Set-top box production line monitoring method, device and equipment based on artificial intelligence

Abstract

The artificial intelligence-based set-top box production line monitoring method, device and equipment provided in the present application relate to the field of artificial intelligence technology. In the present application, first, the local production line monitoring data corresponding to each set-top box component is obtained; secondly, for each set-top box component, the set-top box component is used as the target set-top box component, and, based on the local production line monitoring data corresponding to other set-top box components, the local production line monitoring data corresponding to the target set-top box component is first associated with mining to obtain the target component image feature, and, based on the target reflow soldering data, the target component image feature is second associated with mining to obtain the associated component image feature; then, the associated component image feature corresponding to each set-top box component is respectively subjected to welding quality analysis to obtain the corresponding welding quality analysis result. Based on the above content, the problem of relatively low reliability of set-top box production line monitoring existing in the prior art can be improved.

Description

Technical Field

[0001] The present application relates to the field of artificial intelligence technology, and more specifically, to a set-top box production line monitoring method, device and equipment based on artificial intelligence. Background Art

[0002] With the rapid development of information technology and smart home devices, set-top boxes, as important home entertainment and information receiving devices, are widely used in daily life. In order to ensure the production quality of set-top boxes, production line monitoring and inspection become particularly important. Existing set-top box production line monitoring technologies mainly rely on manual monitoring and traditional automation equipment, but these technologies face multiple challenges.

[0003] Among them, the inspection of welding quality during the set-top box production process usually relies on manual judgment or traditional visual inspection methods. These methods cannot fully identify subtle welding defects and are subject to manual experience and subjective factors. It is difficult to achieve efficient and accurate quality control, and the reliability is relatively low. Summary of the invention

[0004] In view of this, the purpose of the present application is to provide a set-top box production line monitoring method, device and equipment based on artificial intelligence to improve the problem of relatively low reliability of set-top box production line monitoring in the prior art.

[0005] To achieve the above objectives, this application adopts the following technical solutions:

[0006] A set-top box production line monitoring method based on artificial intelligence, comprising:

[0007] Acquire target production line monitoring data and target reflow soldering data corresponding to the target production line monitoring data, and parse the target production line monitoring data to obtain local production line monitoring data corresponding to each set-top box component of a plurality of set-top box components, wherein the local production line monitoring data includes image data formed by image acquisition of solder joints of the corresponding set-top box components, and the target reflow soldering data includes at least temperature timing data;

[0008] For each set-top box component, the set-top box component is used as a target set-top box component, and based on the local production line monitoring data corresponding to other set-top box components, the local production line monitoring data corresponding to the target set-top box component is subjected to first association mining to obtain a target component image feature corresponding to the target set-top box component, and based on the target reflow soldering data, the target component image feature is subjected to second association mining to obtain an associated component image feature corresponding to the target set-top box component;

[0009] The welding quality analysis is performed on the associated component image features corresponding to each set-top box component respectively, and the welding quality analysis result corresponding to each set-top box component is obtained.

[0010] In a preferred embodiment of the present application, in the above-mentioned artificial intelligence-based set-top box production line monitoring method, for each set-top box component, the set-top box component is used as a target set-top box component, and based on the local production line monitoring data corresponding to other set-top box components, the local production line monitoring data corresponding to the target set-top box component is subjected to a first association mining to obtain the target component image feature corresponding to the target set-top box component, and based on the target reflow soldering data, the target component image feature is subjected to a second association mining to obtain the associated component image feature corresponding to the target set-top box component, the step includes:

[0011] Convolving the local production line monitoring data corresponding to the target set-top box component to obtain a corresponding first component image convolution feature, and convolving the local production line monitoring data corresponding to each other set-top box component to obtain a corresponding second component image convolution feature;

[0012] For each other set-top box component, using the first association mining unit corresponding to the other set-top box component in the target quality analysis network, the second component image convolution feature corresponding to the other set-top box component and the first component image convolution feature are subjected to first association mining to obtain the first component image association feature corresponding to the other set-top box component;

[0013] splicing the first component image associated features corresponding to each other set-top box component to obtain the corresponding first component image splicing features, and performing feature space conversion on the first component image splicing features to obtain the corresponding target component image features;

[0014] Based on the target reflow soldering data, a second association mining is performed on the target component image feature to obtain an associated component image feature corresponding to the target set-top box component.

[0015] In a preferred embodiment of the present application, in the above-mentioned artificial intelligence-based set-top box production line monitoring method, for each other set-top box component, using the first association mining unit corresponding to the other set-top box component in the target quality analysis network, the second component image convolution feature corresponding to the other set-top box component and the first component image convolution feature are subjected to first association mining to obtain the first component image association feature corresponding to the other set-top box component, including:

[0016] For each other set-top box component, the second component image convolution feature corresponding to the other set-top box component is loaded into the first association mining unit corresponding to the other set-top box component in the target quality analysis network, and the second component image convolution feature corresponding to the other set-top box component is masked using the masking parameter matrix carried in the first association mining unit to form the corresponding first component image masking feature, wherein the target quality analysis network belongs to a neural network, and the masking parameter matrix is ​​formed by the target quality analysis network in a corresponding training process;

[0017] An operation of determining an association relationship between the first component image mask feature and the first component image convolution feature is performed, and based on the determined association relationship, the first component image convolution feature is updated to form a corresponding first component image association feature, wherein the semantic information in the first component image association feature is semantic information mined from the first component image convolution feature and has an association relationship with the semantic information in the first component image mask feature.

[0018] In a preferred embodiment of the present application, in the above-mentioned artificial intelligence-based set-top box production line monitoring method, the steps of splicing the first component image associated features corresponding to each other set-top box component to obtain the corresponding first component image splicing features, and performing feature space conversion on the first component image splicing features to obtain the corresponding target component image features include:

[0019] splicing the first component image associated features corresponding to each other set-top box component to obtain the corresponding first component image splicing features, and performing feature space transformation on the first component image splicing features to obtain the corresponding first component image transformation features, wherein the feature space transformation includes a self-attention operation, a convolution operation, a pooling operation, and a full connection operation;

[0020] The first component image conversion feature and the first component image convolution feature are fused to form a corresponding target component image feature.

