Method for Automatic Inspection and Management of Sewing Defects and System Therefor

Abstract

A camera and a sewing defect inspector are installed on each sewing machine. The camera captures the sewing results of the matched machine in real time only when a shooting instruction is provided, and outputs the resulting images. The sewing defect inspector detects whether the sewing machine is in operation and provides a shooting instruction to the camera of the matched machine only while it is running to collect the captured images of the sewing results. These images are then fed as input to a sewing defect detection and classification model to detect and classify defects, and the detected sewing defect data is transmitted in real time to a server computer. The server computer receives, stores, and manages sewing defect data from the inspector; provides real-time defect alarms to relevant user terminals; utilizes the collected sewing defect data to generate and visually display statistical data regarding defect status and history information by factory, production line, and worker; monitors the status of the sewing defect inspector; and provides factory / production line management and sewing defect image download services.

Description

Technology Field

[0001] The present invention relates to the field of quality inspection technology for sewing processes, and more specifically, to automated sewing defect inspection technology based on computer vision and artificial intelligence. Background Technology

[0002] In garment manufacturing facilities, the degradation of finished product quality results in significant waste of production time and costs. To reduce this waste, quality inspections are conducted on garment production lines; however, most existing sewing factories rely on manual inspection by workers, which limits their ability to maintain good quality. Figure 1 conceptually illustrates a conventional sewing defect detection method, illustrating a scenario where quality control (QC) is performed only after the work has progressed to the final process, process 37, even though the sewing defect occurred at process 17. Assuming a case where a defect is detected during a quality inspection conducted after the final process and it is belatedly discovered that the defect originated at process 17, the loss incurred from returning the work to process 17 to resolve the defect is substantial. In some cases, it may be necessary to re-perform not only process 17 but also at least some of the subsequent processes. To address these drawbacks, a method of placing an additional Quality Manager (QM) in the middle of the sewing line is sometimes used, but this leads to increased labor costs.

[0003] To overcome the limitations of manual quality inspection, a technology has been disclosed that automatically detects defective products at the point of occurrence by introducing a sewing defect inspection device. For example, a technology has been disclosed that photographs a sewing workpiece with a camera and performs a defect inspection thereon. Korean Published Patent No. 10-2024-0174876 discloses a technology in which 'at least one camera is positioned at a location capable of photographing a worker performing production operations on a production line, thereby acquiring image data regarding the worker's production movements and the products resulting from those movements when the production line is in operation, applying the acquired image data to an artificial intelligence algorithm to verify analysis results for multiple pre-set objects included in stitching items, fabric items, and production operation items within the image data, verifying errors by production line based on the analysis results using the artificial intelligence algorithm, and verifying work efficiency by worker through a separate analysis based on standard data.' While such technology improves upon various disadvantages of manual quality inspection methods, it raises another problem that needs to be solved. This is precisely how to efficiently collect, store, and manage the data collected for defect (quality) inspection. Failure to effectively control camera recording results in an excessive amount of video data to be processed, causing a problem that degrades the performance of the entire system. Perhaps taking this into account, the aforementioned conventional technology reduces the generated video data by installing cameras on each work line rather than on every sewing machine.

[0004] Furthermore, for effective quality control and productivity improvement, it is necessary to systematize the process so that defects can be detected in real time and immediately recognized and responded to by field managers and quality personnel. Unless these challenges are resolved, effective quality control will be difficult.

[0005] There were also limitations in utilizing data after detecting sewing defects. Relying solely on defect detection at the sewing machine unit level made it difficult and limited to effectively control sewing quality across the entire production flow of the sewing line or factory by utilizing the detected defect data. Furthermore, the lack of a proper response system for sewing defects meant that defect situations were not immediately shared, leading to delays in response by quality managers and workers.

[0006] Furthermore, quality standards often vary by buyer, factory, and production line; therefore, to address these differing standards, a server-based system is required to integrate and manage the varying requirements of each supplier. Prior art literature

[0007] (1) Republic of Korea Published Patent No. 10-2024-0174876 (Method and apparatus for managing a sewing factory using artificial intelligence) (2) Republic of Korea Patent Registration No. 10-1917157 (Smart monitoring method and system for a clothing production factory) (3) Republic of Korea Patent Publication No. 10-2022-0065154 (Sewing production management system using IoT technology) The problem to be solved

[0008] The present invention aims to provide an automated sewing defect inspection and management method and a system for the same, which can detect various sewing defects in real time by utilizing computer vision and deep learning technologies, and reduce the inefficiency of over-detection by generating a minimum amount of image data to be used for defect detection.

[0009] In addition, the present invention has another objective of providing an automatic inspection and management method for sewing defects and a system for the same, which collects, manages, and visualizes real-time defect image information detected for each sewing machine and provides appropriate notification and analysis functions for each user.

[0010] Furthermore, another objective of the present invention is to provide an automated sewing defect inspection and management method and a system therefor that support efficient quality control and process improvement by processing defect data generated in various production sites in a standardized manner and providing customized functions according to user authority.

[0011] The problems that the present invention aims to solve are not limited to the problems described above, and may be extended in various ways without departing from the spirit and scope of the present invention. means of solving the problem

[0012] An automatic sewing defect inspection and management system according to embodiments for realizing the objectives of the present invention comprises a plurality of cameras, a plurality of sewing defect inspectors, and a server computer device. The plurality of cameras are installed for each sewing machine installed in a sewing factory, and each captures the sewing work result of the matched sewing machine in real time only while a shooting instruction is provided, and outputs a sewing result image. The plurality of sewing defect inspectors are installed for each sewing machine installed in a sewing factory, and each is configured to detect whether the matched sewing machine is in operation and to provide a shooting instruction to the camera of the matched sewing machine only while it is in operation, thereby collecting the sewing result image captured by the matched camera; to provide the collected sewing result image as input to a sewing defect detection and defect type classification model of an artificial neural network-based computer program running in real time to determine whether a sewing defect has occurred and to classify the type of the detected sewing defect; and to transmit sewing defect data obtained as a result of the detection and classification of defects to a server computer device in real time. The server computer device is configured to be communicably connected to the plurality of sewing defect inspection devices via the Internet and to perform at least one of the following functions by executing a predetermined server computer program: receiving, storing, and managing sewing defect data transmitted from each sewing defect inspection device; providing defect occurrence alarms in real time to the front-end terminal of a relevant user; generating and visually displaying statistical data regarding defect status and history information by factory, production line, and worker using the collected sewing defect data; monitoring the status of sewing defect inspection devices; managing factories / production lines; and downloading defect images.

