CN111077166A - Liquid crystal screen defect detection method, device and terminal equipment - Google Patents

Abstract

The invention provides a flaw detection method, a flaw detection device and terminal equipment of a liquid crystal display, and relates to the technical field of detection, wherein the method comprises the following steps: acquiring a liquid crystal screen image to be detected; inputting a liquid crystal screen image into a defect detection model obtained through pre-training so that the defect detection model can carry out defect detection on the liquid crystal screen image; and acquiring a flaw detection result output by the flaw detection model. The method and the device can detect the defects of the liquid crystal screen in an artificial intelligence mode, obtain the detection result comprising the defect category of the liquid crystal screen, better save the labor cost and improve the detection efficiency of the defects of the liquid crystal screen.

Description

Liquid crystal screen defect detection method, device and terminal equipment

technical field

The invention relates to the technical field of detection, and in particular, to a method, device and terminal equipment for defect detection of a liquid crystal screen.

Background technique

At present, LCD screen electronic products have become an indispensable and important element in our life and work. As an important component of electronic products, the quality of LCD screen is directly related to the performance and appearance of electronic products. In the production process of LCD screens, the current level of technology cannot completely avoid various defects (such as scratches, bubbles, dark/bright spots, etc.). These defects of LCD screens will directly affect the use of electronic products and even cause Electronic products cannot be used normally, resulting in a low user experience. However, most of the existing technologies require professionals to judge the defects of the LCD screen, which not only has high labor cost, but also has low detection efficiency.

SUMMARY OF THE INVENTION

In view of this, the purpose of the present invention is to provide a liquid crystal screen defect detection method, device and terminal equipment, which can better save labor costs and improve the detection efficiency of liquid crystal screen defects.

In order to achieve the above purpose, the technical solutions adopted in the embodiments of the present invention are as follows:

In a first aspect, an embodiment of the present invention provides a method for detecting defects in a liquid crystal screen, the method comprising: acquiring an image of a liquid crystal screen to be detected; inputting the image of the liquid crystal screen into a pre-trained defect detection model, so that the defect detection model Perform defect detection on the LCD screen image; obtain the defect detection result output by the defect detection model; wherein, the defect detection result includes the LCD screen defect category.

With reference to the first aspect, the embodiment of the present invention provides the first possible implementation manner of the first aspect, wherein the step of acquiring the image of the liquid crystal screen to be detected includes: acquiring the image of the liquid crystal screen to be detected by an image acquisition device; wherein , the image acquisition device is set up at a designated position on the LCD screen production line; the LCD screen image to be detected is the image of the LCD screen to be detected collected by the image acquisition device; and/or, the LCD screen image to be detected is obtained through the AOI detection device; wherein , the AOI detection equipment is set up at a designated position on the production line of the LCD screen, and the image of the LCD screen to be detected is the image of the abnormal LCD screen detected by the AOI detection equipment.

In conjunction with the first aspect, the embodiment of the present invention provides a second possible implementation manner of the first aspect, wherein the training process of the defect detection model includes: acquiring multiple sets of training images; The defect category labels carried by the training images in different groups are different; the defect category labels of the training images within the group are the same, but the defect styles are different; multiple groups of training images are input into the preset neural network structure, and the loss function value of the neural network structure is calculated; based on For the loss function value, the network parameters of the neural network structure are trained through the back-propagation algorithm, until the loss function value converges to the preset value, and the training is stopped.

In conjunction with the first aspect, the embodiment of the present invention provides a third possible implementation manner of the first aspect, wherein the defect detection model is a deep convolutional neural network; and the steps of the defect detection model to detect defects on the liquid crystal screen image include: Semantic segmentation is performed on the LCD screen image to obtain a segmentation result; based on the segmentation result, the LCD screen image is detected with flaws.

With reference to the third possible implementation manner of the first aspect, the embodiment of the present invention provides the fourth possible implementation manner of the first aspect, wherein the step of performing semantic segmentation processing on the liquid crystal screen image includes: using hole convolution The LCD screen image is semantically segmented by the method; alternatively, the LCD screen image is semantically segmented by the hole space pyramid pooling method; or the LCD screen image is semantically segmented by the multi-scale convolution cascade method with holes.

