KR-102960956-B1 - ARTIFICIAL INTELLIGENCE BASED VISION INSPECTION METHOD, APPARATUS AND SYSTEM FOR DETECTING DEFECTS IN QUILTED PATTERN

KR102960956B1KR 102960956 B1KR102960956 B1KR 102960956B1KR-102960956-B1

Abstract

One aspect of the present invention discloses an artificial intelligence-based vision inspection method, apparatus, and system for detecting defects in a quilted pattern. The method comprises the steps of: receiving an image of a quilted pattern (the image of the quilted pattern is an image of a fabric being quilted captured in real time using an artificial intelligence edge camera); making a first determination of a defect in the quilted pattern using a defect detection model (the defect detection model is an artificial intelligence-based model trained to detect defects in the quilted pattern); making a second determination of the type of defect in the quilted pattern using the defect detection model based on the result of the first determination; and displaying and outputting the result of the first determination and the result of the second determination on the image.

Inventors

조승원
이진원

Assignees

주식회사 이커버스

Dates

Publication Date: 20260507
Application Date: 20240625

Claims (19)

In an artificial intelligence-based vision inspection method for detecting defects in a quilted pattern by a vision inspection device, A step of receiving an image of a quilting pattern, wherein the image of the quilting pattern is an image of a fabric being quilted captured in real time using an AI Edge Camera; A step of primarily determining defects in the quilting pattern using a defect detection model, wherein the defect detection model is an artificial intelligence-based model trained to detect defects in the quilting pattern; A step of secondarily determining the type of defect for the quilting pattern using the defect detection model based on the first determination result; and An artificial intelligence-based vision inspection method for detecting defects in a quilting pattern, comprising the step of displaying and outputting the result of the first judgment and the result of the second judgment on the image.
In claim 1, the first determining step is, A step of determining the type of the fabric using the defect detection model above; A step of identifying the type of the above quilting pattern and determining whether the above quilting pattern is defective by comparing the above quilting pattern with a reference value of a specific pattern corresponding to the quilting pattern; and An artificial intelligence-based vision inspection method for detecting defects in a quilting pattern, comprising the step of determining the location of defect occurrence within the quilting pattern.
In claim 1, the first determining step is, An AI-based vision inspection method for detecting defects in a quilting pattern, further comprising the step of determining whether there is a defect for each fabric in response to identifying a plurality of fabrics within the above image.
In claim 1, the second determining step is, An artificial intelligence-based vision inspection method for detecting defects in a quilting pattern, comprising a step of secondarily determining the type of defect for the quilting pattern based on the determination that the quilting pattern is defective according to the first determination above.
In claim 1, the second determining step is, A step of determining whether the above quilting pattern can be corrected; and An artificial intelligence-based vision inspection method for detecting defects in a quilting pattern, comprising the step of determining whether to discard the fabric in response to a judgment result.
In claim 5, the step of determining whether to discard the fabric is, An artificial intelligence-based vision inspection method for detecting defects in a quilting pattern, comprising the step of outputting information related to whether the fabric can be modified in response to a determination that the fabric is not to be discarded.
In Article 1, A step of training a first artificial intelligence model trained based on supervised learning, a second artificial intelligence model trained based on unsupervised learning, and a third artificial intelligence model trained based on reinforcement learning, respectively, prior to receiving an image of the above-mentioned quilting pattern; A step of comparing the performance (mAP) of the first artificial intelligence model, the second artificial intelligence model, and the third artificial intelligence model; and An AI-based vision inspection method for detecting defects in a quilting pattern, further comprising the step of determining one of the first AI model, the second AI model, and the third AI model as the defect detection model based on a comparison result.
In claim 7, each of the above learning steps is, A step of training the first artificial intelligence model using labeled training data; A step of training the second artificial intelligence model using unlabeled training data; and An AI-based vision inspection method for detecting defects in a quilting pattern, comprising the step of training the third AI model using a reinforcement learning agent.
In Article 7, The method further includes a step of constructing training data using an image generation model prior to each of the above training, wherein An AI-based vision inspection method for detecting defects in a quilting pattern, comprising the step of constructing training data based on augmenting a plurality of new images based on at least two defect pattern images previously acquired from the AI edge camera.
In Article 9, An artificial intelligence-based vision inspection method for detecting defects in a quilting pattern, wherein the image generation model constructs the training data by repeating the operation of augmenting a third image a certain number of times or more based on calculating the difference between a first image and a second image among at least two defect pattern images.
In Article 8, An AI-based vision inspection method for detecting defects in quilting patterns, wherein the above-mentioned labeled training data includes a final data label in which multiple different labels are superimposed for each image.
In Article 11, The above-mentioned final data labeling is an AI-based vision inspection method for detecting defects in quilting patterns, comprising defect pattern detection area labeling, multi-class labeling, and shape labeling.
In Article 12, The above defect pattern detection area labeling includes a bounding box for identifying the location and size of the fabric within the image, and The above multi-class labeling includes text for classifying patterns within an image, and The above-described shape labeling is an artificial intelligence-based vision inspection method for detecting defects in a quilting pattern, comprising dots or lines for identifying the shape of a defect within an image.
In Article 1, An AI-based vision inspection method for detecting defects in a quilting pattern, further comprising a step of preprocessing the image prior to performing the first judgment above.
In claim 14, the preprocessing step is, A step of extracting a plurality of quilting lines within the image based on applying at least one filter; and An artificial intelligence-based vision inspection method for detecting defects in a quilting pattern, comprising the step of increasing the resolution of the quilting line.
In claim 1, the outputting step is, An artificial intelligence-based vision inspection method for detecting defects in a quilting pattern, comprising the step of visualizing the result of the first judgment and the result of the second judgment within the image and outputting them through a monitoring web or a manufacturing execution system (MES) for sewing products.
In an artificial intelligence-based vision inspection device for detecting defects in quilting patterns, A communication module receiving an image of a quilting pattern, wherein the image of the quilting pattern is an image of a fabric being quilted, captured in real time using an artificial intelligence edge camera; A processor that first determines a defect in the quilting pattern using a defect detection model, and secondarily determines the type of defect in the quilting pattern using the defect detection model based on the result of the first determination, wherein the defect detection model is an artificial intelligence-based model trained to detect defects in the quilting pattern; and An artificial intelligence-based vision inspection device for detecting defects in a quilting pattern, comprising a display that outputs the result of the first judgment and the result of the second judgment on the image based on the control of the processor.
In an artificial intelligence-based vision inspection system for detecting defects in quilting patterns, An AI edge camera transmitting an image of a quilting pattern, wherein the image of the quilting pattern is an image of the fabric being quilted captured in real time; An artificial intelligence-based vision inspection system for detecting defects in a quilting pattern, comprising a server that first determines a defect in the quilting pattern using a defect detection model, secondarily determines the type of defect in the quilting pattern using the defect detection model based on the result of the first determination, and outputs the result of the first determination and the result of the second determination by displaying them on the image.
In Article 18, The above server is an AI-based vision inspection system for detecting defects in quilting patterns, which provides a user interface for visualizing the result of the first judgment and the result of the second judgment within the above image.