[0013] In exemplary embodiments, each of the plurality of sewing defect inspection devices may include a computing device and a sewing machine operation detection sensor unit. The computing device may execute the computer program comprising: a camera manager module that provides the shooting instruction to a matched camera only while an operation detection signal is input to capture the sewing work result, and performs preprocessing by collecting the sewing result image output from the camera, adjusting its size to fit the model input, and normalizing it; a model manager module that inputs the preprocessed image into the sewing defect detection and defect type classification model to detect sewing defects and performs data processing to classify the detected sewing defect images into a predefined class; and a communication manager module that transmits sewing defect data including the detected sewing defect image data to the server computer device. The sewing machine operation detection sensor unit may detect whether the sewing machine is operating and provide an operation detection signal to the computing device only when it is operating.

[0014] In exemplary embodiments, the sewing machine operation detection sensor unit may include a vibration sensor that detects vibrations of the sewing machine occurring while the sewing machine is operating and provides the operation detection signal to the computing device.

[0015] In exemplary embodiments, the sewing defect inspection device may further include a buzzer that sounds an alarm to an operator to provide a notification of defect occurrence when a sewing defect is detected, and / or a display unit that outputs defect occurrence information to a screen.

[0016] In exemplary embodiments, the sewing defect data may include sewing defect image data and metadata of a sewing result determined to be defective, defect type label data, identification information of a matched sewing machine, and identification information of a sewing worker.

[0017] In exemplary embodiments, the server computer device may further perform the function of applying user-specific permissions differentially to set different scopes for real-time defect monitoring dashboard search, history search, and data access according to the permissions.

[0018] In exemplary embodiments, the sewing defect inspection device may be configured to automatically transmit identification information of the matched sewing machine to the server computer device when it is initially installed on the matched sewing machine and connected to the server computer device in a communicable manner. In this case, the factory / workline management function performed by the server computer device may include a function to receive identification information transmitted from the plurality of sewing defect inspection devices and to automatically generate, modify, and delete information regarding the factory configuration, workline configuration, and sewing machines assigned to each workline based on the received identification information, and a function to enable a user to manually generate, modify, and delete information regarding the factory configuration, workline configuration, and sewing machines assigned to each workline.

[0019] In exemplary embodiments, when a worker is newly assigned to a sewing machine and inputs their identification information through a sewing defect inspector matched to the sewing machine, the sewing defect inspector may be configured to automatically transmit the input worker's identification information to the server computer device. In this case, the factory / workline management function performed by the server computer device may further include a function that enables the automatic creation, modification, and deletion of additional information regarding workers assigned to each sewing machine on each workline by utilizing the transmitted worker identification information.

[0020] In exemplary embodiments, the factory / workline management function may further include a function to automatically convert configuration information of worklines established for each factory in the current state, sewing machines, sewing defect inspectors, and worker identification information assigned to each workline into a graphic form and display it on a display screen.

[0021] In exemplary embodiments, the graphic of each work line is represented in a form in which a plurality of nodes are connected, and each node corresponds to a sewing machine, a sewing defect inspector, and a worker placed on the corresponding work line, and can be configured so that when a specific node is activated by a user, identification information of the sewing machine, sewing defect inspector, and worker corresponding to that specific node is displayed.

[0022] In exemplary embodiments, the sewing defect detection and defect type classification model may be a model trained to simultaneously detect multiple defect types when multiple defect types exist in a single sewing result image, and to classify the sewing result image according to the detected types of sewing defects.

[0023] Meanwhile, a method for automatic inspection and management of sewing defects according to embodiments for realizing the objectives of the present invention comprises: a step in which a sewing defect inspection and management system, wherein a camera and a sewing defect inspector are installed for each sewing machine installed in a sewing factory, and the sewing defect inspector is connected to a server computer device via the Internet for communication; a step in which a sewing machine operation detection sensor of the sewing defect inspector monitors whether a sewing machine matched with itself is currently operating; a step in which, while the matching sewing machine is detected to be operating through the monitoring of the operation detection sensor, the sewing defect inspector issues a shooting instruction to the matched camera in real time; a step in which, while the shooting instruction is being issued, the camera captures the sewing work result of the sewing machine in real time and outputs a sewing result image; and a step in which the sewing defect inspector collects the sewing result image output from the camera, provides the preprocessed sewing result image as input to a sewing defect detection and defect type classification model of an artificial neural network-based computer program running at the time, thereby determining in real time whether a sewing defect has occurred and classifying the type of the detected sewing defect. The method includes the step of transmitting sewing defect data obtained as a result of the detection and classification of defects by the sewing defect inspection device to a server computer device in real time; the step of receiving, storing, and managing the sewing defect data transmitted from each sewing defect inspection device at the server computer device, and providing a defect occurrence alarm in real time to a front-end terminal of a relevant user; and the step of providing at least one of a service that creates and visually displays statistical data regarding defect status and history information by factory, work line, and worker using the collected sewing defect data, a service that monitors the status of the sewing defect inspection device, a factory / work line management service, and a defect image download service at the server computer device.

[0024] In exemplary embodiments, the monitoring step may include providing a signal to the computing device of the sewing defect inspection machine indicating that the matched sewing machine is in operation while the vibration sensor employed as the sewing machine operation detection sensor detects vibrations generated by the operation of the matched sewing machine.

[0025] In exemplary embodiments, the preprocessing may be a preprocessing in which the sewing defect inspection device resizes and normalizes the sewing result image collected from the camera to fit the input of the sewing defect detection and defect type classification model.

[0026] In exemplary embodiments, the method may further include the step of providing a notification of defect occurrence to an operator in real time by sounding an alarm and / or outputting a warning light when the sewing defect inspection device detects a sewing defect.

[0027] In exemplary embodiments, the sewing defect data may include sewing defect image data and metadata of a sewing result determined to be defective, defect type label data, identification information of a matched sewing machine, and identification information of a sewing worker.

