Learning model-based dimension measurement device and method

Abstract

The dimension measurement device and method according to an embodiment of the present invention can obtain dimension information of a specific object in an image of interest using an already trained learning model, thereby obtaining highly reliable dimension measurement result data with improved accuracy.

Description

[Technical field]

[0001] This application claims the benefit of the filing dates of Korean Patent Application No. 10-2022-0109769 filed with the Korean Intellectual Property Office on August 31, 2022, and Korean Patent Application No. 10-2023-0103918 filed with the Korean Intellectual Property Office on August 9, 2023, and all of the contents disclosed in the documents of said Korean patent applications are incorporated herein by reference.

[0002] The present invention relates to a dimension measurement apparatus and method, and more particularly to a dimension measurement apparatus and method that uses a pre-trained learning model to obtain dimensional information of a particular object in an image of interest. [Background technology]

[0003] 2. Description of the Related Art As the depletion of fossil fuels has caused the prices of energy sources to rise and as concern over environmental pollution has grown, the demand for secondary batteries as an environmentally friendly alternative energy source has been increasing rapidly.

[0004] Among secondary batteries, lithium batteries are being applied to many industrial fields, such as mobile application devices, automobiles, robots, and energy storage devices, as a countermeasure to the recent problems of environmental regulations and high crude oil prices.

[0005] Such lithium batteries are generally classified into cylindrical, prismatic, and pouch types depending on the shape of the exterior material in which the electrode assembly is housed.

[0006] Among these, cylindrical batteries are provided in a battery pack (Cell To Pack, CTP) structure consisting of a plurality of battery cells. In other words, cylindrical batteries are provided in a structure in which electrodes consisting of a separator between a positive electrode and a negative electrode are wound and inserted into a battery can to assemble them.

[0007] After the assembly process, such cylindrical batteries undergo quality inspection.

[0008] Generally, in quality inspection of cylindrical batteries, dimensions of a corresponding area are measured using an image of a measurement target to determine the position of a weld, the tensile force of a weld, or the presence or absence of defects in the battery structure that occur during a process.

[0009] Conventional dimension measuring devices detect the boundary (edge) of the measurement object corresponding to a preset threshold brightness value from the target image, which is a brightness image (gray model), and extract pixel information of the detected boundary to obtain dimensional information of the measurement object.

[0010] However, conventional dimension measuring devices have a drawback in that if an error occurs in the program logic already designed by an engineer, the boundary of the measurement object on the target image is erroneously detected, reducing the reliability of the dimension information. Summary of the Invention [Problem to be solved by the invention]

[0011] SUMMARY OF THE PRESENT EMBODIMENTS In order to solve the above problems, an object of the present invention is to provide a dimension measuring device.

[0012] Another object of the present invention to solve the above problems is to provide a dimension measuring method. [Means for solving the problem]

[0013] To achieve the above object, one embodiment of the present invention provides a learning model-based dimension measurement device that includes a memory and a processor that executes at least one instruction stored in the memory, the at least one instruction including an instruction to acquire an image of interest including a measurement object whose dimensions are to be measured from an object image, and an instruction to input the image of interest into an already learned learning model and output dimensional information of the measurement object as result data.

[0014] Here, the pre-trained learning model may be a deep-learning based learning model.

[0015] The command to output the dimensional information as result data may include a command to segment the image of interest into physical objects using the previously trained learning model, and label dimensional measurement points of the measurement objects.

[0016] Meanwhile, the command to acquire the image of interest may be pre-trained to acquire the image of interest including the measurement target from the target image using a rule-based algorithm.

[0017] In addition, the command to obtain the image of interest may be pre-trained to perform image pre-processing on the target image to obtain the image of interest.

[0018] To achieve the above object, a learning model-based dimensional measurement method according to another embodiment of the present invention includes a step of acquiring an image of interest including a measurement object whose dimensions are to be measured from an object image, and a step of inputting the image of interest into an already learned learning model and outputting dimensional information of the measurement object as result data.

[0019] Here, the already-trained learning model may be a deep-learning-based learning model.