In conjunction with any implementation of the first aspect, the embodiment of the present invention provides a fifth possible implementation of the first aspect, wherein the liquid crystal screen defect category includes appearance defects and display defects; appearance defects include scratch defects and dents One or more of blemishes, flash blemishes, bubble blemishes, and foreign body blemishes; display blemishes include one or more of point blemishes, line blemishes, and face blemishes.

In a second aspect, an embodiment of the present invention also provides a defect detection device for a liquid crystal screen, including: an image acquisition module for acquiring an image of the liquid crystal screen to be detected; a defect detection module for inputting the liquid crystal screen image into a pre-trained image The defect detection model is used to make the defect detection model perform defect detection on the LCD screen image; the result acquisition module is used to obtain the defect detection result output by the defect detection model; wherein, the defect detection result includes the LCD screen defect category.

In conjunction with the second aspect, an embodiment of the present invention provides a first possible implementation manner of the second aspect, wherein the image acquisition module is configured to: acquire an image of a liquid crystal screen to be detected through an image acquisition device; wherein the image acquisition device is set up in The designated position on the LCD screen production line; the LCD screen image to be detected is the image of the LCD screen to be detected collected by the image acquisition device; and/or, the LCD screen image to be detected is obtained through the AOI detection device; wherein, the AOI detection device is established in At the designated position on the production line of the LCD screen, the image of the LCD screen to be detected is the image of the abnormal LCD screen detected by the AOI detection equipment.

In conjunction with the second aspect, the embodiment of the present invention provides a second possible implementation manner of the second aspect, wherein the defect detection module is further configured to: acquire multiple groups of training images; wherein each group includes multiple training images, and between groups The training images carry different defect category labels; the training images in the group have the same defect category labels and different defect styles; input multiple groups of training images into the preset neural network structure, and calculate the loss function value of the neural network structure; based on the loss function value, the network parameters of the neural network structure are trained through the back-propagation algorithm, and the training is stopped until the loss function value converges to the preset value.

In conjunction with the second aspect, the embodiment of the present invention provides a third possible implementation manner of the second aspect, wherein the defect detection model is a deep convolutional neural network; Semantic segmentation processing, to obtain segmentation results; defect detection unit, used to detect defects on the LCD screen image based on the segmentation results.

With reference to the third possible implementation manner of the second aspect, the embodiment of the present invention provides the fourth possible implementation manner of the second aspect, and the segmentation unit is used for: using a hole convolution method to perform semantic segmentation processing on the LCD screen image; Alternatively, the liquid crystal screen image is semantically segmented by using the hole space pyramid pooling method; or, the liquid crystal screen image is semantically segmented by using a multi-scale convolution cascade method with holes.

In conjunction with any implementation of the second aspect, the embodiment of the present invention provides a fifth possible implementation of the second aspect, wherein the liquid crystal screen defect category includes appearance defects and display defects; appearance defects include scratch defects and dents One or more of blemishes, flash blemishes, bubble blemishes, and foreign body blemishes; display blemishes include one or more of point blemishes, line blemishes, and face blemishes.

In a third aspect, an embodiment of the present invention provides a terminal device, where the terminal device includes a memory and a processor, where the memory is used to store and support the processor to perform any one of the first to eighth possible implementation manners of the first aspect In the program of the method, the processor is configured to execute the program stored in the memory.

In a fourth aspect, an embodiment of the present invention provides a computer storage medium for storing computer software instructions used in the method of any one of the foregoing first aspect to the eighth possible implementation manner of the first aspect.

Embodiments of the present invention provide a method, device, and terminal device for detecting defects in a liquid crystal screen. By acquiring an image of the liquid crystal screen to be detected, and inputting the image of the liquid crystal screen into a pre-trained defect detection model, the defect detection model can be used for the detection of defects. The LCD screen image is used for defect detection, and then the defect detection result output by the defect detection model is obtained. In the embodiment of the present invention, a defect detection model is used to detect the defects of the liquid crystal screen, and the detection results including the categories of the defects of the liquid crystal screen are obtained, which saves the labor cost and improves the detection efficiency of the defects of the liquid crystal screen.

Additional features and advantages of the present disclosure will be set forth in the description that follows, or some may be inferred or unambiguously determined from the description, or may be learned by practicing the above-described techniques of the present disclosure.