[0028] In exemplary embodiments, the method may further include the step of automatically transmitting identification information of the matched sewing machine to the server computer device, together with identification information of the factory and work line to which the matched sewing machine belongs, when the sewing defect inspection machine is initially installed on the matched sewing machine and is communically connected to the server computer device; and the step of receiving identification information transmitted from the plurality of sewing defect inspection machines at the server computer device and automatically generating, modifying, and deleting information regarding the configuration of the factory, the configuration of the work line, and the sewing machine assigned to each work line based on the received identification information.

[0029] In exemplary embodiments, the method may further include the step of, when a worker is newly assigned to a sewing machine, inputting their identification information through a sewing defect inspector matched to the sewing machine, and the sewing defect inspector automatically transmitting the input worker's identification information to the server computer device; and the step of, at the server computer device, further using the transmitted worker identification information to additionally and automatically generate, modify, and delete information regarding the worker assigned to each sewing machine of each work line.

[0030] In exemplary embodiments, the method may further include the step of automatically converting configuration information of work lines established for each factory in the current state, sewing machines, sewing defect inspectors, and worker identification information assigned to each work line into a graphic form and displaying it on a display screen on the server computer device.

[0031] In exemplary embodiments, the graphics of each work line in the server computer device are represented in a form in which a plurality of nodes are connected, and each node corresponds to a sewing machine, a sewing defect inspector, and a worker placed on the corresponding work line, and may be configured so that when a specific node is activated by a user, identification information of the sewing machine, sewing defect inspector, and worker corresponding to that specific node is displayed. Effects of the invention

[0032] According to embodiments of the present invention, a camera and a sewing defect inspection device are installed for each sewing machine in a sewing factory, and defect inspection is performed only when the matched sewing machine is actually in operation. Therefore, images are not captured when the sewing machine is not in operation, thereby suppressing the generation of image data irrelevant to defect detection and minimizing the generation of image data that is subject to defect detection, and accordingly, the burden of data processing for defect detection can be significantly reduced.

[0033] Defects can be detected in real-time during the current stage of the sewing process, rather than at the final stage. Consequently, by accelerating the detection of defective products, rework time can be minimized and unnecessary losses reduced, and the overall efficiency of the production line can be significantly improved, thereby enhancing productivity.

[0034] Information on sewing defects generated from each sewing machine in various sewing factories can be collected in real-time and integrated on a central server computer to take appropriate follow-up actions. In addition, by visually displaying the collected data on dashboards and history management screens to support an intuitive understanding of the quality status within the process, it enables visualization-based quality monitoring and allows for effective quality management at the sewing factory and sewing line levels.

[0035] In addition, the central server computer device can automatically generate production line configuration information for each sewing factory by utilizing various identification information provided by sewing defect inspectors matched to each sewing machine, and can accurately monitor the status of each sewing factory's production lines in real time by automatically reflecting changes when they occur.

[0036] The present invention distinguishes roles into Admin, Manager, and User, and provides user permission-based customized functions by applying differential scopes of function and data access according to each authority. Brief explanation of the drawing

[0037] Figure 1 conceptually illustrates a conventional sewing defect detection method. FIG. 2 is a block diagram conceptually illustrating the configuration of an automatic sewing defect inspection and management system according to an exemplary embodiment of the present invention. FIG. 3 illustrates a state in which a sewing defect inspection device and a camera are installed on a sewing machine according to an exemplary embodiment of the present invention. FIG. 4 is a block diagram illustrating the specific configuration of a sewing defect inspection machine according to an exemplary embodiment of the present invention. FIG. 5 illustrates representative images of defect types to explain a sewing defect detection and defect type classification model according to an exemplary embodiment of the present invention. FIG. 6 is a flowchart illustrating a procedure for detecting sewing defects and classifying defect types by executing a predetermined computer program in a sewing defect inspection machine according to an exemplary embodiment of the present invention and transmitting the results to a server computer device. FIG. 9 illustrates a start page and a main screen provided in a server computer device according to an exemplary embodiment of the present invention. FIG. 10 illustrates a defect statistics chart by defect type line as one of the detailed functions of a dashboard, which is a main menu provided in a server computer device according to an exemplary embodiment of the present invention. FIG. 11 illustrates a page for managing a list of factories and a factory manager provided in a server computer device according to an exemplary embodiment of the present invention. FIG. 12 illustrates a page for managing a list of work lines and a line manager provided in a server computer device according to an exemplary embodiment of the present invention. FIG. 13 illustrates a page for managing a node list and a node manager that constitute a work line provided in a server computer device according to an exemplary embodiment of the present invention. FIG. 14 illustrates a page for managing a list of defect types provided in a server computer device according to an exemplary embodiment of the present invention. FIG. 15 illustrates a defect history screen provided in a server computer device according to an exemplary embodiment of the present invention. FIG. 16 illustrates a real-time notification page for sewing defects provided on a server computer device according to an exemplary embodiment of the present invention. Specific details for implementing the invention

[0038] Hereinafter, preferred embodiments of the present invention will be described in more detail with reference to the attached drawings. Identical components in the drawings are given the same reference numerals, and redundant descriptions of identical components are omitted.

[0039] With respect to the embodiments of the present invention disclosed in the text, specific structural or functional descriptions are merely illustrative for the purpose of explaining the embodiments of the present invention. Embodiments of the present invention may be implemented in various forms and should not be interpreted as being limited to the embodiments described in the text. That is, the present invention may be subject to various modifications and may take various forms, and specific embodiments are illustrated in the drawings and described in detail in the text. However, this is not intended to limit the present invention to specific disclosed forms, and it should be understood that it includes all modifications, equivalents, and substitutions that fall within the spirit and scope of the present invention.

[0040] The terms used in this invention are used merely to describe specific embodiments and are not intended to limit the invention. Singular expressions include plural expressions unless the context clearly indicates otherwise. In this application, terms such as "comprising" or "having" are intended to indicate the presence of the features, numbers, steps, actions, components, parts, or combinations thereof described in the specification, and should be understood as not precluding the existence or addition of one or more other features, numbers, steps, actions, components, parts, or combinations thereof. Furthermore, terms such as "first," "second," etc., may be used to describe various components, but said components should not be limited by said terms. These terms are used solely for the purpose of distinguishing one component from another.

[0041] FIG. 2 conceptually illustrates the configuration of an automatic sewing defect inspection and management system according to an exemplary embodiment of the present invention.