[0020] The step of outputting the dimensional information as result data may include a step of segmenting the image of interest into physical objects using the already-trained learning model and labeling dimensional measurement points of the measurement objects.

[0021] Meanwhile, the step of acquiring the image of interest may include the step of acquiring the image of interest including the measurement target from the target image using a rule-based algorithm.

[0022] Moreover, the step of acquiring the image of interest may include a step of performing image pre-processing on the target image to acquire the image of interest. Effect of the Invention

[0023] The dimension measurement device and method according to an embodiment of the present invention can obtain dimension information of a specific object in an image of interest using an already trained learning model, thereby obtaining highly reliable dimension measurement result data with improved accuracy. [Brief description of the drawings]

[0024] [Figure 1] 1 is an image of a weld that was erroneously detected when measured using a general dimension measuring device. [Diagram 2] 1 is a block diagram of a dimension measuring device according to an embodiment of the present invention. [Diagram 3] FIG. 2 is a flow diagram of a dimension measuring method using the dimension measuring device according to the embodiment of the present invention. [Figure 4] 1 is an image for explaining an image of interest of a dimension measuring device according to an embodiment of the present invention. [Diagram 5] 4 shows a first current collecting plate image of a dimension measuring device according to one embodiment of the present invention and an image of interest of a weld extracted from the first current collecting plate image. [Figure 6] 6 is a pre-processed image for extracting the image of interest of FIG. 5 according to one embodiment of the present invention; [Figure 7] 10 is a second current collecting plate image and an image of interest extracted from the second current collecting plate image according to another embodiment of the present invention; [Figure 8] 8 is a pre-processed image for extracting the image of interest of FIG. 7 according to another embodiment of the present invention; [Figure 9] 13 is an image for explaining a method of measuring the dimensions of an electrode portion using a learning model that has already been trained in a dimension measuring device according to yet another embodiment of the present invention. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0025] Since the present invention can be modified in various ways and has various embodiments, specific embodiments will be illustrated in the drawings and described in detail in the detailed description. However, it is understood that this is not intended to limit the present invention to the specific embodiment, but includes all modifications, equivalents, and alternatives within the spirit and technical scope of the present invention. Similar reference numerals are used for similar components while describing each drawing.

[0026] Terms such as first, second, A, B, etc. may be used to describe various components, but the components should not be limited by the terms. The terms are used only to distinguish one component from another. For example, a first component may be named a second component, and similarly, a second component may be named a first component, without departing from the scope of the present invention. The term "and / or" includes a combination of multiple associated listed items or any of multiple associated listed items.

[0027] When a component is referred to as being "coupled" or "connected" to another component, it should be understood that the component may be directly coupled or connected to the other component, but there may also be other components in between. In contrast, when a component is referred to as being "directly coupled" or "directly connected" to another component, it should be understood that there are no other components in between.

[0028] The terms used in this application are merely used to describe certain embodiments and are not intended to limit the present invention. A singular expression includes a plural expression unless the context clearly indicates otherwise. In this application, the terms "include" or "have" are intended to specify the presence of features, numbers, steps, operations, components, parts, or combinations thereof described in the specification, and are not intended to preclude the presence or additional possibility of one or more other features, numbers, steps, operations, components, parts, or combinations thereof.

[0029] Unless otherwise defined, all terms used herein, including technical or scientific terms, have the same meaning as commonly understood by a person of ordinary skill in the art to which the present invention belongs. Terms as defined in commonly used dictionaries should be interpreted as having a meaning consistent with the meaning they have in the context of the relevant art, and should not be interpreted as having an ideal or overly formal meaning unless expressly defined in this application.

[0030] Figure 1 shows an image of a weld that was erroneously detected when measured using a typical dimensional measuring device.

[0031] Referring to FIG. 1, a typical dimension measuring device can inspect the quality of a completed battery after the assembly process.

[0032] Generally, during quality inspection of a battery, dimensional information on at least one measurement object is required to check the position and tensile force of a weld or to determine whether or not there is a defect in the battery structure that occurs during a process.

