Iptv set-top box quality detection method, device, equipment and medium

Abstract

The IPTV set top box quality detection method, device, equipment and medium provided by the application relate to the technical field of set top box quality detection. In the application, first, set top box image data obtained by image acquisition on a target IPTV set top box is acquired; second, a feature extraction model in an information recognition network formed by pre-training is used to perform feature extraction processing on the set top box image data, and output a set top box image semantic vector corresponding to the set top box image data; then, a feature recognition model in the information recognition network is used to perform feature recognition processing on the set top box image semantic vector in combination with reference set top box image data, and output target set top box quality data corresponding to the target IPTV set top box. Based on the above, the problem that the reliability of set top box quality detection is relatively low in the prior art can be improved.

Description

Technical Field

[0001] This application relates to the field of set-top box quality testing technology, and more specifically, to a method, apparatus, equipment, and medium for testing the quality of IPTV set-top boxes. Background Technology

[0002] Injection-molded IPTV (Internet Protocol Television) set-top boxes are neatly arranged in orientation on the printing fixture and placed in the automatic feeding machine. After startup, the machine self-checks for orientation errors. If correct, the automatic feeder moves the product to the next process. The text to be printed is entered and saved in the 3D printing device. This product can be printed in three directions, and the device can print on multiple sides simultaneously. After printing, the product enters the drying area on the production line. In this area, the temperature is set and the conveyor speed is adjusted to determine the drying time. Sufficient drying time is ensured, with no undried or unprinted areas. After drying, the product enters the inspection and monitoring area. First, product information, specifications, printing color, font size, size requirements, and position requirements are set. Relevant information is entered and saved according to product type for easy parameter retrieval later. Then, the system automatically checks the entered information, imports the inspection information, and only qualified products proceed to the next process. Defective products are automatically transferred to an automatic cleaning and repair system. Based on the received defective product information, the surfaces or areas requiring cleaning are wiped and cleaned, and a self-inspection of the cleaned areas is performed. Products awaiting rework undergo a reprinting process, thus ensuring the reliability of the printed information. In existing technologies, relevant information is typically extracted using traditional character recognition technologies (such as OCR (Optical Character Recognition)), and then compared with reference information. However, the inventors have found that traditional OCR has poor tolerance for image noise, low resolution, blur, rotation, or tilt. When image quality is poor, traditional OCR may fail to correctly recognize characters. Therefore, existing technologies often suffer from relatively low reliability in set-top box quality inspection. Summary of the Invention

[0003] In view of this, the purpose of this application is to provide a method, apparatus, equipment and medium for quality testing of IPTV set-top boxes, so as to improve the problem of relatively low reliability of set-top box quality testing in the prior art.

[0004] To achieve the above objectives, this application adopts the following technical solution:

[0005] A method for quality testing of IPTV set-top boxes, comprising:

[0006] The set-top box image data is obtained by image acquisition of the target IPTV set-top box, wherein the set-top box has set-top box related information formed by printing process on the box body;

[0007] The set-top box image data is processed by a feature extraction model in a pre-trained information recognition network to output the set-top box image semantic vector corresponding to the set-top box image data. The information recognition network is a neural network.

[0008] By using the feature recognition model in the information recognition network and combining it with the reference set-top box image data, feature recognition processing is performed on the semantic vector of the set-top box image to output the target set-top box quality data corresponding to the target IPTV set-top box. The reference set-top box image data contains qualified set-top box related information, and the target set-top box quality data is used to reflect whether the set-top box related information on the box body of the target IPTV set-top box is qualified.

[0009] In a preferred embodiment of this application, in the aforementioned IPTV set-top box quality detection method, the step of performing feature extraction processing on the set-top box image data using a feature extraction model in a pre-trained information recognition network to output the corresponding set-top box image semantic vector includes:

[0010] The convolutional units in the feature extraction model of the pre-trained information recognition network are used to perform convolution processing on multiple channels of the set-top box image data to form convolutional vectors for multiple channels of the set-top box image data. The multiple channels of the data include at least red channel data, green channel data, blue channel data, and grayscale channel data.