[0021] In a preferred embodiment of the present application, in the above-mentioned artificial intelligence-based set-top box production line monitoring method, the step of performing a second association mining on the target component image feature based on the target reflow soldering data to obtain the associated component image feature corresponding to the target set-top box component includes:

[0022] Embedding the target reflow soldering data to obtain corresponding target reflow soldering embedding features;

[0023] Based on the target reflow soldering embedding feature, a second association mining is performed on the target component image feature to obtain an associated component image feature corresponding to the target set-top box component.

[0024] In a preferred embodiment of the present application, in the above-mentioned artificial intelligence-based set-top box production line monitoring method, the step of performing a second association mining on the target component image feature based on the target reflow soldering embedding feature to obtain the associated component image feature corresponding to the target set-top box component includes:

[0025] Utilizing the first branch included in the second association mining unit in the target quality analysis network, the target reflow soldering embedded features are subjected to multiple levels of deep mining to obtain corresponding multiple levels of target reflow soldering deep features, wherein the input features of the deep mining of the latter level are the output features of the deep mining of the previous level, the input features of the deep mining of the first level are the target reflow soldering embedded features, and the deep mining at least includes downsampling;

[0026] Utilizing the second branch included in the second association mining unit, based on the target reflow depth features of the multiple levels, the target component image features are subjected to multiple levels of feature restoration to obtain corresponding target component image restoration features of the multiple levels, wherein, in the feature restoration of the first level, based on the association relationship between the target reflow depth features of the last level and the target component image features, the target component image features are updated to form the target component image restoration features of the first level, and in the feature restoration of each level after the first level, the target component image restoration features of the previous level are upsampled to obtain corresponding upsampled features, and, based on the association relationship between the target reflow depth features of the corresponding level and the upsampled features, the upsampled features are updated to form the target component image restoration features of the corresponding level;

[0027] Based on the target component image restoration features of the multiple levels, the associated component image features corresponding to the target set-top box component are obtained.

[0028] In a preferred embodiment of the present application, in the above-mentioned artificial intelligence-based set-top box production line monitoring method, the step of performing a second association mining on the target component image feature based on the target reflow soldering embedding feature to obtain the associated component image feature corresponding to the target set-top box component includes:

[0029] Utilizing the first branch included in the second association mining unit in the target quality analysis network, the target reflow soldering embedded features are subjected to multiple levels of self-attention processing to obtain corresponding multiple levels of target reflow soldering attention features, wherein the input features of the self-attention processing of the latter level are the output features of the self-attention processing of the previous level, and the input features of the self-attention processing of the first level are the target reflow soldering embedded features;

[0030] Using the second branch included in the second association mining unit, multiple levels of self-attention processing are performed on the target component image features to obtain corresponding multiple levels of target component attention features, wherein the input features of the self-attention processing of the latter level are the output features of the self-attention processing of the previous level, and the input features of the self-attention processing of the first level are the target component image features;

[0031] Utilizing the third branch included in the second association mining unit, based on the target reflow soldering attention features of the multiple levels, multiple levels of association mining are performed on the target component attention features of the multiple levels to obtain corresponding target component association features of the multiple levels, wherein, in the association mining of the first level, based on the association relationship between the target reflow soldering attention features of the last level and the target component attention features of the last level, the target component attention features are updated to form the target component association features of the first level, and in the association mining of each level after the first level, the target component attention features of the corresponding level and the target component association features of the previous level are fused to form corresponding fused features, and, based on the association relationship between the target reflow soldering attention features of the corresponding level and the fused features, the fused features are updated to form the target component association features of the corresponding level;

[0032] Based on the target component association features of the multiple levels, the associated component image features corresponding to the target set-top box component are obtained.

[0033] In a preferred embodiment of the present application, in the above-mentioned artificial intelligence-based set-top box production line monitoring method, the artificial intelligence-based set-top box production line monitoring method further includes:

[0034] Acquire sample production line monitoring data and sample reflow data corresponding to the sample production line monitoring data, and parse the sample production line monitoring data to obtain local sample production line monitoring data corresponding to each set-top box component in a plurality of set-top box components;

[0035] In the candidate quality analysis network, for each set-top box component, the set-top box component is taken as a target set-top box component, and based on the local sample production line monitoring data corresponding to other set-top box components, the local sample production line monitoring data corresponding to the target set-top box component is subjected to first association mining to obtain the sample component image feature corresponding to the target set-top box component, and based on the sample reflow data, the sample component image feature is subjected to second association mining to obtain the associated sample component image feature corresponding to the target set-top box component;

[0036] In the candidate quality analysis network, welding quality analysis is performed on the associated sample component image features corresponding to each set-top box component to obtain welding quality analysis data corresponding to each set-top box component;

[0037] Based on the error between the welding quality analysis data and the corresponding welding quality label, the network parameters of the candidate quality analysis network are updated to form a target quality analysis network, wherein the target quality analysis network is at least used to perform welding quality analysis on the associated component image features corresponding to each set-top box component, respectively, to obtain the welding quality analysis results corresponding to each set-top box component.

[0038] The present application also provides a set-top box production line monitoring device based on artificial intelligence, comprising:

[0039] a data acquisition module, used to acquire target production line monitoring data and target reflow soldering data corresponding to the target production line monitoring data, and parse the target production line monitoring data to obtain local production line monitoring data corresponding to each set-top box component among a plurality of set-top box components, wherein the local production line monitoring data includes image data formed by image acquisition of solder joints of the corresponding set-top box components, and the target reflow soldering data includes at least temperature timing data;

[0040] The association mining module is used for taking each set-top box component as a target set-top box component, and performing first association mining on the local production line monitoring data corresponding to the target set-top box component based on the local production line monitoring data corresponding to other set-top box components to obtain a target component image feature corresponding to the target set-top box component, and performing second association mining on the target component image feature based on the target reflow soldering data to obtain an associated component image feature corresponding to the target set-top box component;

[0041] The quality analysis module is used to perform welding quality analysis on the associated component image features corresponding to each set-top box component, and obtain the welding quality analysis result corresponding to each set-top box component.

[0042] Based on the above, the present application also provides an electronic device, including:

[0043] Memory for storing computer programs;

[0044] A processor connected to the memory is used to execute the computer program stored in the memory to implement the above-mentioned artificial intelligence-based set-top box production line monitoring method.

