Method and system for providing rebar arrangement state based on computer vision

Abstract

To provide a method and system for rebar arrangement state, configured to detect spaces formed between rebars in a captured image of a rebar arrangement, based on computer vision.SOLUTION: A method includes: inputting a rebar image to a pre-trained space detection model so as to estimate mask regions corresponding to spaces between the plurality of rebars; obtaining a plurality of mask regions; classifying the plurality of mask regions into a similar group and an unsimilar group on the basis of similarity in size between the plurality of mask regions; and displaying the plurality of mask regions corresponding to at least one of the similar group or the unsimilar group in the rebar image.SELECTED DRAWING: Figure 7

Description

Technical Field

[0001] The present invention relates to a method and a system for providing a reinforcement arrangement state based on computer vision.

[0002] This research was carried out with the research funds (RS-2023-00251002) of the Korea Agency for Infrastructure Technology Advancement (KAIA) supported by the Ministry of Land, Infrastructure and Transport.

Background Art

[0003] Structural durability is a very important factor for ensuring safety during the design and construction of buildings, bridges, and other civil engineering structures. In particular, structural durability varies greatly depending on the arrangement and distribution of reinforcing bars.

[0004] Specifically, the interval at which the reinforcing bars are arranged effectively disperses the load of the structure, alleviates the phenomenon of stress concentration at a specific point, and enhances the structural durability. However, if the reinforcing bars are not aligned, it will have a significant impact on the supporting force of the structure and pose a danger.

[0005] Therefore, conventionally, the arrangement state of the reinforcing bars has been inspected in various ways. Typically, a method can be used in which an inspector directly inspects the arrangement state of the reinforcing bars using a visual aid device.

[0006] Recently, research has been actively conducted on methods for measuring the interval between reinforcing bars based on fixtures, sensor-based devices, images, and computer vision. In particular, methods have been developed for detecting the arrangement pattern and characteristics of reinforcing bars by utilizing deep learning models such as convolutional neural networks (CNNs).

Summary of the Invention

Problems to be Solved by the Invention

[0007] The present invention relates to a method and system for providing a reinforcing bar arrangement state that detects a space formed between reinforcing bars in an image of a reinforcing bar arrangement.

[0008] The present invention also relates to a method and system for providing a reinforcing bar arrangement state that classifies the spaces detected in an image according to size and provides the arrangement state of the reinforcing bars.

Means for Solving the Problems

[0009] In order to solve the above problems, a method for providing a reinforcing bar arrangement state according to the present invention includes a step of receiving a reinforcing bar image obtained by photographing a space in which a plurality of reinforcing bars are arranged, and inputting the reinforcing bar image into a space detection model pre-trained to estimate a mask region corresponding to the space between the plurality of reinforcing bars, and obtaining a plurality of mask regions shown in the reinforcing bar image, a step of classifying the plurality of mask regions into a similar group and a dissimilar group based on the similarity of the sizes between the plurality of mask regions, and a step of providing the arrangement state of the plurality of reinforcing bars by displaying the plurality of mask regions corresponding to at least one of the similar group and the dissimilar group on the reinforcing bar image.

[0010] A system for providing a reinforcing bar arrangement state according to the present invention includes a storage unit for storing a reinforcing bar image obtained by photographing a space in which a plurality of reinforcing bars are arranged, and a control unit for inputting the reinforcing bar image into a space detection model pre-trained to estimate a mask region corresponding to the space between the plurality of reinforcing bars and obtaining a plurality of mask regions shown in the reinforcing bar image. The control unit can provide the arrangement state of the plurality of reinforcing bars by classifying the plurality of mask regions into a similar group and a dissimilar group based on the similarity of the sizes between the plurality of mask regions and displaying the plurality of mask regions corresponding to at least one of the similar group and the dissimilar group on the reinforcing bar image.