[0011] Multiply the channel data convolution vector corresponding to the grayscale channel data and the channel data convolution vector corresponding to the red channel data to obtain the corresponding correlation parameter distribution. Based on the correlation parameter distribution, perform weighted calculation on the channel data convolution vector corresponding to the grayscale channel data to form the corresponding first correlation mining vector.

[0012] Multiply the channel data convolution vector corresponding to the grayscale channel data and the channel data convolution vector corresponding to the green channel data to obtain the corresponding correlation parameter distribution. Based on the correlation parameter distribution, perform weighted calculation on the channel data convolution vector corresponding to the grayscale channel data to form the corresponding second correlation mining vector.

[0013] Multiply the channel data convolution vector corresponding to the grayscale channel data and the channel data convolution vector corresponding to the blue channel data to obtain the corresponding correlation parameter distribution. Based on the correlation parameter distribution, perform weighted calculation on the channel data convolution vector corresponding to the grayscale channel data to form the corresponding third correlation mining vector.

[0014] The mean value is calculated or spliced ​​on the channel data convolution vector, the first correlation mining vector, the second correlation mining vector and the third correlation mining vector corresponding to the grayscale image channel data to form the corresponding set-top box image semantic vector.

[0015] In a preferred embodiment of this application, in the above-mentioned IPTV set-top box quality detection method, the step of using convolutional units in the feature extraction model of a pre-trained information recognition network to perform convolution processing on multiple channels of the set-top box image data to form convolutional vectors for multiple channels of the set-top box image data includes:

[0016] The convolutional units in the feature extraction model of the pre-trained information recognition network are used to perform convolution processing on the red channel data, green channel data, and blue channel data of the set-top box image data, respectively, to form the channel data convolution vectors corresponding to the red channel data, the channel data convolution vectors corresponding to the green channel data, and the channel data convolution vectors corresponding to the blue channel data.

[0017] The grayscale channel data of the set-top box image data is binarized to obtain the binarized image data corresponding to the set-top box image data;

[0018] Perform at least one morphological combination operation on the binarized image data to obtain at least one morphological image data corresponding to the binarized image data, wherein each morphological combination operation includes one dilation operation and one erosion operation.

[0019] The grayscale image channel data is convolved by the convolution unit to form a grayscale image convolution vector corresponding to the grayscale image channel data. Each morphological image data is convolved by the convolution unit to form a morphological image convolution vector corresponding to each morphological image data.

[0020] Correlation mining is performed on the grayscale convolution vector corresponding to the grayscale channel data and the morphological image convolution vector corresponding to each of the morphological image data to output the channel data convolution vector corresponding to the grayscale channel data.

[0021] In a preferred embodiment of this application, in the above-mentioned IPTV set-top box quality detection method, the step of performing correlation mining processing on the grayscale convolution vector corresponding to the grayscale channel data and the morphological image convolution vector corresponding to each of the morphological image data, and outputting the channel data convolution vector corresponding to the grayscale channel data, includes:

[0022] For each morphological image data corresponding to a morphological image convolution vector, the morphological image convolution vector is multiplied by the reconstruction image convolution vector corresponding to the grayscale image channel data to obtain the corresponding correlation parameter distribution. Based on the correlation parameter distribution, the reconstruction image convolution vector corresponding to the grayscale image channel data is weighted and calculated to output a morphological image correlation mining vector corresponding to the grayscale image channel data.

[0023] The mean value is calculated or concatenated for each morphological image correlation mining vector corresponding to the grayscale image channel data and the grayscale image convolution vector corresponding to the grayscale image channel data to form the channel data convolution vector corresponding to the grayscale image channel data.