[0045] The artificial intelligence-based set-top box production line monitoring method, device and equipment provided in the present application first obtain local production line monitoring data corresponding to each set-top box component; secondly, for each set-top box component, the set-top box component is used as the target set-top box component, and, based on the local production line monitoring data corresponding to other set-top box components, the local production line monitoring data corresponding to the target set-top box component is subjected to first association mining to obtain target component image features, and, based on the target reflow soldering data, the target component image features are subjected to second association mining to obtain associated component image features; then, welding quality analysis is performed on the associated component image features corresponding to each set-top box component to obtain corresponding welding quality analysis results. Based on the above content, before performing the welding quality analysis, the internal correlation mining between the target production line monitoring data will be performed on the set-top box components, and the external correlation mining will be performed with the target reflow soldering data, so that the semantics of the obtained associated component image features are richer and the representation ability is higher. Therefore, the reliability of the basis for analysis is higher, thereby ensuring that the reliability of the obtained welding quality analysis results is also higher. In addition, since each set-top box component is analyzed separately, the granularity of the analysis is smaller, and therefore, it has higher accuracy, that is, the reliability of the welding quality analysis results is further improved, thereby improving the problem of relatively low reliability of set-top box production line monitoring in the prior art. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] In order to make the above-mentioned objects, features and advantages of the present application more obvious and easy to understand, preferred embodiments are specifically cited below and described in detail with reference to the attached drawings.

[0047] Figure 1 A structural block diagram of an electronic device provided in an embodiment of the present application.

[0048] Figure 2 A flowchart of an artificial intelligence-based set-top box production line monitoring method provided in an embodiment of the present application.

[0049] Figure 3 A first schematic diagram of the second association mining provided in an embodiment of the present application.

[0050] Figure 4 A second schematic diagram of the second association mining provided in an embodiment of the present application.

[0051] Figure 5 A block diagram of an artificial intelligence-based set-top box production line monitoring device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0052] In order to make the purpose, technical solution and advantages of the embodiments of the present application clearer, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. The components of the embodiments of the present application described and shown in the drawings here can be arranged and designed in various different configurations.

[0053] Therefore, the following detailed description of the embodiments of the present application provided in the accompanying drawings is not intended to limit the scope of the present application for which protection is sought, but merely represents selected embodiments of the present application. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in the field without creative work are within the scope of protection of the present application.

[0054] like Figure 1 As shown, an embodiment of the present application provides an electronic device, wherein the electronic device may include a memory, a processor, and a set-top box production line monitoring device based on artificial intelligence.

[0055] In detail, the memory and the processor are electrically connected directly or indirectly to realize data transmission or interaction. For example, the memory and the processor can be electrically connected through one or more communication buses or signal lines. The set-top box production line monitoring device based on artificial intelligence includes at least one software function module stored in the memory in the form of software or firmware. The processor is used to execute an executable computer program stored in the memory, for example, the software function module and computer program included in the set-top box production line monitoring device based on artificial intelligence, so as to realize the set-top box production line monitoring method based on artificial intelligence provided in the embodiment of the present application.

[0056] Optionally, the memory may be, but not limited to, a random access memory (RAM), a read only memory (ROM), a programmable read-only memory (PROM), an erasable programmable read-only memory (EPROM), an electrically erasable programmable read-only memory (EEPROM), etc. Furthermore, the processor may be a general-purpose processor, including a central processing unit (CPU), a network processor (NP), a system on chip (SoC), etc.; it may also be a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field programmable gate array (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components.

[0057] Understandably, Figure 1 The structure shown is for illustration only, and the electronic device may also include Figure 1 More or fewer components as shown, or with Figure 1 The different configurations shown, for example, may also include a communication unit for exchanging information with other devices.

[0058] Combination Figure 2 The embodiment of the present application also provides a set-top box production line monitoring method based on artificial intelligence that can be applied to the above electronic device. Among them, the method steps defined in the process related to the set-top box production line monitoring method based on artificial intelligence can be implemented by the electronic device.

[0059] The following will Figure 2 The specific process shown is explained in detail.

[0060] Step S110, acquiring target production line monitoring data and target reflow data corresponding to the target production line monitoring data, and parsing the target production line monitoring data to obtain local production line monitoring data corresponding to each of the plurality of set-top box components.

[0061] In an embodiment of the present application, the electronic device can obtain target production line monitoring data and target reflow data corresponding to the target production line monitoring data, and parse the target production line monitoring data to obtain local production line monitoring data corresponding to each set-top box component in a plurality of set-top box components. The plurality of set-top box components include resistors, capacitors, diodes, inductors, etc., or may also include other electronic components. In addition, the local production line monitoring data includes image data formed by image acquisition of solder joints of corresponding set-top box components, and the target reflow data includes at least temperature timing data, such as temperature changes during the entire process of reflow from heating to end.

[0062] Step S120, for each set-top box component, the set-top box component is taken as a target set-top box component, and, based on the local production line monitoring data corresponding to other set-top box components, the local production line monitoring data corresponding to the target set-top box component is subjected to a first association mining to obtain a target component image feature corresponding to the target set-top box component, and, based on the target reflow data, a second association mining is performed on the target component image feature to obtain an associated component image feature corresponding to the target set-top box component.

[0063] In an embodiment of the present application, after obtaining the local production line monitoring data and the target reflow soldering data, the electronic device can, for each set-top box component, take the set-top box component as the target set-top box component, and, based on the local production line monitoring data corresponding to other set-top box components, perform a first association mining on the local production line monitoring data corresponding to the target set-top box component to obtain a target component image feature corresponding to the target set-top box component, and, based on the target reflow soldering data, perform a second association mining on the target component image feature to obtain an associated component image feature corresponding to the target set-top box component. For example, there are a first set-top box component, a second set-top box component and a third set-top box component. Based on the local production line monitoring data corresponding to the second set-top box component and the local production line monitoring data corresponding to the third set-top box component, a first association mining can be performed on the local production line monitoring data corresponding to the first set-top box component to obtain the target component image feature corresponding to the first set-top box component. Then, based on the target reflow data, a second association mining can be performed on the target component image feature to obtain the associated component image feature corresponding to the first set-top box component. Based on the same method, the associated component image features corresponding to the second set-top box component and the third set-top box component can be obtained.