In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, preferred embodiments are given below, and are described in detail as follows in conjunction with the accompanying drawings.

Description of drawings

In order to illustrate the specific embodiments of the present invention or the technical solutions in the prior art more clearly, the following briefly introduces the accompanying drawings that need to be used in the description of the specific embodiments or the prior art. Obviously, the accompanying drawings in the following description The drawings are some embodiments of the present invention. For those of ordinary skill in the art, other drawings can also be obtained based on these drawings without creative efforts.

FIG. 1 shows a flowchart of a method for detecting defects in a liquid crystal screen provided by an embodiment of the present invention;

2 shows a flowchart of a training process of a flaw detection model provided by an embodiment of the present invention;

FIG. 3 shows a flowchart of another liquid crystal screen defect detection method provided by an embodiment of the present invention;

FIG. 4 shows a structural block diagram of a defect detection device for a liquid crystal screen provided by an embodiment of the present invention;

FIG. 5 shows a schematic structural diagram of a terminal device provided by an embodiment of the present invention.

Detailed ways

In order to make the purposes, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are part of the embodiments of the present invention, but not all of them. example. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

Considering that the liquid crystal screen defect detection method in the prior art is not accurate for the liquid crystal screen defect detection, the embodiments of the present invention provide a liquid crystal screen defect detection method, device and terminal equipment, which can more accurately detect the liquid crystal screen. The flaws are detected, and the embodiments of the present invention are described in detail below.

Referring to the flowchart of a method for detecting flaws in a liquid crystal screen shown in FIG. 1 , the method can be executed by terminal devices such as mobile phones and computers, and the method includes the following steps:

Step S102, acquiring an image of the liquid crystal screen to be detected. The liquid crystal screen image to be detected may be an image after the liquid crystal screen is powered on and lit, or an image of the appearance of the screen when the liquid crystal screen is in a black screen state, which is not limited herein. For the image after the LCD screen is powered on, it can be used to detect the display defects of the LCD screen, such as point defects such as bright spots/dark spots, or line defects such as bright lines/dark lines, or surface defects in small areas. The screen appearance image when the LCD screen is in a black state can be used to detect the appearance defects of the LCD screen, such as scratch defects, bubble defects, foreign matter defects, etc. Specifically, the LCD screen to be tested can be flexibly set according to the detection needs. Bright state or black screen state, and obtain its corresponding image accordingly.

In one embodiment, the image of the LCD screen to be detected is acquired by an image acquisition device such as a camera, a scanner, etc. The image acquisition device is set up at a designated position on the LCD screen production line, and the LCD screen image to be detected is captured by the image acquisition device The image of the LCD screen to be tested.

In another embodiment, the image of the liquid crystal screen to be inspected is obtained by an AOI (Automatic Optical Inspection) detection device, the AOI inspection device is set up at a designated position on the production line of the liquid crystal screen, and the image of the liquid crystal screen to be inspected is The image of the abnormal LCD screen detected by the AOI detection equipment, the abnormal LCD screen is the defective LCD screen. During automatic detection, the AOI detection equipment automatically scans the LCD screen through a high-definition CCD camera, collects the image of the LCD screen, and can initially identify the abnormal LCD screen.

One way to obtain the image of the LCD screen to be detected is that a detection device such as an AOI detection device or an image acquisition device directly sends the collected LCD screen image to be detected to the terminal device, and the other way is to use an AOI detection device or an image acquisition device. The collected liquid crystal screen image to be detected is directly uploaded to the cloud server, so that the terminal device for executing the liquid crystal screen defect detection method downloads the liquid crystal screen image to be detected from the cloud server as required. In addition, the manner of acquiring the liquid crystal screen image to be detected may also be to directly receive the liquid crystal screen image input by the user.

Step S104 , the liquid crystal screen image is input into the pre-trained defect detection model, so that the defect detection model performs defect detection on the liquid crystal screen image.

In one embodiment, the defect detection model can be implemented by using a neural network structure, and the neural network is an operation model composed of a large number of nodes (or neurons) connected to each other. Each node represents a specific output function, called the activation function. The connection between each two nodes represents a weighted value for the signal passing through the connection, called the weight, which is equivalent to the memory of the artificial neural network. The output of the network varies according to the way the network is connected, the weight value and the excitation function. A neural network is usually an approximation of an algorithm or function in nature, or an expression of a logical strategy.