[0024] In a preferred embodiment of this application, in the above-described IPTV set-top box quality detection method, the step of performing feature recognition processing on the semantic vector of the set-top box image using the feature recognition model in the information recognition network, combined with reference set-top box image data, and outputting the target set-top box quality data corresponding to the target IPTV set-top box includes:

[0025] The feature extraction model is used to perform feature extraction processing on the reference set-top box image data, and outputs the reference image semantic vector corresponding to the reference set-top box image data;

[0026] The embedding unit included in the feature recognition model in the information recognition network performs feature embedding processing on the set-top box description text corresponding to qualified set-top box related information in the reference set-top box image data, and outputs the description text embedding vector corresponding to the set-top box description text.

[0027] The feature recognition model includes a correlation mining unit that performs correlation mining on the reference image semantic vector based on the description text embedding vector, and outputs the reference image mining vector corresponding to the reference image semantic vector.

[0028] Based on the vector distance between the reference image mining vector and the set-top box image semantic vector, the target set-top box quality data corresponding to the target IPTV set-top box is determined, so as to complete the feature recognition processing of the set-top box image semantic vector.

[0029] In a preferred embodiment of this application, in the above-mentioned IPTV set-top box quality detection method, the step of performing correlation mining on the reference image semantic vector based on the descriptive text embedding vector using the correlation mining unit included in the feature recognition model, and outputting the reference image mining vector corresponding to the reference image semantic vector, includes:

[0030] The feature recognition model includes a correlation mining unit that performs multiple levels of correlation mining on the reference image semantic vector based on the description text embedding vector, and outputs multiple levels of correlation mining vectors corresponding to the reference image semantic vector.

[0031] The mean or concatenation of multiple deep-related mining vectors corresponding to the semantic vector of the reference image is performed to output the mining vector of the reference image corresponding to the semantic vector of the reference image.

[0032] In a preferred embodiment of this application, in the above-mentioned IPTV set-top box quality detection method, the step of performing multiple depths of correlation mining on the reference image semantic vector based on the descriptive text embedding vector using the correlation mining unit included in the feature recognition model, and outputting multiple depths of correlation mining vectors corresponding to the reference image semantic vector, includes:

[0033] The descriptive text embedding vector and the reference image semantic vector are loaded into the relevance mining unit included in the feature recognition model;

[0034] In the first deep relevance mining, the description text embedding vector and the reference image semantic vector are multiplied to obtain the corresponding relevance parameter distribution. Based on the relevance parameter distribution, the reference image semantic vector is weighted and calculated to output the first deep mining vector.

[0035] In the second depth of relevance mining, the descriptive text embedding vector and the first depth mining vector are convolved to form a second depth convolved descriptive text embedding vector and a convolved reference image semantic vector. The convolved descriptive text embedding vector and the convolved reference image semantic vector are multiplied to obtain the corresponding relevance parameter distribution. The convolved reference image semantic vector is weighted based on the relevance parameter distribution to output the second depth mining vector.

[0036] In the relevance mining at each depth after the second depth, the convolutional descriptive text embedding vector and the depth mining vector of the previous depth are convolved to form the convolutional descriptive text embedding vector and the convolutional reference image semantic vector of the current depth. The convolutional descriptive text embedding vector and the convolutional reference image semantic vector are multiplied to obtain the corresponding relevance parameter distribution. The convolutional reference image semantic vector is weighted based on the relevance parameter distribution to output the current depth mining vector.

[0037] Deconvolution is performed on the depth mining vectors of the second depth and each subsequent depth to obtain multiple vectors of the same size as the first depth mining vector. These multiple vectors and the first depth mining vector are then used as multiple depth correlation mining vectors corresponding to the semantic vector of the reference image.