[0064] Step S130 , performing welding quality analysis on the associated component image features corresponding to each set-top box component, and obtaining welding quality analysis results corresponding to each set-top box component.

[0065] In an embodiment of the present application, after obtaining the associated component image features, the electronic device can perform welding quality analysis on the associated component image features corresponding to each set-top box component respectively, and obtain a welding quality analysis result corresponding to each set-top box component, such as a value between 0-10 (which can be achieved through a corresponding linear regression function), or a classification result such as good quality, general, or poor quality (which can be achieved through a corresponding classification function, such as softmax, etc.).

[0066] Based on the above content, before performing the welding quality analysis, the internal correlation mining between the target production line monitoring data will be performed on the set-top box components, and the external correlation mining will be performed with the target reflow soldering data, so that the semantics of the obtained associated component image features are richer and the representation ability is higher. Therefore, the reliability of the basis for analysis is higher, thereby ensuring that the reliability of the obtained welding quality analysis results is also higher. In addition, since each set-top box component is analyzed separately, the granularity of the analysis is smaller, and therefore, it has higher accuracy, that is, the reliability of the welding quality analysis results is further improved, thereby improving the problem of relatively low reliability of set-top box production line monitoring in the prior art.

[0067] It should be noted that for step S120 , the specific manner of performing the first association mining and the second association mining is not limited and can be selected according to actual needs.

[0068] For example, in an alternative embodiment, in order to ensure that the mined image features of related components have more accurate semantic representation capabilities on the basis of relatively rich semantic information, the above-mentioned step S120 may further include step S121, step S122, step S123 and step S124, and the specific contents of each step are described as follows.

[0069] Step S121, convolving the local production line monitoring data corresponding to the target set-top box component to obtain the corresponding first component image convolution feature, and convolving the local production line monitoring data corresponding to each other set-top box component to obtain the corresponding second component image convolution feature.

[0070] In an embodiment of the present application, the local production line monitoring data corresponding to the target set-top box component can be convolved to obtain the corresponding first component image convolution feature, and the local production line monitoring data corresponding to each other set-top box component can be convolved to obtain the corresponding second component image convolution feature. Specifically, each local production line monitoring data can be convolved by a convolutional neural network to obtain the corresponding first component image convolution feature and second component image convolution feature. Alternatively, in other embodiments, after convolution by a convolutional neural network, the convolved features can be further processed, such as self-attention processing, to obtain the first component image convolution feature and the second component image convolution feature.

[0071] Step S122, for each other set-top box component, using the first association mining unit corresponding to the other set-top box component in the target quality analysis network, the second component image convolution feature corresponding to the other set-top box component and the first component image convolution feature are subjected to first association mining to obtain the first component image association feature corresponding to the other set-top box component.

[0072] In the embodiment of the present application, after obtaining the first component image convolution feature and the second component image convolution feature, for each other set-top box component, the first association mining unit corresponding to the other set-top box component in the target quality analysis network can be used to perform first association mining on the second component image convolution feature corresponding to the other set-top box component and the first component image convolution feature to obtain the first component image association feature corresponding to the other set-top box component. In other words, the target quality analysis network has a first association mining unit corresponding to each set-top box component, for example, the first set-top box component corresponds to the first association mining unit 1, the second set-top box component corresponds to the first association mining unit 2, the third set-top box component corresponds to the first association mining unit 3, and the fourth set-top box component corresponds to the first association mining unit 4.

[0073] Step S123, splicing the first component image associated features corresponding to each other set-top box component to obtain the corresponding first component image splicing features, and performing feature space conversion on the first component image splicing features to obtain the corresponding target component image features.

[0074] In an embodiment of the present application, after obtaining the first component image association feature, the first component image association features corresponding to each other set-top box component can be spliced ​​to obtain the corresponding first component image splicing features, and the first component image splicing features can be transformed in feature space to obtain the corresponding target component image features.

[0075] Step S124: Based on the target reflow soldering data, a second association mining is performed on the target component image feature to obtain an associated component image feature corresponding to the target set-top box component.

[0076] In an embodiment of the present application, after obtaining the target component image feature, a second association mining can be performed on the target component image feature based on the target reflow data to obtain the associated component image feature corresponding to the target set-top box component, that is, the semantic information in the target reflow data can be further aggregated into the target component image feature, or semantic information associated with the target reflow data can be mined from the target component image feature.

[0077] It is understandable that the specific manner of performing the first association mining in the above step S122 is not limited and can be selected according to actual needs.

[0078] For example, in an alternative implementation, in order to improve the accuracy of association mining so that the mined first component image association features have more accurate semantic representation capabilities, the above step S122 may further include the following sub-steps:

[0079] First, for each other set-top box component, the second component image convolution feature corresponding to the other set-top box component is loaded into the first association mining unit corresponding to the other set-top box component in the target quality analysis network, and the second component image convolution feature corresponding to the other set-top box component is masked using the mask parameter matrix carried in the first association mining unit to form the corresponding first component image mask feature, wherein the target quality analysis network belongs to a neural network, and the mask parameter matrix is ​​formed by the target quality analysis network in a corresponding training process (the initial mask parameter matrix can be randomly generated); and the size of the mask parameter matrix can be the same as the size of the second component image convolution feature, so that the mask parameter matrix and the parameters of the corresponding positions in the second component image convolution feature can be multiplied to obtain the parameters of the corresponding positions in the first component image mask feature, and in addition, the parameters of the mask parameter matrix are 0 or 1, and the number of parameters that are 0 can be less than or much less than the number of parameters that are 1;

[0080] Secondly, the association relationship between the first component image mask feature and the first component image convolution feature can be determined, and based on the determined association relationship, the first component image convolution feature can be updated to form a corresponding first component image association feature, wherein the semantic information in the first component image association feature is semantic information mined from the first component image convolution feature and has an association relationship with the semantic information in the first component image mask feature; illustratively, the first component image convolution feature can be transposed to obtain a corresponding output feature, and then the first component image mask feature can be dot-producted with the output feature to obtain a corresponding dot product matrix, and then the first component image convolution feature can be weightedly summed based on the dot product matrix to obtain the corresponding first component image association feature.