In practical applications, the defect detection model can be obtained by training multiple sets of training images in advance. The multiple sets of training images are all defect images of the LCD screen. Each set of training images can be multiple images with a defect, and each set of training images The defect categories of the training images are different; for example, the scratch defect category corresponds to multiple scratch defect images with different expressions, the point defect category corresponds to multiple point defect images with different expressions, and the line defect category corresponds to multiple images with different expressions. Line defect images, etc.; the defect detection model to be trained obtains reasonable detection results through training, that is, when the loss function of the defect detection model converges to a preset threshold, the training is stopped. The defect detection model obtained by pre-training has been able to more accurately detect various types of defects on the LCD screen and various manifestations of each type of defects.

Step S106 , acquiring the defect detection result output by the defect detection model; wherein the defect detection result includes the liquid crystal screen defect category.

The LCD screen defect category can include appearance defects and display defects, wherein appearance defects include one or more of scratch defects, dent defects, burr defects, bubble defects and foreign body defects; display defects include point defects, line defects and one or more of the blemishes. For example, the scratches on the screen of the mobile phone belong to scratch defects, and the bright spot on the screen of the mobile phone cannot be displayed normally, and the defect is a point defect.

The above-mentioned defect detection method for a liquid crystal screen provided by the embodiment of the present invention obtains an image of the liquid crystal screen to be detected, and inputs the image of the liquid crystal screen into a pre-trained defect detection model, so that the defect detection model can perform defect detection on the liquid crystal screen image. , and then obtain the defect detection result output by the defect detection model. In the embodiment of the present invention, a defect detection model is used to detect the defects of the liquid crystal screen, and the detection results including the categories of the defects of the liquid crystal screen are obtained, which saves the labor cost and improves the detection efficiency of the defects of the liquid crystal screen.

For ease of understanding, a specific implementation of training a flaw detection model in the flaw detection method for a liquid crystal screen provided by the present embodiment is given below. Refer to the flowchart of a training process of a flaw detection model shown in FIG. 2 . Figure, the method includes the following steps:

Step S202, acquiring multiple groups of training images; wherein each group includes multiple training images, and the training images between groups carry different defect category labels; the training images within a group have the same defect category labels and different defect styles.

The training images in multiple groups are all defect images of the LCD screen, and the training images are divided into different groups according to the different defect categories. The defect styles are different in each training image, and each training image has a style of defects. For example, the defects of the first group of training images are all foreign body defects, the defects of the second group of training images are all point defects, and each group has multiple training images, wherein the training images in the first group include training images with transparent foreign objects , training images of turbid foreign objects, and other training images with different defect styles. The training images in the second group include training images with dead pixels, training images with dark spots, and training images with bright spots. Multiple training images. The above is only a schematic description, and details are not repeated here.

Step S204 , inputting multiple sets of training images into a preset neural network structure, and calculating a loss function value of the neural network structure.

The flaw detection model is trained based on a preset neural network structure. The neural network structure can be a deep convolutional neural network such as a deeplab neural network. Semantic segmentation is performed on the training image through semantic image segmentation to obtain segmentation results. Based on the segmentation results, the training Image flaw detection, that is, training a flaw detection model.

In step S206, based on the loss function value, the network parameters of the neural network structure are trained by the back-propagation algorithm, and the training is stopped until the loss function value converges to a preset value.

When the loss function converges to the preset threshold, the optimal network parameters are obtained. At this time, the defect detection model can more accurately detect the defects of the LCD screen and output better detection results, and the defect detection model can be stopped. training.

The defect detection model obtained by the above training provided by the embodiment of the present invention detects the defects of the liquid crystal screen, and by training the network parameters of the neural network structure, the loss function is converged to the preset value, so as to obtain the best network parameters, and then This enables the defect detection model to more accurately detect the defects of the LCD screen.

For ease of understanding, a specific implementation of the liquid crystal screen defect detection method provided by this embodiment is given below. Referring to the flowchart of another liquid crystal screen defect detection method shown in FIG. 3 , the method includes the following steps. :

In step S302, an image of the LCD screen to be detected is obtained through an AOI detection device; wherein, the AOI detection device is set up at a designated position on the production line of the LCD screen, and the LCD screen image to be detected is an image of an abnormal LCD screen detected by the AOI detection device.