[0011] In addition, a program stored in a computer-readable recording medium according to the present invention is executed by one or more processes in an electronic device, and is a program stored in a computer-readable recording medium, the program including steps of receiving a rebar image obtained by photographing a space in which a plurality of rebars are arranged, inputting the rebar image into a space detection model pre-trained to estimate a mask region corresponding to a space between the plurality of rebars, and obtaining a plurality of mask regions shown in the rebar image, classifying the plurality of mask regions into a similar group and a dissimilar group based on a similarity with respect to sizes between the plurality of mask regions, and providing an arrangement state of the plurality of rebars by displaying the plurality of mask regions corresponding to at least one of the similar group and the dissimilar group on the rebar image, and may include instruction words for performing the steps.

Advantages of the Invention

[0012] According to various embodiments of the present invention, a rebar arrangement state providing method and system can efficiently verify a structural arrangement of rebar intervals by detecting a space formed between rebars in an image obtained by photographing a rebar arrangement and classifying and providing the detected space according to size.

[0013] For this purpose, according to various embodiments of the present invention, a rebar arrangement state providing method and system can more accurately and precisely detect a space formed between rebars in a rebar image by enhancing learning rebar images in various ways.

[0014] In particular, according to various embodiments of the present invention, a rebar arrangement state providing method and system can more accurately detect a region where an error has occurred in a structural arrangement of rebar intervals by repeatedly performing a process of classifying a space formed between rebars according to size several times.

Brief Description of the Drawings

[0015]

Figure 1

Figure 2

Figure 3

Figure 4

Figure 5

Figure 6

Figure 7

Figure 8

Figure 9

Figure 10

Figure 11

Figure 12

Embodiments for Carrying Out the Invention

[0016] Hereinafter, the embodiments disclosed in this specification will be described in detail with reference to the accompanying drawings. However, the same or similar components are denoted by the same reference numerals regardless of the reference signs, and redundant descriptions thereof are omitted. The suffixes "module" and "unit" for the components used in the following description are given or mixed simply for the purpose of easily creating the specification, and do not have meanings or roles that are distinguishable from each other by themselves. Further, when it is determined that a specific description of related known technologies may obscure the gist of the embodiments disclosed in this specification when explaining the embodiments disclosed in this specification, the detailed description thereof is omitted. Further, the accompanying drawings are merely for facilitating the understanding of the embodiments disclosed in this specification, and the technical idea disclosed in this specification is not limited by the accompanying drawings, and should be understood to include all modifications, equivalents, and alternatives included in the idea and technical scope of the present invention.

[0017] Terms including ordinal numbers such as first and second can be used to describe various components, but the above components are not limited by the above terms. The above terms are used only for the purpose of distinguishing one component from another.

[0018] When it is mentioned that a certain component is "connected" or "coupled" to another component, it should be understood that it may be directly connected or coupled to the other component, but other components may also exist therebetween. On the other hand, when it is mentioned that a certain component is "directly connected" or "directly coupled" to another component, it should be understood that no other components exist therebetween.

[0019] Unless clearly indicated otherwise in the context, singular expressions include plural expressions.

[0020] In this application, terms such as "comprising" or "having" are intended to specify that the features, numbers, steps, operations, components, parts, or combinations thereof described in this specification exist, and it should be understood that they do not preclude the possibility of the existence or addition of one or more other features, numbers, steps, operations, components, parts, or combinations thereof.

[0021] FIG. 1 shows an embodiment of a method for providing a reinforcing bar arrangement state according to the present invention. FIGS. 2 and 3 show an embodiment of detecting a mask region in a reinforcing bar image. FIGS. 4 and 5 show an embodiment of classifying the mask region. FIG. 6 shows a reinforcing bar arrangement state providing system according to the present invention.

[0022] Referring to FIG. 1, a reinforcing bar arrangement state providing system 100 according to the present invention can input a reinforcing bar image into a pre-trained spatial detection model and estimate a mask region for the reinforcing bar image.

[0023] Here, the reinforcing bar image is an image obtained by photographing a space in which a plurality of reinforcing bars are arranged, and may be an image taken by an RGB camera, a black-and-white camera, or the like.