[0033] Therefore, a general dimension measuring device obtains an object image including a measurement object, and obtains the boundary (edge) of the measurement object by applying a pre-designed program logic to the object image.The dimension measuring device then measures the dimension of the measurement object based on pixel information of the boundary.

[0034] For example, a typical dimension measuring device applies a pre-designed program logic to a brightness image (Gray Model) of a measurement object. As a result, the dimension measuring device detects pixels corresponding to a pre-set detection condition value (e.g., a threshold brightness value) and obtains the boundary of the measurement object based on the detected pixels. The dimension measuring device then measures the dimension of the measurement object based on pixel information of the boundary. Here, the pre-designed program logic may be a rule-based algorithm model. In addition, the at least one pixel information may include at least one of pixel height, width, pixel maximum value (Peak White), pixel minimum value (Peak Dark), and aspect ratio.

[0035] Meanwhile, program logic applied to a general dimension measuring device can be pre-designed based on a pre-acquired target image, and therefore, when the brightness of the target image changes or the position of the measurement target in the target image changes, the general dimension measuring device to which the pre-designed program logic is applied has a drawback in that the reliability of the measured dimension information is reduced because the boundary of the measurement target is erroneously detected or the measurement range of the boundary is erroneously specified.

[0036] Therefore, in order to solve such problems, the present invention describes a dimension measurement device and method with improved reliability using a pre-trained machine learning model.

[0037] DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS Hereinafter, preferred embodiments of the present invention will be described in detail with reference to the accompanying drawings.

[0038] FIG. 2 is a block diagram of a dimension measuring device according to an embodiment of the present invention.

[0039] 2, the dimension measuring device according to the embodiment of the present invention can be used in a quality inspection carried out after an assembly process of a cylindrical battery, but is not limited to the above and can be applied to all processes requiring dimension measurement.

[0040] The dimension measuring device can input an object image to a learning model that has already been trained to obtain dimensional information of the object to be measured. Here, the object image may be an image of at least a part of the inside of a battery that has been assembled.

[0041] More specifically, the dimension measuring device can obtain an object image including a measurement object, and extract an image of interest including a Region of Interest (ROI) of the measurement object from the object image.

[0042] The dimension measuring device can then input the image of interest into the already trained learning model to obtain dimensional information of the measurement object.

[0043] The dimension measuring device will be described in detail by components. The dimension measuring device may include a memory 100, a processor 200, a transceiver 300, an input interface 400, an output interface 500 and a storage device 600.

[0044] According to the embodiment, the components 100, 200, 300, 400, 500, and 600 included in the dimension measuring apparatus are connected by a bus 700 and can communicate with each other.

[0045] The memory 100 and the storage device 600 in the above configurations 100, 200, 300, 400, 500, and 600 may be composed of at least one of a volatile storage medium and a non-volatile storage medium. For example, the memory 100 and the storage device 600 may be composed of at least one of a read only memory (ROM) and a random access memory (RAM).

[0046] Among other things, memory 100 may contain at least one instruction that is executed by processor 200 .

[0047] According to an embodiment, at least one instruction may include an instruction to obtain an image of interest including a measurement object whose dimensions are to be measured from the object image, and an instruction to input the image of interest into an already trained learning model and output dimensional information of the measurement object as result data.

[0048] Here, the already-trained learning model may be a deep-learning-based learning model.

[0049] The command to output the dimensional information as result data may include a command to segment the image of interest into physical objects using the previously trained learning model, and label dimensional measurement points of the measurement objects.

[0050] Meanwhile, the command to acquire the image of interest may be pre-trained to acquire the image of interest including the measurement target from the target image using a rule-based algorithm.

[0051] In addition, the command to obtain the image of interest may be pre-trained to perform image pre-processing on the target image to obtain the image of interest.

[0052] Meanwhile, the processor 200 may refer to a central processing unit (CPU), a graphics processing unit (GPU), or a dedicated processor on which the method according to the embodiment of the present invention is performed.

[0053] The processor 200 is capable of executing at least one program command stored in the memory 100, as described above.