[0042] Referring to FIG. 2, it is assumed that there is one or more sewing factories (20-1, 20-2, 20-3,...), and that a number of work lines are established in each sewing factory, and a number of sewing machines (30) are installed in each work line to produce clothing. In this environment, the sewing defect automatic inspection and management system (10) is configured to automatically inspect the quality of sewing work performed in the one or more sewing factories (20-1, 20-2, 20-3,...) and to manage the inspection results in an integrated manner. To this end, for the number of sewing machines (Sewing Machine: SM, 30) installed in each of the one or more sewing factories (20-1, 20-2, 20-3,...), a camera (CAM, 90) and a sewing defect detector (Sewing Defect Detector: SDD, 40) may be installed for each sewing machine (30). The camera (90) is connected to communicate with the matched sewing defect inspection device (40), and the entire sewing defect inspection device (40) can be connected to communicate with the server computer device (200) via the Internet (120). Related users' terminals (300), such as personal computers, tablets, or smartphones, can be connected to the server computer device (200) via the Internet (120).

[0043] FIG. 3 illustrates a state in which a sewing defect inspection device and a camera are installed on a sewing machine according to an exemplary embodiment of the present invention.

[0044] Referring to FIG. 3, a camera (90) may be fixedly installed on one side of a sewing machine (30) so as to be able to photograph in real time the sewing work result (35) that is sewn onto the workpiece by the sewing machine (30). The camera (90) installed on each sewing machine (30) may photograph the sewing work result (35) of the matched sewing machine in real time and output a sewing result image only while a shooting instruction is provided from the sewing defect inspection machine (40).

[0045] FIG. 4 shows the specific configuration of a sewing defect inspection device according to an exemplary embodiment of the present invention.

[0046] Referring to FIG. 4, the sewing defect inspection device (40) may include a calculation control unit (50), a vibration sensor unit (70), a communication unit (80), and an alarm unit (75). The calculation control unit (50), the vibration sensor unit (70), and the communication unit (80) may be built inside a case, and the alarm unit (75) may be mounted outside the case. The sewing defect inspection device (40) may further include an input unit (60) and an output unit (65). The case may be fixed to a sewing machine (30) using a fixing bracket (not shown). The fixing bracket may be configured to stably fix the sewing machine and the sewing defect inspection device even under strong vibrations.

[0047] The sewing machine (30) generates vibrations exceeding a threshold value while in operation. The vibration sensor unit (70) can detect vibrations exceeding that threshold value that occur when the matched sewing machine (30) is in operation and provide an operation detection signal to the operation control unit (50). During the time when no operation detection signal is output from the vibration sensor unit (70), the sewing machine (30) can be seen as being in a stopped state. The reason for adopting the vibration sensor unit (70) is that, due to the characteristics of the sewing site, if filming is done even when not in operation, excessive non-sewing line image data (e.g., sewing machine floor, hands during preparation for work, etc.) that is unnecessary for detecting sewing defects is generated, causing an overload on the computer device that detects defects. Furthermore, if a video is filmed with a worker's hand approaching the workpiece when not in operation, there are problems such as errors in detecting sewing defects. To solve this, the vibration sensor unit (70) can be introduced as a means to minimize the generation of unnecessary data. The vibration sensor unit (70) outputs an operation detection signal only while the sewing machine (30) is in operation. The operation control unit (50) controls the camera (90) to perform real-time shooting only while the operation detection signal is input from the vibration sensor unit (70), thereby preventing the generation of unnecessary sewing workpiece image data and improving the efficiency of the system.

[0048] The vibration sensor unit (70) may include a vibration sensor that detects vibrations to capture images only when the sewing machine is in operation, and a variable resistor that can adjust the sensitivity of the vibration sensor that detects vibrations of the sewing machine.

[0049] The vibration sensor unit (70) is presented as a representative example of a means for detecting whether the sewing machine (30) is in operation, but the means capable of serving as such a sewing machine operation detection sensor unit is not limited to a vibration sensor. Since current is supplied to the sewing machine when it is in operation, the operation of the sewing machine can be detected using a current detector that detects the flow of current. As another example, since the sewing machine operates only while the pedal is pressed, the operation of the sewing machine can be detected using a sensor that detects whether the operator is pressing the pedal.

[0050] The communication unit (80) can connect the sewing defect inspection machine (40) to the server computer device (200) so that it can communicate via the Internet (120). The communication unit (80) may be composed of, for example, a communication module that supports wireless Wi-Fi communication, but it is also possible to implement it as a communication means that supports other types of communication.

[0051] The operation control unit (50) can perform various operation controls as described below by executing a predetermined computer program. The predetermined computer program may include a camera manager module, a model manager module, and a communication manager module. The camera manager module may be a program written to provide a shooting instruction to a matched camera (90) only while an operation detection signal of the sewing machine (30) is input from the vibration sensor unit (70) to capture the sewing work result (35), and to perform preprocessing by collecting the sewing result image output from the camera (90), adjusting its size to match the model input, and normalizing it. The model manager module may be a program written to perform data processing by inputting the preprocessed sewing result image into a sewing defect detection and defect type classification model to detect sewing defects and classify the detected sewing defect image into a predefined class. The communication manager module may be a program written to transmit sewing defect data, including the detected sewing defect image data, to a server computer device (200).

[0052] A pre-existing sewing defect detection and defect type classification model may be applied to the aforementioned computer program. The sewing defect detection and defect type classification model may be obtained by performing machine learning or deep learning based on an artificial neural network using a large amount of sewing defect image data and normal sewing image data, which are labeled according to various sewing defect types and are pre-existing, as training data. When an image of a sewing workpiece is provided as input to the model, the model can detect whether the image corresponds to a sewing defect and, if it corresponds to a sewing defect, classify and output the type of defect. The sewing defect detection and defect type classification model may be a model trained to simultaneously detect multiple defect types when multiple defect types exist in a single sewing result image, and to classify the sewing result image according to the detected types of sewing defects.

[0053] The operation control unit (50) may include a processing device that executes a predetermined computer program to perform various controls and necessary operations, data storage and management, etc., as described above, and a data storage unit capable of storing the predetermined computer program and various data. The processing device may be implemented using one or more general-purpose computers or special-purpose computers, such as a central processing unit (CPU), a processor, a controller, an arithmetic logic unit (ALU), a digital signal processor, a microcomputer, a field programmable array (FPA), a programmable logic unit (PLU), a microprocessor, or any other device capable of executing and responding to instructions.