[0081] Based on this, since the masking parameter matrix is ​​formed during the training process, it can mask some unimportant semantic information in the convolutional features of the second element image, thereby avoiding corresponding interference caused by unimportant information, so that the accuracy of subsequent association mining can be improved.

[0082] It is understandable that the specific method of performing feature space conversion in the above step S123 is not limited and can be selected according to actual needs.

[0083] For example, in an alternative implementation, in order to avoid the problem of loss of important semantic information in the aforementioned first association mining, so that the obtained target component image features can reliably represent the actual data, the above step S123 may further include the following sub-steps:

[0084] First, the first component image associated features corresponding to each other set-top box component can be spliced ​​to obtain the corresponding first component image splicing features, and the first component image splicing features can be transformed in feature space to obtain the corresponding first component image conversion features, wherein the feature space conversion includes a self-attention operation, a convolution operation, a pooling operation and a full connection operation, that is, the first component image splicing features can be subjected to a self-attention operation to obtain the corresponding self-attention features, and the self-attention features can be subjected to a convolution operation to obtain the corresponding convolution features, and the convolution features can be subjected to a pooling operation to obtain the corresponding pooling features, and finally, the pooling features can be subjected to a full connection operation to obtain the first component image conversion features;

[0085] Secondly, the first component image conversion feature and the first component image convolution feature can be fused to form the corresponding target component image feature. For example, the first component image conversion feature and the first component image convolution feature can be added or averaged to achieve fusion to obtain the corresponding target component image feature. In this way, both the semantic information in the first component image convolution feature can be represented and the important information therein can be represented in detail.

[0086] It is understandable that the specific manner of performing the second association mining in the above step S124 is not limited and can be selected according to actual needs.

[0087] For example, in an alternative implementation, in order to ensure reliable implementation of the second association mining, the above-mentioned step S124 may further include step S124a and step S124b, and the specific content of each step is described as follows.

[0088] Step S124a, embedding the target reflow data to obtain corresponding target reflow embedding features.

[0089] In an embodiment of the present application, the target reflow data may be embedded to obtain corresponding target reflow embedding features. For example, the target reflow data may be embedded based on a corresponding word embedding model (the specific processing process may refer to the relevant prior art and will not be described in detail here) to obtain corresponding target reflow embedding features; or, the result of the embedding process may be further processed by self-attention to obtain target reflow embedding features.

[0090] Step S124b: Based on the target reflow soldering embedding feature, a second association mining is performed on the target component image feature to obtain an associated component image feature corresponding to the target set-top box component.

[0091] In an embodiment of the present application, after obtaining the target reflow soldering embedding feature, a second association mining may be performed on the target component image feature based on the target reflow soldering embedding feature to obtain the associated component image feature corresponding to the target set-top box component.

[0092] It can be understood that, in the above step S124b, the specific method of performing the second association mining is not limited and can be selected according to actual needs.

[0093] For example, in an alternative implementation, in order to capture more detailed information in the second association mining, the above step S124b may further include the following sub-steps:

[0094] First, the target reflow soldering embedded features can be subjected to multiple levels of deep mining using the first branch included in the second association mining unit in the target quality analysis network to obtain corresponding multiple levels of target reflow soldering deep features, wherein the input features of the deep mining of the latter level are the output features of the deep mining of the previous level, the input features of the deep mining of the first level are the target reflow soldering embedded features, and the deep mining at least includes downsampling, such as Figure 3 As shown, the target reflow soldering embedded feature is downsampled to obtain the target reflow soldering depth feature of the first level, and the target reflow soldering depth feature of the first level is downsampled to obtain the target reflow soldering depth feature of the first level, and the target reflow soldering depth feature of the second level is downsampled to obtain the target reflow soldering depth feature of the third level, wherein, in order to achieve downsampling, pooling or convolution processing with a stride greater than 1 can be performed;

[0095] Secondly, the second branch included in the second association mining unit can be used to perform multiple-level feature restoration on the target component image features based on the target reflow depth features of the multiple levels to obtain corresponding target component image restoration features of the multiple levels, wherein, in the feature restoration of the first level, based on the association relationship between the target reflow depth features of the last level and the target component image features, the target component image features are updated to form the target component image restoration features of the first level, and in the feature restoration of each level after the first level, the target component image restoration features of the previous level are restored. Upsampling (such as deconvolution or interpolation to achieve upsampling) to obtain corresponding upsampling features, and, based on the correlation between the target reflow depth feature of the corresponding level and the upsampling feature, updating the upsampling feature to form the target component image restoration feature of the corresponding level, for example, in the feature restoration of the second level, upsampling the target component image restoration feature of the first level to obtain the corresponding upsampling feature, and, based on the correlation between the target reflow depth feature of the second-to-last level and the upsampling feature, updating the upsampling feature to form the target component image restoration feature of the second level;

[0096] Finally, based on the target component image restoration features of the multiple levels, the associated component image features corresponding to the target set-top box component can be obtained; exemplarily, the target component image restoration features of the multiple levels can be fused to obtain the corresponding associated component image features, or the target component image restoration features of the last level can be used as the associated component image features.

[0097] For another example, in another alternative implementation, in order to capture more associated information in the second associated mining, the above step S124b may also include the following sub-steps:

[0098] First, the first branch included in the second association mining unit in the target quality analysis network can be used to perform multiple levels of self-attention processing on the target reflow soldering embedded features to obtain corresponding multiple levels of target reflow soldering attention features, wherein the input features of the self-attention processing of the latter level are the output features of the self-attention processing of the previous level, and the input features of the self-attention processing of the first level are the target reflow soldering embedded features; for example, the target reflow soldering embedded features can be self-attention processed to obtain the target reflow soldering attention features of the first level, and the target reflow soldering attention features of the first level can be self-attention processed to obtain the target reflow soldering attention features of the second level, and the target reflow soldering attention features of the second level can be self-attention processed to obtain the target reflow soldering attention features of the third level. For specific content, please refer to Figure 4 The content shown;

[0099] Secondly, the second branch included in the second association mining unit can be used to perform multiple levels of self-attention processing on the target component image features to obtain corresponding multiple levels of target component attention features, wherein the input features of the self-attention processing of the latter level are the output features of the self-attention processing of the previous level, and the input features of the self-attention processing of the first level are the target component image features, as described above;