Description

Artificial Intelligence-Based Vision Inspection Method, Apparatus, and System for Detecting Defects in Quilted Patterns The present invention relates to an artificial intelligence-based vision inspection method, apparatus, and system for detecting defects in quilting patterns, and more specifically, to a method, apparatus, and system for detecting defects in quilting patterns inserted into fabrics of multi-variety sewn products using an artificial intelligence edge camera and an artificial intelligence model. Sewing products are manufactured by placing filling material, such as cotton or feathers, between two layers of fabric and sewing them into a desired shape. They are mainly used in the production of products such as blankets, pillows, and dolls, and various types of sewing products are mass-produced through automated production lines. Unique quilting patterns reflecting the manufacturer's intent or customer needs can be incorporated into the surface of these sewn products. For example, if the product is a quilt, a grid pattern may be applied. Similar to product manufacturing, quilting patterns are automatically inserted into the product using sewing machines. However, unexpected variables such as wrinkled fabric or broken needles during the quilting process can result in defective patterns; to address this, camera-based inspection processes, such as vision inspection systems, are additionally implemented. Recently, AI model-based vision inspection techniques have been utilized to perform more accurate inspections. However, there has been a problem with a low defect detection rate. Therefore, to address this, vision inspection techniques that improve defect detection accuracy are required. FIG. 1 is a schematic diagram showing an artificial intelligence-based vision inspection system for detecting defects in a quilting pattern according to one embodiment of the present invention. Figure 2 is a diagram showing the implementation form of an artificial intelligence-based vision inspection system that detects defects in a quilting pattern. Figure 3 is a diagram showing the definition of defects for a quilting pattern image. Figure 4 is a diagram showing the definition of defect types for a quilting pattern image. Figure 5 is a diagram showing a screen of a production management system. FIG. 6 is a flowchart of an artificial intelligence-based vision inspection method for detecting defects in a quilting pattern according to an embodiment of the present invention. Figure 7 is a flowchart of an unsupervised learning method for an artificial intelligence model. Figure 8 is a flowchart of the reinforcement learning method for an artificial intelligence model. FIG. 9 is a block diagram of an artificial intelligence-based vision inspection device for detecting defects in a quilting pattern according to one embodiment of the present invention. The present invention is capable of various modifications and may have various embodiments, and specific embodiments are illustrated in the drawings and described in detail. However, this is not intended to limit the invention to specific embodiments, and it should be understood that it includes all modifications, equivalents, and substitutions that fall within the spirit and scope of the invention. Terms such as "first," "second," etc., may be used to describe various components, but said components should not be limited by said terms. Such terms are used solely for the purpose of distinguishing one component from another. For example, without departing from the scope of the present invention, the first component may be named the second component, and similarly, the second component may be named the first component. The term "and/or" includes a combination of a plurality of related described items or any of a plurality of related described items. When it is stated that one component is "connected" or "connected" to another component, it should be understood that while it may be directly connected or connected to that other component, there may also be other components in between. On the other hand, when it is stated that one component is "directly connected" or "directly connected" to another component, it should be understood that there are no other components in between. The terms used in this application are used merely to describe specific embodiments and are not intended to limit the invention. The singular expression includes the plural expression unless the context clearly indicates otherwise. In this application, terms such as "comprising" or "having" are intended to specify 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. Unless otherwise defined, all terms used herein, including technical or scientific terms, have the s