The image of the abnormal liquid crystal screen is the image of the defective liquid crystal screen. The types of the defects have been described in the above-mentioned embodiments, and will not be repeated here.

Step S304, the LCD screen image is input into the pre-trained defect detection model.

Flaw detection models can be implemented using deep convolutional neural networks (CNNs). A deep convolutional neural network is a feedforward neural network in which artificial neurons respond to surrounding units and can perform large-scale image processing. Compared with traditional neural networks, deep convolutional neural networks have better recognition effects in feature recognition-related tasks, and can be better applied to image detection.

Step S306, performing semantic segmentation processing on the LCD screen image through the defect detection model to obtain a segmentation result.

In specific implementation, the deep convolutional neural network performs semantic segmentation processing on the LCD screen image through semantic image segmentation. This embodiment provides the following three semantic segmentation methods:

Method 1: Semantic segmentation is performed on the LCD screen image by using the hole convolution method;

The atrous convolution method obtains more LCD image information through the atrous (with hole) algorithm, and uses a fully connected conditional random field (CRF) to improve the model's ability to capture details, without pooling loss information. Let each convolution output contain a larger range of information, so as to obtain the segmentation result of the LCD screen image.

Method 2: Perform semantic segmentation processing on the LCD screen image by using the empty space pyramid pooling method;

Atrous spatial pyramid pooling (ASPP) method is to remove sampling in the last few max pooling layers of neural network pooling layers, and instead use atrous convolution to calculate feature maps with higher sampling density . The LCD screen image is divided into sub-blocks according to the preset size, and the sub-blocks are subjected to the convolution operation to obtain the corresponding feature maps. The hole space pyramid pooling method enables the feature maps of any size to be converted into fixed-size feature vectors. LCD screen images achieve better semantic segmentation.

Method 3: Semantic segmentation processing is performed on the LCD screen image by a multi-scale convolution cascade method with holes.

The multi-scale atrous convolution cascade or parallel is adopted to capture the multi-scale background, and the previously proposed atrous spatial pyramid pooling module will be modified, which is used to explore the multi-scale convolution features and encode the global background based on the image level. Get features.

During specific implementation, any one of the above-mentioned manners may be selected for implementation according to requirements, which is not limited herein.

Step S308, performing flaw detection on the LCD screen image based on the segmentation result.

According to the segmentation result, the defect of the LCD screen image is detected in the defect detection model, and the defect type of the LCD screen in the LCD screen image is determined.

Step S310 , acquiring the defect detection result output by the defect detection model; wherein the defect detection result includes the liquid crystal screen defect category.

Specifically, if the detected defect of the LCD screen is a scratch defect, the defect detection result includes that the defect of the LCD screen is a scratch defect; the detected defect of the LCD screen is a bubble defect, the defect detection result includes the defect of the LCD screen It is a bubble defect; the detected defect of the LCD screen is a dark spot/bright spot defect, and the defect detection result includes the defect of the LCD screen as a point defect.

To sum up, using the above LCD defect detection method provided in this embodiment, an AOI detection device is used to obtain an image of an LCD screen to be detected, and a deep convolutional neural network is used to perform semantic segmentation processing on the LCD screen image through semantic image segmentation. , to obtain the segmentation result, so as to perform more accurate detection of the defects of the LCD screen image based on the segmentation result.

In a specific implementation, the flaw detection model may be a deeplab neural network, which is a top-notch deep learning model for image semantic segmentation, whose goal is to convert semantic labels (such as people, dogs, cats, etc.) Assigned to each pixel of the input image, the first function of deeplab neural network is combined with deep convolutional neural network, using atrous convolution for semantic segmentation, and the second function is based on the optimization of the first function, using atrous spatial pyramid pooling The third function is to use multi-scale atrous convolution cascades or parallel to capture the multi-scale background, and optimize the segmentation of ASPP based on image features. There is also a function for the third function. Extension to include a simple and efficient decoder module to improve segmentation results. The deep convolutional neural network mainly includes a convolution layer and a pooling layer. The convolutional layer and the pooling layer are used to perform convolution processing and pooling processing on the LCD screen image respectively, and the detection of the LCD screen image is realized by feature extraction. This embodiment can combine the advantages of deeplab neural network and apply it to defect detection of liquid crystal screens, replace the traditional inefficient method of manually discriminating defects, and improve the defect detection efficiency, and is suitable for all liquid crystal screens that need to be inspected. Defective occasions.