[0054] The input unit (60) can receive identification information of the factory where the sewing defect inspection machine (40) is installed, identification information of the matched sewing machine (30), and identification information of the worker performing sewing work on the matched sewing machine (30) as needed, and transmit it to the operation control unit (50). In addition, the input unit (60) may include an input section capable of installing and updating the aforementioned computer program.

[0055] The output unit (65) can be configured to output information provided by the operation control unit (50) through a simple display screen. For example, it can output detected sewing defect information provided by the operation control unit (50).

[0056] The alarm unit (75) may be composed of a buzzer that sounds a warning sound to the worker to provide a notification of the occurrence of a defect when a sewing defect is detected, and / or a light-emitting unit that outputs a warning light to provide a notification of the occurrence of a defect.

[0057] FIG. 5 illustrates representative images of defect types to explain a sewing defect detection and defect type classification model according to an exemplary embodiment of the present invention.

[0058] Referring to Figure 5, the dataset shows sewing defects classified into 11 types of defects (Bobbin thread pulling up, Broken stitch, Crooked seam, Needle mark, Overlapped stitch, Pleated fabric, Puckering, Skipped stitch, Stain & damage, Thread sagging, and Twisted) in addition to normal defects. Normal and defective sewing images can be prepared as training data by acquiring a large number of images for each type and labeling them according to the defect type. To prepare the training data, 11 types of fabric and thread combinations were used to reflect various sewing environments, and the data can be prepared by taking photos under conditions similar to reality using industrial sewing machines and camera modules that are actually in use. After preparing the training data, a sewing defect detection and defect type classification program written based on artificial neural networks is trained (machine learning or deep learning) to obtain a sewing defect detection and defect type classification model. The performance of the model can be improved by reinforcing the training dataset through monitoring of the sewing work site. The obtained sewing defect detection and defect type classification model can be applied to the aforementioned computer program executed by the operation control unit (50).

[0059] FIG. 6 illustrates a procedure for executing a predetermined computer program in a sewing defect inspection machine according to an exemplary embodiment of the present invention to detect sewing defects and classify defect types, and then transmitting the results to a server computer device.

[0060] The operation control unit (50) executing the above-mentioned predetermined computer program can determine whether the matched sewing machine (30) is in operation based on the operation detection signal provided by the vibration sensor unit (70) (step S10).

[0061] When the operation control unit (50) starts inputting an operation detection signal, it determines that the sewing machine (30) has started operating, and from then on, while the operation detection signal is being input, it can continuously issue a shooting instruction to the camera (90) installed on the matched sewing machine (30) to take pictures of the sewing work result (35) (step S20). When the operation detection signal is not input, the operation control unit (50) can instruct the camera (90) to stop shooting.

[0062] In addition, the operation control unit (50) executing the above-mentioned computer program can collect and store the sewing result image captured by the matched camera (90) in real time, and perform preprocessing to adjust the size and normalize the collected sewing result image so that it is suitable for providing as input to a sewing defect detection and defect type classification model (step S30).

[0063] Then, the operation control unit (50) provides the previously processed image as input to the sewing defect detection and defect type classification model of the running artificial neural network-based computer program to determine in real time whether a sewing defect has occurred and, if a sewing defect is detected, classify the type of the sewing defect (step S40).

[0064] When a sewing defect is detected, the operation control unit (50) can output an alarm in real time through the alarm unit (75) indicating that a sewing defect has been detected (step S50). Through this, the operator of the matched sewing machine can immediately know that the current operation has caused a defect and can quickly take appropriate follow-up measures accordingly.

[0065] In addition, in parallel with outputting an alarm, the operation control unit (50) can perform in real time transmission of sewing defect data obtained as a result of detection and classification of defects to the server computer device (200) when a sewing defect is detected (step S60). Here, the sewing defect data transmitted to the server computer device (200) may basically include sewing defect image data of the sewing result that is judged as defective and metadata (shooting date and time, camera information, etc.). In addition, the transmitted sewing defect data may further include defect type label data, identification information of the matched sewing machine, and identification information of the sewing worker.

[0066] After processing the alarm output and the defect report for the server computer device (200), or if no sewing defect is detected in step S40, the input image is initialized (step S70), and the process can be returned to step S20, where the sewing work result is captured again, and the procedure of steps S30-S60 can be repeated for a new captured image.

[0067] FIG. 7 illustrates the tasks performed by a sewing defect inspection device and a server computer device of an automatic sewing defect inspection and management system according to an exemplary embodiment of the present invention. FIG. 8 illustrates the configuration of a server computer device.

[0068] Referring to FIGS. 7 and 8, the sewing defect inspection device (40), as previously described, detects vibrations occurring in the sewing machine (30) and captures images of the sewing workpiece (35) only while the sewing machine (30) is operating. The captured images are preprocessed and provided as input to a sewing defect detection and defect type classification model to detect sewing defects and classify defect types. When a sewing defect is detected, the sewing defect inspection device (40) displays an alarm or similar, and can transmit the detected sewing defect image and its metadata to a server computer device (200) via the Internet (120). At this time, identification information of the sewing machine (30) that caused the sewing defect and the worker can also be transmitted together.

[0069] The server computer device (200) may be configured to receive, store, and manage sewing defect data transmitted by the sewing defect inspection device (40) of each sewing factory (20-1, 20-2, 20-3), and to provide functions for effectively managing and supervising sewing defects. That is, the server computer device (200) can receive, store, and manage sewing defect data transmitted from each sewing defect inspection device (40) by executing a predetermined server computer program, and provide a defect occurrence alarm in real time to the front-end terminal (300) of the relevant user.

[0070] The server computer device (200) may include a defective image / metadata server (220) and a sewing defect detection application server (240) separated according to their respective roles.

[0071] The defective image / metadata server (220) receives sewing defect images, metadata, and various identification information transmitted by sewing defect inspection devices (40), classifies the received data by category, stores and manages it in a database (230), and can provide data requested by the sewing defect detection application server (240).