[0100] Then, the third branch included in the second association mining unit can be used to perform multiple-level association mining on the target component attention features of the multiple levels based on the target reflow soldering attention features of the multiple levels to obtain the corresponding target component association features of the multiple levels, wherein, in the association mining of the first level, based on the association relationship between the target reflow soldering attention features of the last level and the target component attention features of the last level, the target component attention features are updated to form the target component association features of the first level, and in the association mining of each level after the first level, the target component attention features of the corresponding level are updated with the target component association features of the previous level. Fusion to form a corresponding fusion feature, and, based on the correlation relationship between the target reflow soldering attention feature of the corresponding level and the fusion feature, the fusion feature is updated to form the target component association feature of the corresponding level; for example, in the association mining of the second level, the target component attention feature of the second-to-last level and the target component association feature of the first level are fused (such as addition or mean calculation, etc.) to form a corresponding fusion feature, and, based on the correlation relationship between the target reflow soldering attention feature of the second-to-last level and the fusion feature, the fusion feature is updated to form the target component association feature of the second level, so that the target component association feature of the third level can be obtained in sequence, etc.;

[0101] Finally, based on the target component association features of the multiple levels, the associated component image features corresponding to the target set-top box component can be obtained; exemplarily, the target component association features of the multiple levels can be fused to obtain the corresponding associated component image features, or the target component association features of the last level can be used as the associated component image features.

[0102] It should be noted that for step S130, the specific method of performing welding quality analysis on the associated component image features corresponding to each set-top box component is not limited and can be selected according to actual needs.

[0103] For example, in an alternative implementation, in order to reliably perform welding quality analysis, the above step S130 may further include the following sub-steps:

[0104] First, the associated component image features can be fully connected to obtain corresponding fully connected features, and the fully connected features can be output processed to obtain corresponding welding quality analysis results. For example, the fully connected features can be processed by the output function included in the target quality analysis network, such as a linear regression function or a classification function, to obtain corresponding welding quality analysis results.

[0105] It should be further explained that, in order to ensure the reliable execution of step S120 and step S130, that is, to ensure that the target quality analysis network can effectively perform association mining and energy demand prediction, the artificial intelligence-based set-top box production line monitoring method may also include the following steps:

[0106] First, sample production line monitoring data and sample reflow data corresponding to the sample production line monitoring data may be obtained, and the sample production line monitoring data may be parsed to obtain local sample production line monitoring data corresponding to each of the plurality of set-top box components, as described above;

[0107] Secondly, in the candidate quality analysis network, for each set-top box component, the set-top box component is taken as a target set-top box component, and, based on the local sample production line monitoring data corresponding to other set-top box components, the local sample production line monitoring data corresponding to the target set-top box component is subjected to first association mining to obtain the sample component image feature corresponding to the target set-top box component, and, based on the sample reflow soldering data, the sample component image feature is subjected to second association mining to obtain the associated sample component image feature corresponding to the target set-top box component;

[0108] Then, in the candidate quality analysis network, welding quality analysis is performed on the associated sample component image features corresponding to each set-top box component to obtain welding quality analysis data corresponding to each set-top box component;

[0109] Finally, the network parameters of the candidate quality analysis network can be updated based on the error between the welding quality analysis data and the corresponding welding quality label (formed by expert annotation or obtained based on other neural network analysis) (such as calculating the cross entropy loss between the welding quality analysis data and the welding quality label) to form a target quality analysis network (for example, the network parameters of the candidate quality analysis network can be updated in the direction of reducing the error so that the error converges, such as the error is reduced to the target value or the error reduction is less than a preset value, thereby using the current candidate quality analysis network as the target quality analysis network, or constructing a corresponding target quality analysis network based on the current network parameters of the candidate quality analysis network), wherein the target quality analysis network is at least used to perform welding quality analysis on the associated component image features corresponding to each set-top box component respectively, to obtain the welding quality analysis results corresponding to each set-top box component, for example, executing the above-mentioned steps S120 and S130.

[0110] Combination Figure 5The embodiment of the present application also provides a set-top box production line monitoring device based on artificial intelligence that can be applied to the above electronic device. The set-top box production line monitoring device based on artificial intelligence can include a data acquisition module, a correlation mining module and a quality analysis module.

[0111] The data acquisition module is used to acquire target production line monitoring data and target reflow data corresponding to the target production line monitoring data, and parse the target production line monitoring data to obtain local production line monitoring data corresponding to each of the multiple set-top box components, wherein the local production line monitoring data includes image data formed by image acquisition of the solder joints of the corresponding set-top box components, and the target reflow data includes at least temperature timing data. In the embodiment of the present application, the data acquisition module can be used to execute Figure 2 As shown in step S110, the relevant contents of the data acquisition module can refer to the above description of step S110.

[0112] The association mining module is used to take each set-top box component as a target set-top box component, and, based on the local production line monitoring data corresponding to other set-top box components, perform a first association mining on the local production line monitoring data corresponding to the target set-top box component to obtain the target component image feature corresponding to the target set-top box component, and, based on the target reflow soldering data, perform a second association mining on the target component image feature to obtain the associated component image feature corresponding to the target set-top box component. In the embodiment of the present application, the association mining module can be used to perform Figure 2 As shown in step S120, the relevant contents of the association mining module can refer to the above description of step S120.

[0113] The quality analysis module is used to perform welding quality analysis on the image features of the associated components corresponding to each set-top box component, and obtain the welding quality analysis results corresponding to each set-top box component. Figure 2 As shown in step S130, the relevant contents of the quality analysis module can refer to the above description of step S130.

[0114] In an embodiment of the present application, corresponding to the above-mentioned artificial intelligence-based set-top box production line monitoring method applied to the electronic device, a computer-readable storage medium is also provided, in which a computer program is stored, and when the computer program is run, each step of the artificial intelligence-based set-top box production line monitoring method is executed.

[0115] Among them, the steps executed when the aforementioned computer program is running will not be described one by one here, and reference may be made to the previous explanation of the artificial intelligence-based set-top box production line monitoring method.