Corresponding to the aforementioned method for detecting defects in a liquid crystal screen, an embodiment of the present invention provides a device for detecting defects in a liquid crystal screen. Referring to the structural block diagram of a device for detecting defects in a liquid crystal screen shown in FIG. 4 , the device includes the following modules:

an image acquisition module 402, configured to acquire an image of the liquid crystal screen to be detected;

The defect detection module 404 is used for inputting the LCD screen image into the pre-trained defect detection model, so that the defect detection model can perform defect detection on the LCD screen image;

The result acquisition module 406 is configured to acquire the defect detection result output by the defect detection model, wherein the defect detection result includes the LCD screen defect category.

The above-mentioned liquid crystal screen defect detection device provided by the embodiment of the present invention obtains the liquid crystal screen image to be detected, and inputs the liquid crystal screen image into the pre-trained defect detection model, so that the defect detection model can perform defect detection on the liquid crystal screen image. , and then obtain the defect detection result output by the defect detection model. The embodiments of the present invention use a defect detection model formed by a neural network to detect the defects of the LCD screen and obtain detection results including the categories of LCD screen defects, which saves labor costs and improves the detection efficiency of LCD screen defects. The above-mentioned image acquisition module 402 is further configured to: acquire the image of the LCD screen to be detected through the image acquisition device; wherein, the image acquisition device is set up at a designated position on the LCD screen production line; the LCD screen image to be detected is the image to be detected collected by the image acquisition device The image of the LCD screen; and/or obtain the image of the LCD screen to be detected by the AOI detection device; wherein, the AOI detection device is set up at a designated position on the production line of the LCD screen, and the LCD screen image to be detected is the abnormality detected by the AOI detection device LCD screen image.

The above-mentioned defect detection module 404 is also used to: obtain multiple groups of training images; wherein, each group contains multiple training images, and the training images between the groups carry different defect category labels; the training images in the group have the same defect category labels, but the defect category labels are the same. Different styles; input multiple sets of training images into the preset neural network structure, and calculate the loss function value of the neural network structure; based on the loss function value, train the network parameters of the neural network structure through the back propagation algorithm until the loss function value Convergence to the preset value and stop training.

In the specific implementation, the neural network structure is a deep convolutional neural network; the above-mentioned defect detection module 404 includes: a segmentation unit for performing semantic segmentation processing on the LCD screen image to obtain segmentation results; a defect detection unit for based on the segmentation results. LCD screen image for defect detection. Among them, the segmentation unit is used to: use the hole convolution method to perform semantic segmentation processing on the LCD screen image; or use the hole space pyramid pooling method to perform semantic segmentation processing on the LCD screen image; or use a multi-scale hole convolution stage. Semantic segmentation is performed on the LCD screen image in a linked manner.

In the device of any one of the above embodiments, the liquid crystal screen defect category includes appearance defects and display defects; appearance defects include one or more of scratch defects, dent defects, burr defects, bubble defects and foreign matter defects; display defects Include one or more of point flaws, line flaws, and face flaws.

The implementation principle and the technical effects of the device provided in this embodiment are the same as those in the foregoing embodiments. For brief description, for the parts not mentioned in the device embodiment, reference may be made to the corresponding content in the foregoing method embodiments.

An embodiment of the present invention provides a terminal device. Referring to the schematic structural diagram of a terminal device shown in FIG. 5 , the terminal device includes: a processor 50 , a memory 51 , a bus 52 and a communication interface 53 , the processor 50 , The communication interface 53 and the memory 51 are connected through a bus 52 ; the processor 50 is used to execute executable modules, such as computer programs, stored in the memory 51 .

The memory 51 may include a high-speed random access memory (RAM, Random Access Memory), and may also include a non-volatile memory (non-volatile memory), such as at least one disk memory. The communication connection between the system network element and at least one other network element is realized through at least one communication interface 53 (which may be wired or wireless), and the Internet, wide area network, local area network, metropolitan area network, etc. may be used.