[0024] At this time, the plurality of reinforcing bars may be arranged so as to intersect along different axes from each other. That is, the reinforcing bar image may be an image obtained by photographing a space in which a plurality of reinforcing bars are arranged so as to intersect each other.

[0025] Also, the mask region may be a space formed by a plurality of reinforcing bars arranged so as to intersect each other. For example, in a reinforcing bar image in which a plurality of reinforcing bars arranged along a first axis and a second axis intersecting vertically are photographed, the mask region may be a rectangular space formed by two reinforcing bars arranged adjacent to each other along the first axis and two reinforcing bars arranged adjacent to each other along the second axis.

[0026] At this time, the mask region may be realized in various shapes such as a rhombus or a circle according to the angles at which a plurality of steel bars intersect and the respective shapes of the plurality of steel bars.

[0027] The space detection model may be an artificial neural network trained to estimate the mask region from the steel bar image. For example, the space detection model may be an artificial neural network realized based on a DNN (Deep Neural Network) and a CNN (Convolutional Neural Network).

[0028] For this purpose, the space detection model may be trained by a space detection model learning method, and such a space detection model learning method may be executed by the steel bar arrangement state providing system 100 according to the embodiment, or may be executed by another learning system (or device).

[0029] At this time, the learning system can store the learning steel bar image and enhance the learning steel bar image to generate a plurality of learning steel bar images.

[0030] Specifically, the learning system can arrange imaging devices (for example, cameras) at different positions with respect to the space in which a plurality of steel bars are arranged, and store the captured steel bar images as learning steel bar images.

[0031] For example, the learning system can store a plurality of learning steel bar images 11 captured at positions separated from each other by different distances with respect to the space in which a plurality of steel bars are arranged.

[0032] As another example, the learning system can store a plurality of learning steel bar images 12 captured at positions separated by a predetermined distance at different angles with respect to the space in which a plurality of steel bars are arranged.

[0033] As still another example, the learning system can also store a plurality of learning steel bar images captured under different illumination states with respect to the space in which a plurality of steel bars are arranged.

[0034] As another example, the learning system can capture a space in which a plurality of reinforcing bars having different dimensions are arranged, or store a plurality of learning reinforcing bar images that capture a space in which a plurality of reinforcing bars are arranged at different intervals from each other.

[0035] Furthermore, the learning system can edit the learning reinforcing bar images to generate a plurality of learning reinforcing bar images 13 that are enhanced (Augmented) into different shapes from each other.

[0036] For example, the learning system can edit the learning reinforcing bar images so that they are flipped horizontally and generate the flipped learning reinforcing bar images. Thereby, the learning system can train the spatial detection model to estimate mask regions existing in various directions and positions.

[0037] As another example, the learning system can edit the learning reinforcing bar images so that they are rotated at a predetermined angle (for example, 45 degrees or 135 degrees, etc.) and generate the rotated learning reinforcing bar images. Thereby, the learning system can train the spatial detection model to accurately estimate the mask region in a wider direction in the reinforcing bar image.

[0038] As still another example, the learning system can edit the learning reinforcing bar images to adjust the contrast and generate the learning reinforcing bar images with the adjusted contrast. For this purpose, the learning system calculates the maximum pixel intensity and the minimum pixel intensity for a plurality of pixels belonging to the learning reinforcing bar images, and normalizes the values of the plurality of pixels based on the calculation results, thereby improving the visibility and sharpness of the objects corresponding to the plurality of reinforcing bars in the learning reinforcing bar images. Thereby, the learning system can reduce the influence of noise and artifacts included in the reinforcing bar images and train the spatial detection model to subdivide the mask region.

[0039] As another example, the learning system can edit to adjust the saturation of the training rebar image and generate a training rebar image with adjusted saturation. For this purpose, the learning system can adjust the saturation of the training rebar image by adjusting the pixel intensity for a plurality of pixels belonging to the training rebar image. Thereby, the learning system can train the spatial detection model to more accurately distinguish the region corresponding to the rebar and the mask region in the rebar image.