[0038] This application also provides an IPTV set-top box quality testing device, including:

[0039] The set-top box image acquisition module is used to acquire set-top box image data obtained by image acquisition of a target IPTV set-top box, wherein the target IPTV set-top box has set-top box related information formed by printing process on the box body;

[0040] The feature extraction module is used to perform feature extraction processing on the set-top box image data through a feature extraction model in a pre-trained information recognition network, and output the set-top box image semantic vector corresponding to the set-top box image data, wherein the information recognition network belongs to a neural network;

[0041] The feature recognition module is used to perform feature recognition processing on the semantic vector of the set-top box image by combining the feature recognition model in the information recognition network with the reference set-top box image data, and output the target set-top box quality data corresponding to the target IPTV set-top box. The reference set-top box image data contains qualified set-top box related information, and the target set-top box quality data is used to reflect whether the set-top box related information on the box body of the target IPTV set-top box is qualified.

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

[0043] Memory, used to store computer programs;

[0044] A processor connected to the memory is used to execute the computer program stored in the memory to implement the above-described IPTV set-top box quality testing method.

[0045] Based on the above, this application also provides a computer-readable storage medium storing a computer program that, when executed, performs the various steps of the IPTV set-top box quality detection method described above.

[0046] The IPTV set-top box quality inspection method, apparatus, equipment, and medium provided in this application first acquire set-top box image data obtained by image acquisition of the target IPTV set-top box; second, through a feature extraction model in a pre-trained information recognition network, feature extraction processing is performed on the set-top box image data, outputting the set-top box image semantic vector corresponding to the set-top box image data; then, through a feature recognition model in the information recognition network, combined with reference set-top box image data, feature recognition processing is performed on the set-top box image semantic vector, outputting the target set-top box quality data corresponding to the target IPTV set-top box. Based on the above, on the one hand, the powerful learning ability of neural networks can be utilized to perform reliable feature recognition processing, thereby obtaining reliable target set-top box quality data; on the other hand, since reference set-top box image data is used as an aid during the feature recognition processing, the reliability of feature recognition processing is improved. Therefore, it can effectively ensure that the obtained target set-top box quality data can truly and reliably reflect the actual printing condition of the set-top box, thereby improving the problem of relatively low reliability of set-top box quality inspection in the prior art. Attached Figure Description

[0047] To make the above-mentioned objectives, features and advantages of this application more apparent and understandable, preferred embodiments are described below in detail with reference to the accompanying drawings.

[0048] Figure 1 This is a structural block diagram of an electronic device provided in an embodiment of this application.

[0049] Figure 2 This is a flowchart illustrating the IPTV set-top box quality testing method provided in an embodiment of this application.

[0050] Figure 3 This is a schematic diagram illustrating correlation mining at multiple depths provided in embodiments of this application.

[0051] Figure 4 This is a block diagram of the IPTV set-top box quality testing device provided in an embodiment of this application. Detailed Implementation

[0052] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. The components of the embodiments of this application described and shown in the accompanying drawings can generally be arranged and designed in various different configurations.

[0053] Therefore, the following detailed description of the embodiments of this application provided in the accompanying drawings is not intended to limit the scope of the claimed application, but merely to illustrate selected embodiments of the application. All other embodiments obtained by those skilled in the art based on the embodiments of this application without inventive effort are within the scope of protection of this application.

[0054] like Figure 1 As shown in the figure, this application provides an electronic device. The electronic device may include a memory, a processor, and an IPTV set-top box quality testing device.

[0055] In detail, the memory and the processor are electrically connected directly or indirectly to enable data transmission or interaction. For example, the memory and the processor can be electrically connected via one or more communication buses or signal lines. The IPTV set-top box quality testing device includes at least one software functional module stored in the memory in the form of software or firmware. The processor is used to execute executable computer programs stored in the memory, such as the software functional modules and computer programs included in the IPTV set-top box quality testing device, to implement the IPTV set-top box quality testing method provided in this application embodiment.

[0056] Optionally, the memory may be, but is not limited to, random access memory (RAM), read-only memory (ROM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), etc.