[0116] In summary, the artificial intelligence-based set-top box production line monitoring method, device and equipment provided by the present application first obtain the local production line monitoring data corresponding to each set-top box component; secondly, for each set-top box component, the set-top box component is used as the target set-top box component, and, based on the local production line monitoring data corresponding to other set-top box components, the local production line monitoring data corresponding to the target set-top box component is subjected to a first association mining to obtain the target component image feature, and, based on the target reflow soldering data, the target component image feature is subjected to a second association mining to obtain the associated component image feature; then, the associated component image feature corresponding to each set-top box component is subjected to a welding quality analysis to obtain the corresponding welding quality analysis result. Based on the above content, before performing the welding quality analysis, the internal correlation mining between the target production line monitoring data will be performed on the set-top box components, and the external correlation mining will be performed with the target reflow soldering data, so that the semantics of the obtained associated component image features are richer and the representation ability is higher. Therefore, the reliability of the basis for analysis is higher, thereby ensuring that the reliability of the obtained welding quality analysis results is also higher. In addition, since each set-top box component is analyzed separately, the granularity of the analysis is smaller, and therefore, it has higher accuracy, that is, the reliability of the welding quality analysis results is further improved, thereby improving the problem of relatively low reliability of set-top box production line monitoring in the prior art.

[0117] In several embodiments provided in the embodiments of the present application, it should be understood that the disclosed device and method can also be implemented in other ways. The device and method embodiments described above are merely schematic, for example, the flowcharts and block diagrams in the accompanying drawings show the possible architecture, functions and operations of the device, method and computer program product according to multiple embodiments of the present application. In this regard, each box in the flowchart or block diagram can represent a module, a program segment or a part of a code, and the module, a program segment or a part of a code contains one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions marked in the box can also occur in a different order from the order marked in the accompanying drawings. For example, two consecutive boxes can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each box in the block diagram and / or flowchart, and the combination of boxes in the block diagram and / or flowchart can be implemented with a dedicated hardware-based system that performs a specified function or action, or can be implemented with a combination of dedicated hardware and computer instructions.

[0118] In addition, the functional modules in the various embodiments of the present application may be integrated together to form an independent part, or each module may exist separately, or two or more modules may be integrated to form an independent part.

[0119] If the function is implemented in the form of a software function module and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application is essentially or the part that contributes to the prior art or the part of the technical solution can be embodied in the form of a software product, which is stored in a storage medium, including several instructions to enable a computer device (which can be a personal computer, electronic device, or network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present application. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk and other media that can store program code. It should be noted that in this article, the term "include", "include" or any other variant thereof is intended to cover non-exclusive inclusion, so that the process, method, article or device including a series of elements includes not only those elements, but also includes other elements that are not explicitly listed, or also includes elements inherent to such process, method, article or device. Without more constraints, an element defined by the phrase "comprising a..." does not exclude the existence of other identical elements in the process, method, article or apparatus comprising the element.

[0120] The above description is only the preferred embodiment of the present application and is not intended to limit the present application. For those skilled in the art, the present application may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application shall be included in the protection scope of the present application.

Claims

1. A set-top box production line monitoring method based on artificial intelligence, characterized in that: include: Acquire target production line monitoring data and target reflow soldering data corresponding to the target production line monitoring data, and parse the target production line monitoring data to obtain local production line monitoring data corresponding to each set-top box component of a plurality of set-top box components, wherein the local production line monitoring data includes image data formed by image acquisition of solder joints of the corresponding set-top box components, and the target reflow soldering data includes at least temperature timing data; For each set-top box component, the set-top box component is taken as a target set-top box component, and the local production line monitoring data corresponding to the target set-top box component is convolved to obtain the corresponding first component image convolution feature, and the local production line monitoring data corresponding to each other set-top box component is convolved to obtain the corresponding second component image convolution feature; for each other set-top box component, the first association mining unit corresponding to the other set-top box component in the target quality analysis network is used to perform first association mining on the second component image convolution feature corresponding to the other set-top box component and the first component image convolution feature to obtain the first component image association feature corresponding to the other set-top box component; the first component image association feature corresponding to each other set-top box component is spliced ​​to obtain the corresponding first component image splicing feature, and the first component image splicing feature is converted into a feature space to obtain the corresponding target component image feature; based on the target reflow data, the target component image feature is second associated with the mining to obtain the associated component image feature corresponding to the target set-top box component; The welding quality analysis is performed on the associated component image features corresponding to each set-top box component respectively, and the welding quality analysis result corresponding to each set-top box component is obtained.

2. The method for monitoring a set-top box production line based on artificial intelligence according to claim 1, characterized in that: The step of performing first association mining on the second component image convolution feature corresponding to the other set-top box component and the first component image convolution feature using the first association mining unit corresponding to the other set-top box component in the target quality analysis network for each other set-top box component to obtain the first component image association feature corresponding to the other set-top box component comprises: For each other set-top box component, the second component image convolution feature corresponding to the other set-top box component is loaded into the first association mining unit corresponding to the other set-top box component in the target quality analysis network, and the second component image convolution feature corresponding to the other set-top box component is masked using the masking parameter matrix carried in the first association mining unit to form the corresponding first component image masking feature, wherein the target quality analysis network belongs to a neural network, and the masking parameter matrix is ​​formed by the target quality analysis network in a corresponding training process; An operation of determining an association relationship between the first component image mask feature and the first component image convolution feature is performed, and based on the determined association relationship, the first component image convolution feature is updated to form a corresponding first component image association feature, wherein the semantic information in the first component image association feature is semantic information mined from the first component image convolution feature and has an association relationship with the semantic information in the first component image mask feature.

3. The method for monitoring a set-top box production line based on artificial intelligence according to claim 1, characterized in that: The steps of splicing the first component image associated features corresponding to each other set-top box component to obtain the corresponding first component image splicing features, and performing feature space conversion on the first component image splicing features to obtain the corresponding target component image features include: splicing the first component image associated features corresponding to each other set-top box component to obtain the corresponding first component image splicing features, and performing feature space transformation on the first component image splicing features to obtain the corresponding first component image transformation features, wherein the feature space transformation includes a self-attention operation, a convolution operation, a pooling operation, and a full connection operation; The first component image conversion feature and the first component image convolution feature are fused to form a corresponding target component image feature.

4. The method for monitoring a set-top box production line based on artificial intelligence according to claim 1, characterized in that: The step of performing second association mining on the target component image features based on the target reflow soldering data to obtain the associated component image features corresponding to the target set-top box component includes: Embedding the target reflow soldering data to obtain corresponding target reflow soldering embedding features; Based on the target reflow soldering embedding feature, a second association mining is performed on the target component image feature to obtain an associated component image feature corresponding to the target set-top box component.