[0072] The sewing defect detection application server (240) can utilize collected sewing defect data to create statistical data regarding defect status by factory, work line, and individual, and create and visually display statistical data regarding the list and history information of defects, monitor the status of the sewing defect inspection machine (40), manage factories and work lines, and provide services in response to various requests from users. Specifically, the sewing defect detection application server (240) can be configured to provide services such as providing defect statistics charts by factory and line through a dashboard by running sewing defect inspection machine manager software, creating and providing defect history so that real-time defect lists and records can be checked, managing information regarding the configuration of sewing factories, work lines configured within each sewing factory, and sewing machine devices placed in each work line, and updating changes, downloading sewing defect images (learning) requested by the user, and visually displaying work status by date and by factory in an easy-to-understand manner on a display. Data generated during the process of providing these services can be stored and managed in a separate database (250).

[0073] The images in FIGS. 9 to 16 are images captured of output screens related to these services.

[0074] Figure 9 shows the main screen, which includes a DASHBOARD menu at the top providing defect statistics charts by factory and line, and a DEFECT HISTORY menu providing a real-time list and records of defects. The left side menu box includes a Factory Management menu for managing factories and factory managers, a Line Management menu for managing lines and line users by factory, an Update Device menu for managing updates to sewing defect inspection devices, and a Download Images menu for downloading defect images (learning). The central screen displays the percentage of sewing defects by type that occurred in each factory on a specific day as a pie chart.

[0075] Data on the status of sewing defect detection by production line in a sewing factory can be organized and provided as a graph. When a user selects a desired factory and production line from the management menu on the main screen of Fig. 9, statistical data is generated by classifying sewing defects detected over time in the selected factory and production line by type, and this data can be provided as a graph so that the user can conveniently check it. This allows the status of sewing defect detection to be checked by production line and time period for each factory according to the user's intent, thereby making it easier to identify problems on the production line. Furthermore, the location where defects frequently occur can be accurately and easily identified, and the defect detection status can be used as a basis for resolving problems in the future.

[0076] Figure 10 illustrates the start page and main screen of the main menu, the dashboard. As one of the detailed functions of the main screen, if the user selects a factory and a line, or additionally selects a date, a defect detection statistics chart by time period can be provided for each production line accordingly. The user can check the defect detection status by time period in real time.

[0077] Figure 11 illustrates a page for managing a list of factories and factory managers. On this page, you can register a new factory, enter factory names and factory codes, delete factories, and add / delete / modify factory managers.

[0078] Figure 12 illustrates a page for managing a list of work lines and line managers. On this page, you can register a new work line, enter line names and line codes, delete lines, and add / delete / modify line managers.

[0079] FIG. 13 illustrates a page for managing a list of nodes constituting a work line and a node manager. Here, a node refers to a device (sewing machine and sewing defect inspection machine) registered within the work line. Through this page, new nodes can be registered and the number of nodes managed. With an easy node management system, node management and modification are easy, allowing for flexible adaptation to new processes.

[0080] Figure 14 illustrates a page for managing a list of defect types. This page is intended for registering new defects and assigning defect codes, and the identification and management of defect items can be facilitated through the defect code and name.

[0081] Figure 15 illustrates a defect history screen. This page displays a list of sewing defect data transmitted from a sewing factory. On this page, users can set detailed conditions (date, factory, line, process) to select and quickly retrieve only the necessary data. Defect images are also displayed along with the list; if a user right-clicks on an image with a mouse, the clicked image can pop up as an enlarged sewing defect image. By updating sewing defect data in real time, it enables the rapid identification of defect data for today's date, and the CONFIRM indicator function allows the manager to verify whether defect supervision is being properly performed.

[0082] FIG. 16 illustrates a real-time notification page for sewing defects. This page is designed to notify the relevant work line manager in real time of the sewing defects occurring on the line in charge when sewing defects occur. Sewing defect data (photo) and location information where the defect occurred (line number, process sequence, time of occurrence) are displayed on the screen of the manager terminal (300) along with a list of data that has not yet been confirmed (the leftmost item is the newest). The line manager can then check the defect alarm data and press Confirm or Not Defect to confirm or remove it from the defect list. Through this, the manager can respond immediately when sewing defects occur, thereby contributing to minimizing defects and reducing costs.

[0083] In providing the above information to the user from the server computer device (200), the sewing defect inspection machine manager software may apply user-specific permissions differentially and provide functions to set different real-time defect monitoring dashboard search, history search, and data access scopes according to the permissions. Through these functions, only the necessary information is provided to the relevant parties according to their role-specific permissions, thereby improving work efficiency, and the relevant parties can intuitively check and process the necessary data within their respective scopes of responsibility.

[0084] Meanwhile, the server computer device (200) can obtain more information from sewing defect inspectors (40) to effectively and quickly determine the configuration status of the work lines of the sewing factories (20-1, 20-2, 20-3). For example, when a sewing defect inspector (40) is initially installed on a matched sewing machine (30) and is connected to the server computer device (200) for communication, it can be configured to automatically transmit the identification information of the matched sewing machine (30) to the server computer device (200) along with the identification information of the factory to which the matched sewing machine belongs and the identification information of the work line. By using this information, it is possible to determine which sewing machine (30) each sewing defect inspector (40) is matched with, and which work line of which factory the matched sewing defect inspector (40) and the sewing machine (30) belong to. Accordingly, the server computer device (200) can automatically generate information regarding the configuration of the factory, the configuration of the work line, and the sewing machines placed in each work line based on the identification information transmitted from the sewing defect inspection machine (40).

[0085] In addition, if the sewing machine matched to the existing sewing defect inspection machine (40) changes or the work line to which it belongs changes, the identification information of the changed sewing machine and / or the work line to which it belongs can be transmitted to the server computer device (200). Accordingly, the server computer device (200) can identify the configuration of the work lines currently established in the sewing factory and the arrangement status of the sewing machines in real time. In this way, the server computer device (200) can automatically identify in real time which work lines are established for each sewing factory and which sewing machines (30) and sewing defect inspection machines (40) are placed in each work line through the identification information provided by the sewing defect inspection machine (40).

[0086] Furthermore, when a worker is newly assigned to a sewing machine (30), they may be configured to input their identification information through a sewing defect inspection device (40) matched to the sewing machine (30), and the sewing defect inspection device (40) may be configured to automatically transmit the input worker's identification information to a server computer device (200). By further utilizing the worker identification information provided in this way, the server computer device (200) can determine which worker is assigned to each sewing machine (30) of each work line. Based on this information, the server computer device (200) can automatically determine in real time which work lines are established for each factory and which sewing machine and which worker are assigned to each work line.