[0040] Through the above configuration, the learning system can generate a plurality of training rebar images 13 that have been edited by at least one of inversion, rotation, contrast adjustment, and saturation adjustment for a specific training rebar image.

[0041] Furthermore, the learning system can label a training mask region corresponding to the space between a plurality of rebars for each of the plurality of training rebar images (14).

[0042] Specifically, for each of the plurality of training rebar images, the learning system can input a bounding box corresponding to the mask region as the training mask region and label each training rebar image with the training mask region input above.

[0043] Thereby, the learning system can train the spatial detection model to estimate the mask region in the rebar image using the plurality of training rebar images and the training mask regions labeled for each of the plurality of training rebar images (15).

[0044] Referring to FIG. 2, in one embodiment, the spatial detection model can be trained to detect the mask region in the rebar image based on DVNet (Deep Vision Net), SPPNet (Segmentation Pyramid Pooling Network), and DCNet (Deep CNN Network).

[0045] In connection with this, referring to FIG. 3, it is possible to confirm the form in which the space detection model according to one embodiment is realized.

[0046] Referring to FIG. 1 again, the reinforcing bar arrangement state providing system 100 according to the present invention classifies a plurality of mask regions estimated for a reinforcing bar image into a similar group and a dissimilar group based on the similarity with respect to the sizes between the respective mask regions (16), and can provide a reinforcing bar arrangement state according to the classification result (17).

[0047] Here, the similar group may be a group in which the arrangements of a plurality of reinforcing bars are uniform and the sizes of the spaces between the plurality of reinforcing bars are similar. That is, the similar group may include a plurality of mask regions having similar sizes among the plurality of mask regions detected in the reinforcing bar image.

[0048] For example, the similar group may include a plurality of mask regions belonging to a range predetermined from the average size value for the plurality of mask regions.

[0049] As another example, the similar group may include a plurality of mask regions belonging to a range (for example, the number of mask regions or the ratio to the entire mask region) predetermined from the maximum size value (or the minimum size value) based on the average size value for the plurality of mask regions.

[0050] The dissimilar group may be a group in which the arrangements of a plurality of reinforcing bars are non-uniform and the sizes of the spaces between the plurality of reinforcing bars are different, or may be a group including mask regions excluding the plurality of mask regions belonging to the similar group. That is, the dissimilar group may include a plurality of mask regions excluding the plurality of mask regions belonging to the similar group among the plurality of mask regions detected in the reinforcing bar image.

[0051] Referring to FIG. 4, in one embodiment, the rebar arrangement state providing system 100 can classify a plurality of mask regions into a first group (e.g., A, B, C) and a second group (e.g., P, Q, R, S, T, U, X, Y, Z) based on the number of pixels of each of the plurality of mask regions detected in the rebar image.

[0052] Next, the rebar arrangement state providing system 100 can identify the group (e.g., the second group) to which more mask regions belong among the first group and the second group, and further classify the identified group into a first sub-group (e.g., P, Q, R, S, T, U) and a second sub-group (e.g., X, Y, Z).

[0053] Thereby, the rebar arrangement state providing system 100 can identify the group (e.g., the first sub-group) to which more mask regions belong among the first sub-group and the second sub-group as a similar group, and identify other mask regions excluding the plurality of mask regions included in the similar group as a non-similar group.

[0054] In this regard, referring to FIG. 5, a method by which the rebar arrangement state providing system 100 classifies a plurality of mask regions into a similar group and a non-similar group can be confirmed according to one embodiment.

[0055] On the other hand, referring to FIG. 6, the rebar arrangement state providing system 100 according to the present invention may include an input unit 110, a storage unit 120, an output unit 130, and a control unit 140.

[0056] The input unit 110 can input a rebar image. For this purpose, the input unit 110 is connected via a wireless or wired network to another device, system, server, etc. equipped with the rebar image, and can receive the rebar image from the device, server, etc. Alternatively, the input unit 110 may be connected via a wireless or wired network to a photographing device such as a camera, and a rebar image photographed by the photographing device may be input.