[0054] FIG. 3 is a flow diagram of a dimension measuring method using the dimension measuring device according to the embodiment of the present invention.

[0055] 3, the dimension measuring device according to the embodiment of the present invention may obtain an object image (S1000). At this time, the object image may be an image of at least a part of the inside of a battery after assembly is completed, and may include a measurement object for dimension measurement.

[0056] According to one embodiment, the target image may be a video image of at least one weld to be measured. For example, the weld may be a weld formed on a first or second current collector in a battery. Here, the first and second current collectors may be a negative and positive current collectors, respectively.

[0057] According to another embodiment, the target image may be an X-ray image of at least one electrode portion to be measured, for example, the electrode portion may be a first electrode or a second electrode of a battery.

[0058] FIG. 4 is an image for explaining an image of interest of the dimension measuring device according to the embodiment of the present invention.

[0059] 3 and 4, the dimension measuring device can obtain an image of interest from an object image (S3000). Here, the image of interest may be a partial image obtained by extracting a region of interest (ROI) of the object from the object image. According to an embodiment, the dimension measuring device can extract the image of interest after pre-processing the object image using a rule-based algorithm.

[0060] Thereafter, the dimension measuring device can acquire dimensional information on the measurement object from the image of interest using the already trained learning model (S5000). In other words, the dimension measuring device can input the image of interest to the already trained learning model to acquire dimensional information on the measurement object. Here, the machine learning model may be a classification learning model using deep learning.

[0061] More specifically, the dimension measuring device may segment an image of interest, which is input data (Feature), into measurement objects (Class) whose dimensions are to be measured, using a classification learning model using deep learning that is pre-trained using learning data. Then, the dimension measuring device may label measurement points of the measurement objects, measure the pixel distance between the two labeled points, and output the label as result data.

[0062] 5 and 6 are images for explaining a method for measuring dimensions of a weld using a weld dimension measuring device according to an embodiment of the present invention. More specifically, FIG. 5 is a first current collecting plate image and an image of interest of a weld extracted from the first current collecting plate image, and FIG. 6 is an image preprocessed to extract the image of interest of FIG. 5.

[0063] 5, the dimension measuring device according to the embodiment of the present invention can obtain a first current collector image, which is an object image captured of a battery that has been assembled as described above. For example, the first current collector may be a negative current collector.

[0064] The dimension measuring device can then pre-process the first current collecting plate image, which is a target image, to obtain an image of interest, which may be an image including a weld, which is a region of interest (ROI) in the first current collecting plate image.

[0065] For example, if the first current collecting plate image includes a plurality of welds, the dimension measuring device can individually extract a plurality of images of interest each including the welds.

[0066] To explain the method of extracting an image of interest in more detail with reference to FIG. 6, the dimension measuring device can pre-process the first current collector image to detect the position of at least one weld from the first current collector image.

[0067] For example, in the case of a cylindrical battery in which an assembly process has been completed, a plurality of electrode tabs may be provided in contact with a region on the first current collector plate in a state where they are overlapped by an electrode wound inside a can. At this time, at least one weld may be formed on the first current collector plate for electrical contact of the electrode tabs. At this time, the at least one weld may be formed in a diagonal direction on the first current collector plate (see FIG. 5). Thus, the dimension measuring device may perform a pre-processing to rotate an image of the first current collector plate so that the weld is positioned horizontally or vertically.

[0068] The dimension measuring device can then confirm the location of the weld, which is an area of ​​interest, based on the rotated first current collector plate image.

[0069] To explain in more detail with reference to an embodiment, the dimension measuring device may convert the rotated first current collecting plate image into a gray value image according to a preset rule-based algorithm. Then, the dimension measuring device may input a preset condition value to primarily obtain an edge of a weld in the first current collecting plate image. For example, the preset condition value may be a preset pixel brightness value. In other words, the dimension measuring device may primarily extract at least one pixel having a preset pixel brightness value as the edge of a weld. Then, the dimension measuring device may confirm the position of the weld based on information of at least one pixel extracted as the edge of the weld.