[0087] Of course, the server computer device (200) may also allow the user to manually input the configuration of the factory, the configuration of the work line, identification information of the sewing machine placed on each work line and the sewing defect inspection machine (40) matched thereto, and / or identification information regarding the worker.

[0088] As such, in the server computer device (200), the factory / workline management function provided by the sewing defect inspector manager software can automatically configure worklines built for each sewing factory by utilizing identification information of the factory, workline, sewing machine, sewing defect inspector, and worker sent by the sewing defect inspector (40), and can also include a function to automatically match the sewing machine, sewing defect inspector, and worker assigned to each workline. That is, the information displayed on the pages illustrated in FIGS. 12 to 14 can be automatically configured and associated with each other.

[0089] Furthermore, the factory / workline management function provided by the sewing defect inspector manager software can automatically convert the configuration and matching information of these factory worklines—specifically, the configuration information of worklines established for each factory in their current state, and the identification information of sewing machines, sewing defect inspectors, and workers assigned to each workline—into a graphical format rather than a table format, allowing for visual verification on the display screen. Through such output screens, the manager can identify the worklines established for each factory, as well as the sewing machines and worker information assigned to each line, in real time. The graphic of each workline can be represented in a form where multiple nodes are connected. Each node corresponds to the sewing machine, sewing defect inspector, and worker assigned to that workline; the system can be configured so that when a specific node is activated by a user, the identification information of the sewing machine, sewing defect inspector, and worker corresponding to that specific node is displayed. Industrial applicability

[0090] The present invention can be used for quality control of sewing operations in a sewing factory.

[0091] Although the embodiments have been described above with reference to the limited drawings, those skilled in the art will understand that various modifications and changes can be made to the invention without departing from the spirit and scope of the invention as described in the following claims. For example, suitable results may be achieved even if the described techniques are performed in a different order than described, and / or the components of the described system, structure, device, circuit, etc. are combined or assembled in a form different from the described method, or are replaced or substituted by other components or equivalents. Therefore, other implementations, other embodiments, and equivalents to the claims are also within the scope of the claims set forth below. Explanation of the symbols

[0092] 10: Automatic Sewing Defect Inspection and Management System 30: Sewing machine 35: Sewing result 40: Sewing defect inspection machine 50: Operational control unit 70: Vibration sensor unit 75: Alarm unit 90: Camera 200: Server computer device