[0057] The storage unit 120 can store the data and command words necessary for the operation of the steel bar arrangement state providing system 100 according to the present invention. For example, the storage unit 120 can store a steel bar image and information regarding the steel bar arrangement state generated for the steel bar image (e.g., similar groups and dissimilar groups). Further, the storage unit 120 can store a pre-learned spatial detection model.

[0058] The output unit 130 can output at least one of a steel bar image and information regarding the steel bar arrangement state generated for the steel bar image. For this purpose, the output unit 130 can be connected to an output device such as a display device via a wireless or wired network. Therefore, the output unit 130 can output a steel bar image and information regarding the steel bar arrangement state so that the user can visually confirm them.

[0059] On the other hand, the output unit 130 can also be connected to another device, system, and server via a wireless or wired network. In such a case, the output unit 130 can transmit at least one of a steel bar image and information regarding the steel bar arrangement state to the device, system, and server.

[0060] The control unit 140 can control the overall operation of the steel bar arrangement state providing system 100 according to the present invention. For example, the control unit 140 can input a steel bar image into the spatial detection model to obtain a plurality of mask regions, classify the plurality of mask regions into similar groups and dissimilar groups, and output a steel bar arrangement state based on at least one of the similar groups and dissimilar groups.

[0061] Based on the configuration of the steel bar arrangement state providing system 100 described above, the steel bar arrangement state providing method will be described more specifically below.

[0062] FIG. 7 is a flowchart showing a method for providing a reinforcing bar arrangement state according to the present invention. FIG. 8 shows an embodiment of detecting a mask region in a reinforcing bar image. FIGS. 9 to 11 show an embodiment of classifying a mask region. FIG. 12 shows an embodiment of providing a reinforcing bar arrangement state.

[0063] Referring to FIG. 7, a reinforcing bar arrangement state providing system 100 according to the present invention receives a reinforcing bar image obtained by photographing a space in which a plurality of reinforcing bars are arranged (S100), inputs the reinforcing bar image to a space detection model pre-trained to estimate a mask region corresponding to the space between the plurality of reinforcing bars, and can obtain a plurality of mask regions shown in the reinforcing bar image (S200).

[0064] Specifically, the reinforcing bar arrangement state providing system 100 receives a reinforcing bar image obtained by photographing a space in which a plurality of reinforcing bars are arranged in a lattice pattern, inputs the received reinforcing bar image to a pre-trained space detection model, and can obtain each of a plurality of regions corresponding to the spaces between the plurality of reinforcing bars formed by the plurality of reinforcing bars arranged in a lattice pattern as a mask region.

[0065] Referring to FIG. 8, for example, the reinforcing bar arrangement state providing system 100 inputs a reinforcing bar image to a space detection model, and in the reinforcing bar image 20 of a lattice-shaped reinforcing bar arrangement 23 formed by a plurality of reinforcing bars arranged along the first axis 21 and a plurality of reinforcing bars arranged along the second axis 22, a rectangular space formed by two adjacent reinforcing bars among the plurality of reinforcing bars arranged along the first axis 21 and two adjacent reinforcing bars among the plurality of reinforcing bars arranged along the second axis 22 can be obtained as a mask region 24.

[0066] Referring back to FIG. 7, the reinforcing bar arrangement state providing system 100 according to the present invention can classify a plurality of mask regions into a similar group and a dissimilar group based on the similarity of the sizes of the plurality of mask regions (S300).

[0067] Specifically, the reinforcing bar arrangement state providing system 100 can count the number of a plurality of pixels belonging to each of a plurality of mask regions, and classify the plurality of mask regions into a first group and a second group based on the counted number of the plurality of pixels.

[0068] Referring to FIG. 9, for example, the reinforcing bar arrangement state providing system 100 can calculate the average pixel number for a plurality of mask regions 31 based on the number of a plurality of pixels belonging to each of the plurality of mask regions 31 obtained for the reinforcing bar image 30.