[0070] Then, the dimension measuring device can convert the first current collector plate image of a gray value into a binary image. Then, the dimension measuring device can obtain the start and end coordinate points of the weld as shown in FIG. 6.

[0071] According to one embodiment, when the weld in the first current collecting plate image is located in a horizontal direction, the start coordinate point may be a pixel coordinate value located on the left edge of the weld based on the x-axis, and the end coordinate point may be a pixel coordinate value located on the right edge of the weld based on the x-axis.

[0072] According to another embodiment, when the weld in the first current collector image is located in a vertical direction, the start coordinate point may be a pixel coordinate value of the weld located at the top end based on the y axis, and the end coordinate point may be a pixel coordinate value of the weld located at the bottom end based on the y axis.

[0073] According to the embodiment, the dimension measuring device may obtain start and end coordinate points (Start Point, End Point) in the first current collecting plate image based on a predefined pixel brightness value. For example, when the welded portion in the first current collecting plate image is located in a horizontal direction, the dimension measuring device may check the pixel brightness value of each coordinate in the x-axis direction and compare it with the predefined pixel brightness value. In this case, the dimension measuring device may set the first coordinate point having a brightness value larger or smaller than the predefined pixel brightness value as the start coordinate point (Start Point). Also, the dimension measuring device may obtain the last coordinate point having a brightness value larger or smaller than the predefined pixel brightness value as the end coordinate point (End Point).

[0074] Meanwhile, when the welded portion in the first current collecting plate image is located in a vertical direction, the dimension measuring device may check pixel brightness values ​​for each coordinate in a y-axis direction and compare them with a preset pixel brightness value. At this time, the dimension measuring device may set a first coordinate point having a brightness value larger or smaller than the preset pixel brightness value as a start coordinate point. Also, the dimension measuring device may obtain a last coordinate point having a brightness value larger or smaller than the preset pixel brightness value as an end coordinate point.

[0075] Thereafter, the dimension measuring device can obtain at least one dimension measuring point based on the extracted boundary information of the weld and the start and end coordinate points (Start Point, End Point). For example, the dimension measuring device can obtain at least one dimension measuring point of the weld using a sub-pixel algorithm. Here, the sub-pixel algorithm may be a method of determining a point having a brightness value of 1 pixel or less as a corner of an object using an interpolation method. In this way, the dimension measuring device can finally obtain the boundary of the weld in the first current collecting plate image based on the at least one dimension measuring point information.

[0076] Therefore, the dimension measuring device according to the embodiment of the present invention can extract an image of interest from the first current collecting plate image based on the boundary information of the weld and the dimension measuring point information finally acquired by the pre-processing process. Then, the extracted image of interest can be image-processed (overlaid) by the dimension measuring device.

[0077] The dimensioning device can then input the image-processed image of interest into a pre-trained deep-learning based classification learning model.

[0078] Thereafter, the dimension measuring device may use the pre-trained classification learning model to segment the image of interest into physical units, i.e., the welds of the first current collecting plate, and then label the dimension measuring points of the welds. For example, the dimension measuring device may classify objects in pixel units for the image of interest, and then label the dimension measuring points of the welds.

[0079] Thereafter, the dimension measuring device may obtain the dimension of the measurement object as result data by measuring the pixel distance between the two labeled dimension measuring points based on a pre-trained deep-learning based classification learning model. Here, the classification learning model may be a model that transforms at least one learning data into various forms and then uses the transformed data as learning data in order to increase the amount of learning data. For example, the learning data may be transformed in various ways, such as by flipping up and down or left and right, changing brightness, saturation or color tone, changing contrast, rotating or moving, and adding noise.

[0080] 7 and 8 are images for explaining a method for measuring the dimensions of a weld using a dimension measuring device according to another embodiment of the present invention. More specifically, FIG. 7 is a second current collector plate image and an image of interest extracted from the second current collector plate image, and FIG. 8 is an image preprocessed to extract the image of interest in FIG. 7.

[0081] 7 and 8, the dimension measuring device according to another embodiment of the present invention can obtain the second current collector image, which is the target image, as described above. Here, the second current collector image is an image taken after the assembly process and may be an image of the positive current collector plate.

