respectively. ) can be connected to The interface 1124 for external drive implementation includes at least one or both of USB (Universal Serial Bus) and IEEE 1394 interface technologies.

These drives and their associated computer readable media provide non-volatile storage of data, data structures, computer executable instructions, and the like. In the case of computer 1102, drives and media correspond to storing any data in a suitable digital format. Although the description of computer readable media above refers to HDDs, removable magnetic disks, and removable optical media such as CDs or DVDs, those skilled in the art can use zip drives, magnetic cassettes, flash memory cards, and cartridges. It will be appreciated that other tangible computer readable media such as , , and the like may also be used in the exemplary operating environment and that any such media may include computer executable instructions for performing the methods of the present disclosure. .

A number of program modules may be stored on the drive and RAM 1112, including an operating system 1130, one or more application programs 1132, other program modules 1134, and program data 1136. All or portions of the operating system, applications, modules and/or data may also be cached in RAM 1112. It will be appreciated that the present disclosure may be implemented in a variety of commercially available operating systems or combinations of operating systems.

A user may enter commands and information into the computer 1102 through one or more wired/wireless input devices, such as a keyboard 1138 and a pointing device such as a mouse 1140. Other input devices (not shown) may include a microphone, IR remote control, joystick, game pad, stylus pen, touch screen, and the like. Although these and other input devices are often connected to the processing unit 1104 through an input device interface 1142 that is connected to the system bus 1108, a parallel port, IEEE 1394 serial port, game port, USB port, IR interface, may be connected by other interfaces such as the like.

A monitor 1144 or other type of display device is also connected to the system bus 1108 through an interface such as a video adapter 1146. In addition to the monitor 1144, computers typically include other peripheral output devices (not shown) such as speakers, printers, and the like.

Computer 1102 may operate in a networked environment using logical connections to one or more remote computers, such as remote computer(s) 1148 via wired and/or wireless communications. Remote computer(s) 1148 may be a workstation, computing device computer, router, personal computer, handheld computer, microprocessor-based entertainment device, peer device, or other common network node, and generally includes It includes many or all of the components described for, but for simplicity, only memory storage device 1150 is shown. The logical connections shown include wired/wireless connections to a local area network (LAN) 1152 and/or a larger network, such as a wide area network (WAN) 1154 . Such LAN and WAN networking environments are common in offices and corporations and facilitate enterprise-wide computer networks, such as intranets, all of which can be connected to worldwide computer networks, such as the Internet.

When used in a LAN networking environment, computer 1102 connects to local network 1152 through wired and/or wireless communication network interfaces or adapters 1156. Adapter 1156 may facilitate wired or wireless communications to LAN 1152, which also includes a wireless access point installed therein to communicate with wireless adapter 1156. When used in a WAN networking environment, computer 1102 may include a modem 1158, be connected to a communicating computing device on WAN 1154, or establish communications over WAN 1154, such as over the Internet. have other means. A modem 1158, which may be internal or external and a wired or wireless device, is connected to the system bus 1108 through a serial port interface 1142. In a networked environment, program modules described for computer 1102, or portions thereof, may be stored on remote memory/storage device 1150. It will be appreciated that the network connections shown are exemplary and other means of establishing a communication link between computers may be used.

Computer 1102 is any wireless device or entity that is deployed and operating in wireless communication, eg, printers, scanners, desktop and/or portable computers, portable data assistants (PDAs), communication satellites, wireless detectable tags associated with It operates to communicate with arbitrary equipment or places and telephones. This includes at least Wi-Fi and Bluetooth wireless technologies. Thus, the communication may be a predefined structure as in conventional networks or simply an ad hoc communication between at least two devices.

Wi-Fi (Wireless Fidelity) makes it possible to connect to the Internet without wires. Wi-Fi is a wireless technology, such as a cell phone, that allows such devices, eg, computers, to transmit and receive data both indoors and outdoors, i.e. anywhere within coverage of a base station. Wi-Fi networks use a radio technology called IEEE 802.11 (a, b, g, etc.) to provide secure, reliable, and high-speed wireless connections. Wi-Fi can be used to connect computers to each other, to the Internet, and to wired networks (using IEEE 802.3 or Ethernet). Wi-Fi networks can operate in the unlicensed 2.4 and 5 GHz radio bands, for example, at 11 Mbps (802.11a) or 54 Mbps (802.11b) data rates, or in products that include both bands (dual band) .

Those skilled in the art will understand that information and signals may be represented using any of a variety of different technologies and techniques. For example, data, instructions, instructions, information, signals, bits, symbols and chips that may be referenced in the above description are voltages, currents, electromagnetic waves, magnetic fields or particles, optical fields s or particles, or any combination thereof.

Those skilled in the art will understand that the various illustrative logical blocks, modules, processors, means, circuits, and algorithm steps described in connection with the embodiments disclosed herein are electronic hardware, (for convenience) , may be implemented by various forms of program or design code (referred to herein as software) or a combination of both. To clearly illustrate this interchangeability of hardware and software, various illustrative components, blocks, modules, circuits, and steps have been described above generally in terms of their functionality. Whether such functionality is implemented as hardware or software depends on the particular application and the design constraints imposed on the overall system. Skilled artisans may implement the described functionality in varying ways for each particular application, but such implementation decisions should not be interpreted as causing a departure from the scope of the present disclosure.