The bus 52 may be an ISA bus, a PCI bus, an EISA bus, or the like. The bus can be divided into an address bus, a data bus, a control bus, and the like. For ease of representation, only one bidirectional arrow is shown in FIG. 5, but it does not mean that there is only one bus or one type of bus.

The memory 51 is used to store a program, and the processor 50 executes the program after receiving the execution instruction, and the method executed by the device defined by the stream process disclosed in any of the foregoing embodiments of the present invention can be applied to processing in the processor 50 , or implemented by the processor 50 .

The processor 50 may be an integrated circuit chip with signal processing capability. In the implementation process, each step of the above-mentioned method may be completed by a hardware integrated logic circuit in the processor 50 or an instruction in the form of software. The above-mentioned processor 50 may be a general-purpose processor, including a central processing unit (CPU for short), a network processor (NP for short), etc.; it may also be a digital signal processor (Digital Signal Processing, DSP for short) , Application Specific Integrated Circuit (ASIC for short), Field-Programmable Gate Array (FPGA for short) or other programmable logic devices, discrete gate or transistor logic devices, and discrete hardware components. Various methods, steps, and logical block diagrams disclosed in the embodiments of the present invention can be implemented or executed. A general purpose processor may be a microprocessor or the processor may be any conventional processor or the like. The steps of the method disclosed in conjunction with the embodiments of the present invention may be directly embodied as executed by a hardware decoding processor, or executed by a combination of hardware and software modules in the decoding processor. The software modules may be located in random access memory, flash memory, read-only memory, programmable read-only memory or electrically erasable programmable memory, registers and other storage media mature in the art. The storage medium is located in the memory 51, and the processor 50 reads the information in the memory 51 and completes the steps of the above method in combination with its hardware.

Embodiments of the present invention further provide a computer-readable storage medium, where a computer program is stored on the computer-readable storage medium, and when the computer program is run by a processor, the steps of the method in any one of the foregoing embodiments are executed.

Those skilled in the art can clearly understand that, for the convenience and brevity of description, for the specific working process of the system described above, reference may be made to the corresponding process in the foregoing embodiment, and details are not repeated here.

The liquid crystal screen defect detection method, device, and computer program product of the terminal device provided by the embodiments of the present invention include a computer-readable storage medium storing program codes, and the instructions included in the program codes can be used to execute the foregoing method embodiments. For the specific implementation of the method, reference may be made to the method embodiment, which will not be repeated here.

The functions, if implemented in the form of software functional units and sold or used as independent products, may be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention can be embodied in the form of a software product in essence, or the part that contributes to the prior art or the part of the technical solution. The computer software product is stored in a storage medium, including Several instructions are used to cause a computer device (which may be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, Read-Only Memory (ROM, Read-Only Memory), Random Access Memory (RAM, Random Access Memory), magnetic disk or optical disk and other media that can store program codes .

Finally, it should be noted that the above-mentioned embodiments are only specific implementations of the present invention, and are used to illustrate the technical solutions of the present invention, but not to limit them. The protection scope of the present invention is not limited thereto, although referring to the foregoing The embodiment has been described in detail the present invention, those of ordinary skill in the art should understand: any person skilled in the art who is familiar with the technical field within the technical scope disclosed by the present invention can still modify the technical solutions described in the foregoing embodiments. Or can easily think of changes, or equivalently replace some of the technical features; and these modifications, changes or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present invention, and should be covered in the present invention. within the scope of protection. Therefore, the protection scope of the present invention should be based on the protection scope of the claims.

Claims (14)