[0082] Then, the dimension measuring apparatus can perform image pre-processing based on the second collector plate image to obtain an image of interest.

[0083] More specifically, in general, the welds on the second current collecting plate image may be arranged in a specific pattern symmetrical with respect to the origin as shown in Fig. 7. Thus, the dimension measuring device may deform the target image (polar space data) so that the weld area, which is the region of interest of the second current collecting plate image, is positioned horizontally.

[0084] The dimension measuring device can then determine the location of the weld, which is an area of ​​interest, based on the deformed second current collector plate image.

[0085] To explain in more detail with reference to an embodiment, the dimension measuring device can convert the rotated second current collecting plate image into a gray value image according to a preset rule-based algorithm, and then input a predefined condition value to obtain the edge of the weld in the second current collecting plate image.

[0086] Then, the dimension measuring device can convert the first current collecting plate image of a brightness image (gray value) into a binary image. Then, the dimension measuring device can obtain start and end coordinate points (Start Point, End Point) of the weld. According to an embodiment, the dimension measuring device can obtain start and end coordinate points (Start Point, End Point) in the second current collecting plate image based on a predefined pixel brightness value.

[0087] Thereafter, the dimension measuring device can obtain at least one dimension measuring point based on the extracted boundary information of the weld and the start and end coordinate points (Start Point, End Point), thereby allowing the dimension measuring device to extract an image of interest from the second current collecting plate image based on the boundary information of the weld and the dimension measuring point information obtained by the pre-processing process.

[0088] The extracted image of interest can then be overlaid by the dimension measuring device.

[0089] The dimensioning device can then input the image-processed image of interest into a pre-trained deep-learning based classification learning model.

[0090] The dimension measurement device can then use the pre-trained classification learning model to segment the image of interest into physical units, that is, the welds of the second current collecting plate.

[0091] The dimension measurement device can then label the dimension measurement points of the weld based on a pre-trained deep-learning based classification learning model.

[0092] Then, the dimension measuring device can obtain the dimension of the measurement object as result data by measuring the pixel distance between the two labeled dimension measuring points. Here, the classification learning model may be a model that transforms at least one learning data into various forms and then uses the transformed data as learning data in order to increase the amount of learning data. For example, the learning data may be transformed in various ways, such as by flipping up and down or left and right, changing brightness, saturation or color tone, changing contrast, rotating or moving, and adding noise.

[0093] FIG. 9 is an image for explaining a method of measuring the dimensions of an electrode portion using a learning model that has already been trained in a dimension measuring device according to still another embodiment of the present invention.

[0094] 9, the dimension measuring device according to another embodiment of the present invention can obtain an image of an electrode part photographed after an assembly process as a target image, as described above. For example, the image of the electrode part may be an X-ray image.

[0095] Thereafter, the dimension measuring device can extract an image of interest including the connection portion of the positive and negative electrodes, which is a region of interest (ROI), from the image of the electrode portion.

[0096] The extracted image of interest can then be overlaid by the dimension measuring device.

[0097] Then, the dimension measuring device can input the processed image of interest to a classification learning model using pre-trained deep learning, thereby obtaining length data of an electrode portion connecting a negative electrode end and a positive electrode end of an electrode.

[0098] More specifically, the dimension measuring device may segment an image of interest input to the classification learning model into physical units based on a pre-trained deep-learning based classification learning model. According to an embodiment, the classification learning model may be a semantic segmentation model. For example, the dimension measuring device may segment an image of interest into electrode part images.

[0099] Then, the dimension measuring device can label the dimension measuring points of the weld based on the classification learning model. For example, the dimension measuring points in the electrode part can be one end (first point) of the negative electrode of the electrode located adjacent to the tab area and one end (second point) of the positive electrode of the electrode located adjacent to the electrode body (area where the holding part of the electrode is formed). In this case, the dimension measuring device can be labeled in a plurality of places corresponding to the number of turns of the electrode.