5. The method for monitoring a set-top box production line based on artificial intelligence according to claim 4, characterized in that: The step of performing second association mining on the target component image feature based on the target reflow soldering embedding feature to obtain the associated component image feature corresponding to the target set-top box component includes: Utilizing the first branch included in the second association mining unit in the target quality analysis network, the target reflow soldering embedded features are subjected to multiple levels of deep mining to obtain corresponding multiple levels of target reflow soldering deep features, wherein the input features of the deep mining of the latter level are the output features of the deep mining of the previous level, the input features of the deep mining of the first level are the target reflow soldering embedded features, and the deep mining at least includes downsampling; Utilizing the second branch included in the second association mining unit, based on the target reflow depth features of the multiple levels, the target component image features are subjected to multiple levels of feature restoration to obtain corresponding target component image restoration features of the multiple levels, wherein, in the feature restoration of the first level, based on the association relationship between the target reflow depth features of the last level and the target component image features, the target component image features are updated to form the target component image restoration features of the first level, and in the feature restoration of each level after the first level, the target component image restoration features of the previous level are upsampled to obtain corresponding upsampled features, and, based on the association relationship between the target reflow depth features of the corresponding level and the upsampled features, the upsampled features are updated to form the target component image restoration features of the corresponding level; Based on the target component image restoration features of the multiple levels, the associated component image features corresponding to the target set-top box component are obtained.

6. The method for monitoring a set-top box production line based on artificial intelligence according to claim 4, characterized in that: The step of performing second association mining on the target component image feature based on the target reflow soldering embedding feature to obtain the associated component image feature corresponding to the target set-top box component includes: Utilizing the first branch included in the second association mining unit in the target quality analysis network, the target reflow soldering embedded features are subjected to multiple levels of self-attention processing to obtain corresponding multiple levels of target reflow soldering attention features, wherein the input features of the self-attention processing of the latter level are the output features of the self-attention processing of the previous level, and the input features of the self-attention processing of the first level are the target reflow soldering embedded features; Using the second branch included in the second association mining unit, multiple levels of self-attention processing are performed on the target component image features to obtain corresponding multiple levels of target component attention features, wherein the input features of the self-attention processing of the latter level are the output features of the self-attention processing of the previous level, and the input features of the self-attention processing of the first level are the target component image features; Utilizing the third branch included in the second association mining unit, based on the target reflow soldering attention features of the multiple levels, multiple levels of association mining are performed on the target component attention features of the multiple levels to obtain corresponding target component association features of the multiple levels, wherein, in the association mining of the first level, based on the association relationship between the target reflow soldering attention features of the last level and the target component attention features of the last level, the target component attention features are updated to form the target component association features of the first level, and in the association mining of each level after the first level, the target component attention features of the corresponding level and the target component association features of the previous level are fused to form corresponding fused features, and, based on the association relationship between the target reflow soldering attention features of the corresponding level and the fused features, the fused features are updated to form the target component association features of the corresponding level; Based on the target component association features of the multiple levels, the associated component image features corresponding to the target set-top box component are obtained.

7. The method for monitoring a set-top box production line based on artificial intelligence according to any one of claims 1 to 6, characterized in that: The artificial intelligence-based set-top box production line monitoring method also includes: Acquire sample production line monitoring data and sample reflow data corresponding to the sample production line monitoring data, and parse the sample production line monitoring data to obtain local sample production line monitoring data corresponding to each set-top box component in a plurality of set-top box components; In the candidate quality analysis network, for each set-top box component, the set-top box component is taken as a target set-top box component, and based on the local sample production line monitoring data corresponding to other set-top box components, the local sample production line monitoring data corresponding to the target set-top box component is subjected to first association mining to obtain the sample component image feature corresponding to the target set-top box component, and based on the sample reflow data, the sample component image feature is subjected to second association mining to obtain the associated sample component image feature corresponding to the target set-top box component; In the candidate quality analysis network, welding quality analysis is performed on the associated sample component image features corresponding to each set-top box component to obtain welding quality analysis data corresponding to each set-top box component; Based on the error between the welding quality analysis data and the corresponding welding quality label, the network parameters of the candidate quality analysis network are updated to form a target quality analysis network, wherein the target quality analysis network is at least used to perform welding quality analysis on the associated component image features corresponding to each set-top box component, respectively, to obtain the welding quality analysis results corresponding to each set-top box component.

8. A set-top box production line monitoring device based on artificial intelligence, characterized in that: include: a data acquisition module, used to acquire target production line monitoring data and target reflow soldering data corresponding to the target production line monitoring data, and parse the target production line monitoring data to obtain local production line monitoring data corresponding to each set-top box component among a plurality of set-top box components, wherein the local production line monitoring data includes image data formed by image acquisition of solder joints of the corresponding set-top box components, and the target reflow soldering data includes at least temperature timing data; The association mining module is used for, for each set-top box component, taking the set-top box component as a target set-top box component, convolving the local production line monitoring data corresponding to the target set-top box component to obtain a corresponding first component image convolution feature, and convolving the local production line monitoring data corresponding to each other set-top box component to obtain a corresponding second component image convolution feature; for each other set-top box component, using the first association mining unit corresponding to the other set-top box component in the target quality analysis network, performing first association mining on the second component image convolution feature corresponding to the other set-top box component and the first component image convolution feature to obtain a first component image association feature corresponding to the other set-top box component; splicing the first component image association feature corresponding to each other set-top box component to obtain a corresponding first component image splicing feature, and performing feature space conversion on the first component image splicing feature to obtain a corresponding target component image feature; based on the target reflow data, performing second association mining on the target component image feature to obtain an association component image feature corresponding to the target set-top box component; The quality analysis module is used to perform welding quality analysis on the associated component image features corresponding to each set-top box component, and obtain the welding quality analysis result corresponding to each set-top box component.

9. An electronic device, characterized in that: include: Memory for storing computer programs; A processor connected to the memory is used to execute the computer program stored in the memory to implement the artificial intelligence-based set-top box production line monitoring method described in any one of claims 1-7.

Images