1. a defect detection method of liquid crystal screen, is characterized in that, comprises: Obtain the LCD screen image to be detected; inputting the liquid crystal screen image into a pre-trained flaw detection model, so that the flaw detection model performs flaw detection on the liquid crystal screen image; Acquiring a defect detection result output by the defect detection model; wherein the defect detection result includes a liquid crystal screen defect category. 2. The method according to claim 1, wherein the step of acquiring the liquid crystal screen image to be detected comprises: The image of the LCD screen to be detected is acquired by an image acquisition device; wherein, the image acquisition device is set up at a designated position on the LCD screen production line; the LCD screen image to be detected is the image of the LCD screen to be detected collected by the image acquisition device. image; and / or Obtain the image of the LCD screen to be detected by the AOI detection device; wherein, the AOI detection device is set up at a designated position on the production line of the LCD screen, and the LCD screen image to be detected is the abnormal LCD screen detected by the AOI detection device. Image. 3. The method according to claim 1, wherein the training process of the flaw detection model comprises: Acquiring multiple sets of training images; wherein each group contains multiple training images, and the training images between groups carry different defect category labels; the training images within the group have the same defect category labels, but different defect styles; Inputting multiple groups of the training images into a preset neural network structure, and calculating the loss function value of the neural network structure; Based on the loss function value, the network parameters of the neural network structure are trained through a back-propagation algorithm, and the training is stopped until the loss function value converges to a preset value. 4. The method according to claim 1, wherein the flaw detection model is a deep convolutional neural network; The step of performing flaw detection on the LCD screen image by the flaw detection model includes: Perform semantic segmentation processing on the LCD screen image to obtain a segmentation result; Defect detection is performed on the LCD screen image based on the segmentation result. 5. The method according to claim 4, wherein the step of performing semantic segmentation processing on the LCD screen image comprises: Semantic segmentation processing is performed on the LCD screen image by using a hole convolution method; or, Semantic segmentation processing is performed on the LCD screen image by using the empty space pyramid pooling method; or, The liquid crystal screen image is semantically segmented by a multi-scale convolution cascade with holes. 6. The method according to any one of claims 1 to 5, wherein, The LCD screen defect categories include appearance defects and display defects; The appearance flaws include one or more of scratch flaws, dent flaws, burr flaws, bubble flaws and foreign body flaws; The display flaws include one or more of point flaws, line flaws, and surface flaws. 7. A defect detection device for a liquid crystal screen, characterized in that, comprising: an image acquisition module for acquiring an image of the LCD screen to be detected; A defect detection module, configured to input the LCD image into a pre-trained defect detection model, so that the defect detection model can perform defect detection on the LCD image; The result acquisition module is used for acquiring the defect detection result output by the defect detection model; wherein, the defect detection result includes the liquid crystal screen defect category. 8. The device according to claim 7, wherein the image acquisition module is used for: The image of the LCD screen to be detected is acquired by an image acquisition device; wherein, the image acquisition device is set up at a designated position on the LCD screen production line; the LCD screen image to be detected is the image of the LCD screen to be detected collected by the image acquisition device. image; and / or Obtain the image of the LCD screen to be detected by the AOI detection device; wherein, the AOI detection device is set up at a designated position on the production line of the LCD screen, and the LCD screen image to be detected is the abnormal LCD screen detected by the AOI detection device. Image. 9. The device according to claim 7, wherein the defect detection module is further used for: Acquiring multiple sets of training images; wherein, each group contains multiple training images, and the training images between the groups carry different defect category labels; the training images within the group have the same defect category labels, but different defect styles; Inputting multiple groups of the training images into a preset neural network structure, and calculating the loss function value of the neural network structure; Based on the loss function value, the network parameters of the neural network structure are trained through a back-propagation algorithm, and the training is stopped until the loss function value converges to a preset value. 10. The device according to claim 7, wherein the defect detection model is a deep convolutional neural network; The defect detection module includes: a segmentation unit, configured to perform semantic segmentation processing on the LCD screen image to obtain segmentation results; A defect detection unit, configured to perform defect detection on the liquid crystal screen image based on the segmentation result. 11. The device according to claim 10, wherein the dividing unit is used for: Semantic segmentation processing is performed on the LCD screen image by using a hole convolution method; or, Semantic segmentation processing is performed on the LCD screen image by using the empty space pyramid pooling method; or, The liquid crystal screen image is semantically segmented by a multi-scale convolution cascade with holes. 12. The device according to any one of claims 7 to 11, wherein the liquid crystal screen defect categories include appearance defects and display defects; The appearance flaws include one or more of scratch flaws, dent flaws, burr flaws, bubble flaws and foreign body flaws; The display flaws include one or more of point flaws, line flaws, and surface flaws. 13. A terminal device, characterized in that the terminal device comprises a memory and a processor, the memory is used to store a program that supports the processor to execute the method according to any one of claims 1 to 6, the processor is is configured to execute a program stored in the memory. 14. A computer storage medium, characterized in that it is used for storing computer software instructions used in the method of any one of claims 1 to 6.

Images