Claims

Claim 1 A plurality of cameras installed on each sewing machine in a sewing factory, each of which captures the sewing work result of the matched sewing machine in real time and outputs a sewing result video only when a shooting instruction is provided; a plurality of sewing defect inspection devices, each of which is installed on each sewing machine in the sewing factory, each comprising a computing device that executes an artificial neural network-based computer program, and a sewing machine operation detection sensor unit that detects in real time whether the matched sewing machine is operating by detecting vibration or current flow occurring only while the matched sewing machine is operating, and outputs an operation detection signal to the computing device only when it is operating;And the computing device is configured to perform in real time the following: providing a shooting instruction to the camera of a matched sewing machine only while the operation detection signal is input to collect an image of the sewing result captured by the matched camera; providing the collected image of the sewing result as input to a sewing defect detection and defect type classification model of the running artificial neural network-based computer program to determine in real time whether a sewing defect has occurred and to classify the type of the detected sewing defect; and transmitting sewing defect data obtained as a result of the detection and classification to a server computer device; and is connected to communicate with the plurality of sewing defect inspection machines via the Internet, and by executing a predetermined server computer program, performs the functions of receiving, storing, and managing the sewing defect data transmitted from each sewing defect inspection machine; providing a defect occurrence alarm in real time to the front-end terminal of a relevant user; utilizing the collected sewing defect data to create and visually display statistical data regarding the defect status and history information by sewing factory, by work line where multiple sewing machines are deployed, and by the worker in charge of each sewing machine; a sewing defect inspection machine status monitoring function; a factory and work line management function; and a defect image download function. An automatic sewing defect inspection and management system characterized by having a server computer device configured to perform at least one of the functions. Claim 2 The automatic sewing defect inspection and management system according to claim 1, wherein the artificial neural network-based computer program comprises: a camera manager module that provides the shooting instruction to a matched camera only while an operation detection signal is input to capture the sewing work result, and performs preprocessing by collecting the sewing result image output from the camera, adjusting its size to match the model input, and normalizing it; a model manager module that performs data processing by inputting the preprocessed image into the sewing defect detection and defect type classification model to detect sewing defects and classify the detected sewing defect images into a predefined class; and a communication manager module that transmits sewing defect data including the detected sewing defect image data to the server computer device. Claim 3 The sewing machine operation detection sensor unit according to claim 1 is characterized by including a vibration sensor that detects vibrations of the sewing machine occurring while the sewing machine is operating and provides the operation detection signal to the computing device, or a current sensor that detects the current flow supplied to the sewing machine when the sewing machine is operating. Claim 4 The sewing defect inspection and management system according to claim 1, further comprising a buzzer that sounds a warning sound to an operator to provide a notification of defect occurrence when a sewing defect is detected, and / or a display unit that outputs defect occurrence information to a screen. Claim 5 The sewing defect automatic inspection and management system according to claim 1, wherein the sewing defect data comprises sewing defect image data of a sewing result determined to be defective, metadata including at least shooting date and time information, defect type label data, identification information of a matched sewing machine, and identification information of a sewing worker. Claim 6 The sewing defect automatic inspection and management system according to claim 1, wherein the server computer device further performs the function of differentially applying user-specific permissions and setting different real-time defect monitoring dashboard search, history search, and data access ranges according to the permissions. Claim 7 In claim 1, the sewing defect inspection device is configured to automatically transmit identification information of the matched sewing machine to the server computer device when it is initially installed on the matched sewing machine and connected to the server computer device in a communicable manner, together with identification information of the factory and work line to which the matched sewing machine belongs, and the factory and work line management function performed on the server computer device includes a function of receiving identification information transmitted from the plurality of sewing defect inspection devices and automatically generating, modifying, and deleting information regarding the factory configuration, work line configuration, and sewing machines assigned to each work line based on the received identification information, and a function that enables a user to manually generate, modify, and delete information regarding the factory configuration, work line configuration, and sewing machines assigned to each work line. Claim 8 A sewing defect automatic inspection and management system according to claim 7, wherein when a worker is newly assigned to a sewing machine, the sewing defect inspector is configured to automatically transmit the worker's identification information through a sewing defect inspector matched to the sewing machine, and the factory and production line management function performed by the server computer device further includes a function that enables additional automatic generation, modification, and deletion of information regarding workers assigned to each sewing machine of each production line by further utilizing the transmitted worker identification information. Claim 9 A sewing defect automatic inspection and management system according to claim 7 or 8, characterized in that the factory and work line management function further includes a function to automatically convert configuration information of work lines established for each factory in the current state, sewing machines, sewing defect inspectors, and worker identification information assigned to each work line into a graphic form and display it on a display screen. Claim 10 A sewing defect automatic inspection and management system according to claim 9, wherein the graphic of each work line is represented in a form in which multiple nodes are connected, and each node corresponds to a sewing machine, sewing defect inspector, and worker placed on the corresponding work line, and is configured such that when a specific node is activated by a user, identification information of the sewing machine, sewing defect inspector, and worker corresponding to that specific node is displayed. Claim 11 The sewing defect automatic inspection and management system according to claim 1, characterized in that the sewing defect detection and defect type classification model is a model trained to simultaneously detect multiple defect types when multiple defect types exist in a single sewing result image and to classify the sewing result image according to the detected types of sewing defects. Claim 12 A method for automatic inspection and management of sewing defects performed using an automatic inspection and management system for sewing defects, wherein a camera and a sewing defect inspector are installed on each sewing machine in a sewing factory, and the sewing defect inspector is connected to a server computer device via the Internet for communication. The method comprises: a sewing machine operation detection sensor unit of the sewing defect inspector detecting in real time whether a matched sewing machine is operating by detecting vibrations or current flows that occur only while the matching sewing machine is operating, and only when it is operating A step of monitoring whether the matched sewing machine is currently operating by providing an operation detection signal indicating that the matched sewing machine is operating to the computing device of the sewing defect inspection machine; a step of the sewing defect inspection machine issuing a real-time shooting instruction to the matched camera while the matched sewing machine is detected to be operating through monitoring of the operation detection sensor unit; a step of the camera shooting the sewing work result of the sewing machine in real-time and outputting a sewing result image while the shooting instruction is being issued; a step of the sewing defect inspection machine collecting the sewing result image output from the camera and providing the preprocessed sewing result image as input to a sewing defect detection and defect type classification model of an artificial neural network-based computer program running in real-time to determine whether a sewing defect has occurred and classify the type of the detected sewing defect; a step of the sewing defect inspection machine transmitting sewing defect data obtained as a result of defect detection and classification to a server computer device in real-time. A method for automatic inspection and management of sewing defects, characterized by comprising: a step of receiving, storing, and managing sewing defect data transmitted from each sewing defect inspection machine at a server computer device, and providing a defect occurrence alarm in real time to a front-end terminal of a relevant user; and a step of providing at least one of a service that utilizes the collected sewing defect data to create and visually display statistical data regarding defect status and history information by sewing factory, a work line where multiple sewing machines are arranged, and a worker in charge of each sewing machine, a service for monitoring the status of sewing defect inspection machines, a factory and work line management service, and a defect image download service. Claim 13 A method for automatic inspection and management of sewing defects according to claim 12, wherein the sewing machine operation detection sensor unit used in the monitoring step comprises a vibration sensor that detects vibrations of the sewing machine occurring while the sewing machine is operating and provides the operation detection signal to the computing device, or a current sensor that detects the current flow supplied to the sewing machine when the sewing machine is operating. Claim 14 A method for automatic inspection and management of sewing defects according to claim 12, wherein the above preprocessing is a preprocessing in which the sewing defect inspection device resizes and normalizes the sewing result images collected from the camera to fit the input of the sewing defect detection and defect type classification model. Claim 15 A method for automatic inspection and management of sewing defects according to claim 12, further comprising the step of providing a notification of defect occurrence to a worker in real time by outputting a warning sound and / or a warning light when the sewing defect inspection device detects a sewing defect. Claim 16 A method for automatic inspection and management of sewing defects according to claim 12, characterized in that the sewing defect data includes sewing defect image data of a sewing result determined to be defective, metadata including at least shooting date and time information, defect type label data, identification information of a matched sewing machine, and identification information of a sewing worker. Claim 17 A method for automatic inspection and management of sewing defects according to claim 12, further comprising: a step of automatically transmitting identification information of the matched sewing machine to the server computer device together with identification information of the factory and work line to which the matched sewing machine belongs when the sewing defect inspection machine is initially installed on the matched sewing machine and connected to the server computer device in a communicable manner; and a step of receiving identification information transmitted from each sewing defect inspection machine at the server computer device and automatically generating, modifying, and deleting information regarding the configuration of the factory, the configuration of the work line, and the sewing machine assigned to each work line based on the received identification information. Claim 18 A method for automatic inspection and management of sewing defects according to claim 17, wherein when a worker is newly assigned to a sewing machine, the sewing defect inspector inputs their identification information through a sewing defect inspector matched to the sewing machine, and the sewing defect inspector automatically transmits the input worker's identification information to the server computer device; and further comprising the step of, in the server computer device, additionally automatically generating, modifying, and deleting information regarding the worker assigned to each sewing machine of each work line by further utilizing the transmitted worker identification information. Claim 19 A method for automatic inspection and management of sewing defects according to claim 18, further comprising the step of automatically converting configuration information of work lines established for each factory in the current state, sewing machines, sewing defect inspectors, and worker identification information placed in each work line into a graphic form and displaying it on a display screen in the server computer device. Claim 20 A method for automatic inspection and management of sewing defects according to claim 19, wherein, in the server computer device, the graphics of each work line are represented in a form in which a plurality of nodes are connected, and each node corresponds to a sewing machine, a sewing defect inspector, and a worker placed on the corresponding work line, and is configured such that when a specific node is activated by a user, identification information of the sewing machine, sewing defect inspector, and worker corresponding to that specific node is displayed.

Images