Various embodiments presented herein may be implemented as a method, apparatus, or article of manufacture using standard programming and/or engineering techniques. The term article of manufacture includes a computer program, carrier, or media accessible from any computer-readable storage device. For example, computer-readable storage media include magnetic storage devices (eg, hard disks, floppy disks, magnetic strips, etc.), optical disks (eg, CDs, DVDs, etc.), smart cards, and flash memory devices (eg, EEPROM, cards, sticks, key drives, etc.), but are not limited thereto. Additionally, various storage media presented herein include one or more devices and/or other machine-readable media for storing information.

It is to be understood that the specific order or hierarchy of steps in the processes presented is an example of example approaches. Based upon design priorities, it is to be understood that the specific order or hierarchy of steps in the processes may be rearranged within the scope of this disclosure. The accompanying method claims present elements of the various steps in a sample order, but are not meant to be limited to the specific order or hierarchy presented.

The description of the presented embodiments is provided to enable any person skilled in the art to make or use the present disclosure. Various modifications to these embodiments will be apparent to those skilled in the art of this disclosure, and the general principles defined herein may be applied to other embodiments without departing from the scope of this disclosure. Thus, the present disclosure is not to be limited to the embodiments presented herein, but is to be interpreted in the widest scope consistent with the principles and novel features presented herein.

Claims (8)

A method for predicting whether an item is defective, performed by a computing device, comprising:
acquiring one or more images;
inputting the obtained image to a neural network model and generating a plurality of feature maps;
extracting a first feature vector based on the plurality of feature maps;
extracting a second feature vector for identifying a feature map related to whether or not there is a defect among the plurality of feature maps, based on the extracted first feature vector;
extracting a third feature vector for identifying a feature map region related to a defect, based on the plurality of feature maps; and
predicting whether there is a defect based on the first feature vector, the second feature vector, and the third feature vector by using the neural network model;
including,
The step of extracting a third feature vector for identifying a feature map region related to whether or not there is a defect based on the plurality of feature maps,
inputting the plurality of feature maps to one or more second convolution layers and applying a convolution operation; and
performing additional pooling on the feature maps to which the convolution operation is applied, and extracting a third feature vector for identifying a feature map region related to whether or not there is a defect;
including,
method.
According to claim 1,
The one or more images,
contains one or more 2D images;
Acquiring the one or more images,
Acquiring at least one image of a front image, a back image, a right image, a left image, a top image, or a bottom image of an article,
method.
According to claim 1,
The step of inputting the obtained image to the neural network model and generating a plurality of feature maps,
performing gamma correction on the obtained image; and
inputting the gamma-corrected image to the neural network model;
including,
method.
According to claim 1,
The step of inputting the obtained image to the neural network model and generating a plurality of feature maps,
Features of the image are extracted using a plurality of first convolution layers and a plurality of first pooling layers included in the neural network model, and a plurality of feature maps are obtained based on the extracted features. generating them;
including,
method.
According to claim 1,
Extracting a first feature vector based on the plurality of feature maps,
Extracting the first feature vector by performing additional pooling on the plurality of feature maps,
method.
According to any one of claims 1 or 5,
The additional pooling,
Including global average pooling,
method.
A computer program stored on a computer-readable storage medium, which, when executed by one or more processors, causes the one or more processors to perform operations for predicting whether an article is defective, the operations comprising:
acquiring one or more images;
inputting the obtained image to a neural network model and generating a plurality of feature maps;
extracting a first feature vector based on the plurality of feature maps;
extracting a second feature vector for identifying a feature map related to whether or not there is a defect among the plurality of feature maps, based on the extracted first feature vector;
extracting a third feature vector for identifying a feature map region related to a defect, based on the plurality of feature maps; and
predicting a defect based on the first feature vector, the second feature vector, and the third feature vector by using the neural network model;
including,
Based on the plurality of feature maps, the operation of extracting a third feature vector for identifying a feature map region related to whether or not there is a defect,
inputting the plurality of feature maps to one or more second convolution layers and applying a convolution operation; and
performing additional pooling on the feature maps to which the convolution operation is applied, and extracting a third feature vector for identifying a feature map region related to whether or not there is a defect;
including,
A computer program stored on a computer readable storage medium.
As a computing device,
at least one processor; and
Memory
including,
The at least one processor,
acquire one or more images;
inputting the obtained image to a neural network model and generating a plurality of feature maps;
extracting a first feature vector based on the plurality of feature maps;
extracting a second feature vector for identifying a feature map related to whether or not there is a defect among the plurality of feature maps, based on the extracted first feature vector;
based on the plurality of feature maps, extracting a third feature vector for identifying a feature map region related to whether or not there is a defect; and
Using the neural network model, it is configured to predict whether or not a defect is present based on the first feature vector, the second feature vector, and the third feature vector,
Based on the plurality of feature maps, extracting a third feature vector for identifying a feature map region related to whether or not a defect exists,
inputting the plurality of feature maps to one or more second convolution layers and applying a convolution operation; and
performing additional pooling on the feature maps to which the convolution operation is applied, and extracting a third feature vector for identifying a feature map region related to whether or not there is a defect;
including,
computing device.

Images