[0069] Thereby, the reinforcing bar arrangement state providing system 100 can identify, as the first group 41 (or the second group 42), a plurality of mask regions (such as the first mask region 31a and the third mask region 31c, etc.) in which the number of pixels is larger than the average pixel number among the plurality of mask regions 31, and identify, as the second group 42 (or the first group 41), a plurality of mask regions (such as the second mask region 31b, etc.) in which the number of pixels is smaller than the average pixel number.

[0070] As another example, the reinforcing bar arrangement state providing system 100 can classify (or cluster) a plurality of mask regions obtained for the reinforcing bar image into a first group and a second group by using the K - means clustering technique.

[0071] Furthermore, the reinforcing bar arrangement state providing system 100 can count the number of a plurality of mask regions classified into the first group and the second group respectively, identify either one of the first group and the second group in which the counted number of the plurality of mask regions is the largest, and classify the plurality of mask regions into a first lower group and a second lower group based on the number of pixels for each of the plurality of mask regions belonging to the identified group.

[0072] Referring to FIG. 10, for example, when the number of mask regions included in the first group 41 is larger than the number of mask regions included in the second group among the first group 41 and the second group, a plurality of mask regions (e.g., 41a, 41b, 41c, 41d) belonging to the first group 41 can be classified into a first lower group 45 and a second lower group 46.

[0073] Alternatively, when the number of mask regions included in the first group 41 is smaller than the number of mask regions included in the second group among the first group 41 and the second group, the reinforcing bar arrangement state providing system 100 can also classify a plurality of mask regions belonging to the second group into the first lower group 45 and the second lower group 46.

[0074] At this time, based on the number of pixels for each of the plurality of mask regions included in the first group 41 (or the second group), the reinforcing bar arrangement state providing system 100 calculates the average number of pixels for the plurality of mask regions included in the first group 41 (or the second group), and among the plurality of mask regions included in the first group 41 (or the second group), a plurality of mask regions (e.g., the first mask region 41a, the third mask region 41b, and the seventh mask region 41d, etc.) having a larger number of pixels than the average number of pixels are specified as the first lower group 45 (or the second lower group 46), and a plurality of mask regions (e.g., the sixth mask region 41c, etc.) having a smaller number of pixels than the average number of pixels can be specified as the second lower group 46 (or the first lower group 45).

[0075] Alternatively, the reinforcing bar arrangement state providing system 100 can classify (or cluster) a plurality of mask regions included in the first group 41 (or the second group) into the first lower group 45 and the second lower group 46 by using the K-Means clustering technique.

[0076] As another example, the reinforcing bar arrangement state providing system 100 identifies the group to which either one of the plurality of mask regions classified into the first group and the second group and having the largest (or smallest) number of pixels counted above belongs, and can classify the plurality of mask regions belonging to the identified group into a first sub-group and a second sub-group.

[0077] Furthermore, the reinforcing bar arrangement state providing system 100 counts the number of the plurality of mask regions respectively classified into the first sub-group and the second sub-group, and identifies either one of the first sub-group and the second sub-group having the largest number of the plurality of mask regions counted above as a similar group, and can identify, as a non-similar group, the other mask regions excluding the plurality of mask regions belonging to the similar group among the plurality of mask regions obtained from the reinforcing bar image.

[0078] Referring to FIG. 11, for example, when the number of mask regions included in the first sub-group 45 (or the second sub-group 46) among the first sub-group 45 and the second sub-group 46 is larger than the number of mask regions included in the second sub-group 46 (or the first sub-group 45), the first sub-group 45 (or the second sub-group 46) can be identified as the similar group 47.

[0079] Also, the reinforcing bar arrangement state providing system 100 can identify either one of the first sub-group 45 and the second sub-group 46 (for example, the first sub-group 45) identified as the similar group 47 and the other sub-group (for example, the second sub-group 46) as non-similar groups among the first sub-group 45 and the second sub-group 46.