[0100] Thereafter, the dimension measuring device may obtain the length of the electrode to be measured as result data by measuring the pixel distance between the labeled first and second points based on a pre-trained deep-learning based classification learning model. Here, the classification learning model may be a model that transforms at least one learning data into various forms and then uses the transformed data as learning data in order to increase the amount of learning data. For example, the learning data may be transformed in various ways, such as by flipping up and down or left and right, changing brightness, saturation or color tone, changing contrast, rotating or moving, and adding noise.

[0101] The dimension measuring apparatus and method according to the embodiment of the present invention have been described above.

[0102] The dimension measurement device and method according to an embodiment of the present invention can obtain dimension information of a specific object in an image of interest using an already trained learning model, thereby obtaining highly reliable dimension measurement result data with improved accuracy.

[0103] The operation of the method according to the embodiment and the experimental example of the present invention can be embodied as a computer readable program or code in a computer readable recording medium. The computer readable recording medium includes all kinds of recording devices in which data that can be read by a computer system is stored. In addition, the computer readable recording medium can be distributed to computer systems connected through a network, and the computer readable program or code can be stored and executed in a distributed manner.

[0104] Additionally, the computer-readable recording medium may include hardware devices specially configured to store and execute program instructions, such as ROM, RAM, flash memory, etc. The program instructions may include not only machine code, such as produced by a compiler, but also high-level language code that can be executed by a computer using an interpreter, etc.

[0105] Some aspects of the invention have been described in the context of an apparatus, but it may also be presented in terms of a corresponding method, where a block or apparatus corresponds to a method step or a feature of a method step. Similarly, aspects described in the context of a method may be presented in terms of a corresponding block or item or feature of a corresponding apparatus. Some or all of the method steps may be performed by (or using) a hardware apparatus, such as, for example, a microprocessor, a programmable computer, or an electronic circuit. In some embodiments, one or more of the most important method steps may be performed by such an apparatus.

[0106] Although the present invention has been described with reference to preferred embodiments thereof, those skilled in the art will understand that various modifications and variations of the present invention can be made without departing from the spirit and scope of the present invention as set forth in the following claims. [Explanation of symbols]

[0107] 100: Memory 200: Processor 300: Transmitter / receiver 400: Input interface device 500: Output interface device 600: Storage device 700: Bus

Claims

1. A learning model-based dimension measurement device, comprising: memory; and a processor for executing at least one instruction stored in the memory; The at least one instruction: instructions for acquiring an image of interest including an object to be measured from the object image; and A dimension measuring device including instructions for inputting the image of interest into a previously trained learning model and outputting dimensional information of the measurement object as result data.

2. The dimension measuring device according to claim 1 , wherein the already-trained learning model is a deep-learning based learning model.

3. The command to output the dimensional information as result data is The dimension measurement device of claim 1 , further comprising instructions for segmenting the image of interest into physical objects and labeling dimension measurement points of the objects using the already trained learning model.

4. The instructions for acquiring the image of interest include:

2. The dimension measuring apparatus of claim 1, further comprising instructions for obtaining an image of interest including the measurement object from the object image using a rule-based algorithm.

5. The instructions for acquiring the image of interest include: The dimensioning apparatus of claim 4 , further comprising instructions for performing image pre-processing on the object image to obtain the image of interest.

6. 1. A learning model based dimensioning method, comprising: acquiring an image of interest including a measurement object whose dimensions are to be measured from the object image; and A dimensional measurement method comprising the steps of inputting the image of interest into an already trained learning model and outputting dimensional information of the object to be measured as result data.

7. The dimension measurement method according to claim 6 , wherein the already-trained learning model is a deep-learning based learning model.

8. The step of outputting the dimensional information as result data includes: The dimension measurement method according to claim 6, further comprising a step of segmenting the image of interest into physical objects using the already-trained learning model, and labeling dimension measurement points of the measurement objects.

9. The step of acquiring an image of interest includes: The dimension measurement method according to claim 6, further comprising the step of obtaining an image of interest including the measurement object from the object image using a rule-based algorithm.

10. The step of acquiring an image of interest includes: The method of claim 9 , further comprising performing image pre-processing on the object image to obtain the image of interest.

Images