[0080] That is, the reinforcing bar arrangement state providing system 100 can identify a plurality of mask regions (for example, the first mask region 45a, the third mask region 45b, and the seventh mask region 45c) included in either one of the first lower group 45 and the second lower group 46 that has a larger number of mask regions as the similar group 47, and identify a plurality of mask regions (for example, the sixth mask region 46a and the eighth mask region 46b) included in the other lower group (for example, the second lower group 46) as the dissimilar group 48.

[0081] Referring to FIG. 7 again, the reinforcing bar arrangement state providing system 100 according to the present invention can provide the arrangement state of a plurality of reinforcing bars by displaying a plurality of mask regions corresponding to at least one of the similar group and the dissimilar group on the reinforcing bar image (S400).

[0082] Specifically, the reinforcing bar arrangement state providing system 100 can generate a highlighted display for a plurality of mask regions belonging to the dissimilar group among the plurality of mask regions obtained from the reinforcing bar image so as to be distinguished from other mask regions.

[0083] Referring to FIG. 12, for example, the reinforcing bar arrangement state providing system 100 can display a plurality of mask regions 51 specified as the dissimilar group on the reinforcing bar image 50 by designating the first color (for example, red), and display a plurality of mask regions 52 specified as the similar group by designating the second color (for example, green).

[0084] As another example, the reinforcing bar arrangement state providing system 100 can also selectively display only one of a plurality of mask regions specified as the dissimilar group and a plurality of mask regions specified as the similar group on the reinforcing bar image.

[0085] Through the above configuration, the rebar arrangement state providing system 100 according to the present invention can efficiently verify the structural arrangement of the rebar spacing by detecting the space formed between the rebars in the image of the rebar arrangement and classifying and providing the detected space according to the size.

[0086] For this purpose, the rebar arrangement state providing system 100 according to the present invention can detect the space formed between the rebars in the rebar image more accurately and precisely by enhancing the learning rebar images in various ways.

[0087] In particular, the rebar arrangement state providing system 100 according to the present invention can more accurately detect the area where an error has occurred in the structural arrangement of the rebar spacing by repeatedly performing the process of classifying the space formed between the rebars according to the size several times.

[0088] Furthermore, the present invention described above can be executed by one or more processes in an electronic device and can be realized as a program stored in a computer-readable recording medium.

[0089] Therefore, the present invention can be realized as computer-readable code or instruction words on a medium on which a program is recorded. That is, various control methods according to the present invention can be provided in the form of an integrated or individual program.

[0090] On the other hand, a computer-readable medium includes any type of recording device in which data readable by a computer system is stored. Examples of computer-readable media include HDD (Hard Disk Drive), SSD (Solid State Disk), SDD (Silicon Disk Drive), ROM, RAM, CD-ROM, magnetic tape, floppy disk, optical data storage device, and the like.

[0091] Furthermore, a computer-readable medium may include storage, which may be a server or cloud storage accessible by an electronic device via communication. In this case, the computer can download the program according to the present invention from the server or cloud storage via wired or wireless communication.

[0092] Furthermore, in the present invention, the computer described above is an electronic device equipped with a processor, that is, a CPU (Central Processing Unit), and its type is not particularly limited.

[0093] On the other hand, the above detailed description should not be construed as limiting in all respects and should be considered as an example. The scope of the present invention should be determined by a reasonable interpretation of the appended claims, and all changes within the equivalent scope of the present invention are included in the scope of the present invention.

Claims

1. Receiving a rebar image obtained by photographing a space in which a plurality of rebars are arranged; Inputting the rebar image into a pre-trained space detection model for estimating a mask area corresponding to the space between the plurality of rebars, and obtaining a plurality of mask areas shown in the rebar image; Classifying the plurality of mask areas into a similar group and a dissimilar group based on the similarity of the sizes of the plurality of mask areas; Providing an arrangement state of the plurality of rebars by displaying the plurality of mask areas corresponding to at least one of the similar group and the dissimilar group on the rebar image. A method for providing a rebar arrangement state, comprising:

2. The step of classifying the plurality of mask areas into a similar group and a dissimilar group includes: Counting the number of a plurality of pixels belonging to each of the plurality of mask areas; Classifying the plurality of mask areas into a first group and a second group based on the counted number of the plurality of pixels. The method for providing a rebar arrangement state according to claim 1, comprising:

3. The step of classifying the plurality of mask areas into a similar group and a dissimilar group includes: Counting the number of the plurality of mask areas classified into the first group and the second group respectively; Identifying one of the first group and the second group in which the counted number of the plurality of mask areas is the largest; Classifying the plurality of mask areas belonging to the identified group into a first sub-group and a second sub-group based on the number of pixels for each of the plurality of mask areas belonging to the identified group. The method for providing a rebar arrangement state according to claim 2, comprising:

4. The step of classifying the plurality of mask areas into a similar group and a dissimilar group includes: Counting the number of the plurality of mask areas classified into the first sub-group and the second sub-group respectively; Among the first lower group and the second lower group, identifying as the similar group any one of the lower groups with the largest number of the counted plurality of mask regions, and identifying, as the dissimilar group, the other mask regions among the plurality of mask regions obtained from the reinforcing bar image, excluding the plurality of mask regions belonging to the similar group. The method for providing a reinforcing bar arrangement state according to claim 3 includes this step.

5. The step of obtaining a plurality of mask regions shown in the reinforcing bar image includes: The method for providing a reinforcing bar arrangement state according to claim 1, wherein each of the plurality of regions corresponding to the spaces between the plurality of reinforcing bars formed by the plurality of reinforcing bars arranged in a grid pattern is obtained as the mask region.

6. The pre-trained space detection model: is trained based on a space detection model training method, The space detection model training method includes: storing a training reinforcing bar image; enhancing the training reinforcing bar image to generate a plurality of training reinforcing bar images; for each of the plurality of training reinforcing bar images, labeling a training mask region corresponding to the space between the plurality of reinforcing bars; training the space detection model to estimate the mask region in the reinforcing bar image by using the plurality of training reinforcing bar images and the training mask regions labeled for each of the plurality of training reinforcing bar images. The method for providing a reinforcing bar arrangement state according to claim 1 includes this step.

7. The training reinforcing bar image: is a reinforcing bar image taken by arranging a photographing device at different positions with respect to the space in which the plurality of reinforcing bars are arranged, according to the method for providing a reinforcing bar arrangement state according to claim 6.

8. The step of generating a plurality of training reinforcing bar images includes: generating the plurality of training reinforcing bar images obtained by editing the training reinforcing bar image by at least one of inversion, rotation, contrast adjustment, and saturation adjustment, according to the method for providing a reinforcing bar arrangement state according to claim 6.

9. a storage unit for storing a reinforcing bar image obtained by photographing a space in which a plurality of reinforcing bars are arranged; and a control unit for inputting the reinforcing bar image into a pre-trained space detection model for estimating a mask region corresponding to the space between the plurality of reinforcing bars, and obtaining a plurality of mask regions shown in the reinforcing bar image. The control unit: A steel bar arrangement state providing system that classifies the plurality of mask regions into a similar group and a dissimilar group based on the similarity of the sizes between the plurality of mask regions, and provides the arrangement state of the plurality of steel bars by displaying the plurality of mask regions corresponding to at least one of the similar group and the dissimilar group on the steel bar image.

10. A program that is executed by one or more processes in an electronic device and is stored in a computer-readable recording medium, wherein the program receives a steel bar image obtained by photographing a space in which a plurality of steel bars are arranged; inputs the steel bar image into a space detection model that has been pre-trained to estimate a mask region corresponding to the space between the plurality of steel bars, and obtains a plurality of mask regions shown in the steel bar image; classifies the plurality of mask regions into a similar group and a dissimilar group based on the similarity of the sizes between the plurality of mask regions; and includes instruction words for performing a step of providing the arrangement state of the plurality of steel bars by displaying the plurality of mask regions corresponding to at least one of the similar group and the dissimilar group on the steel bar image. A program stored in a computer-readable recording medium.

Images