[0057] Furthermore, the processor can be a general-purpose processor, including a central processing unit (CPU), a network processor (NP), a system on chip (SoC), etc.; it can 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, or discrete hardware components.

[0058] Understandable. Figure 1 The structure shown is for illustrative purposes only; the electronic device may also include components that are more advanced than those shown. Figure 1 The more or fewer components shown, or having the same Figure 1 The different configurations shown may include, for example, a communication unit for exchanging information with other devices (such as image acquisition devices).

[0059] Combination Figure 2 This application also provides a method for quality testing of IPTV set-top boxes applicable to the aforementioned electronic devices. The method steps defined in the process of the IPTV set-top box quality testing method can be implemented by the electronic devices.

[0060] The following will be about Figure 2 The specific process shown will be explained in detail.

[0061] Step S110: Obtain set-top box image data obtained by image acquisition of the target IPTV set-top box.

[0062] In this embodiment, the electronic device can acquire set-top box image data obtained by image acquisition of the target IPTV set-top box. For example, an image acquisition device can be used to acquire images of the target IPTV set-top box. The set-top box image data may include one frame or multiple frames, which can be processed separately. For example, one frame may correspond to one surface of the target IPTV set-top box. The target IPTV set-top box has set-top box related information printed on its surface. The specific content of this information is not limited; for example, it may include product information, specifications, and model number.

[0063] Step S120: The set-top box image data is processed by feature extraction model in the pre-trained information recognition network to output the set-top box image semantic vector corresponding to the set-top box image data.

[0064] In this embodiment, after acquiring the set-top box image data, the electronic device can perform feature extraction processing on the set-top box image data using a feature extraction model in a pre-trained information recognition network, and output a set-top box image semantic vector corresponding to the set-top box image data. In other words, the semantic information in the set-top box image data can be mined and represented in vector form. The information recognition network is a neural network.

[0065] Step S130: Using the feature recognition model in the information recognition network and combining it with the reference set-top box image data, perform feature recognition processing on the semantic vector of the set-top box image, and output the target set-top box quality data corresponding to the target IPTV set-top box.

[0066] In this embodiment, after obtaining the set-top box image semantic vector, the electronic device can perform feature recognition processing on the set-top box image semantic vector by combining the feature recognition model in the information recognition network with reference set-top box image data, and output the target set-top box quality data corresponding to the target IPTV set-top box. The reference set-top box image data contains qualified set-top box related information (for example, an image acquisition operation can be performed on a qualified IPTV set-top box to obtain the corresponding reference set-top box image data). The target set-top box quality data reflects whether the set-top box related information on the target IPTV set-top box is qualified. For example, if the content, color, font, size, and position of the set-top box related information in the set-top box image data are consistent with the content, color, font, size, and position of the set-top box related information in the reference set-top box image data, it is determined to be qualified; otherwise, it is determined to be unqualified. In other words, during the training process of the information recognition network, the corresponding label is only qualified if the content, color, font, size, and position of the set-top box related information in the sample set-top box image data are consistent with the content, color, font, size, and position of the set-top box related information in the reference set-top box image data; otherwise, the corresponding label is unqualified.

[0067] Based on the above, on the one hand, the powerful learning ability of neural networks can be utilized to perform reliable feature recognition processing, thereby obtaining reliable target set-top box quality data; on the other hand, since reference set-top box image data is also used as an aid during the feature recognition process, the reliability of feature recognition processing is improved. Therefore, it can effectively ensure that the obtained target set-top box quality data can truly and reliably reflect the actual printing situation of the set-top box, thereby improving the problem of relatively low reliability of set-top box quality detection in the existing technology.

[0068] It should be noted that the specific method for feature extraction processing of the set-top box image data in step S120 is not limited and can be selected according to actual needs.

[0069] For example, in an alternative implementation, the set-top box image data can be directly convolved, and the semantic vector obtained from the convolution process can be used as the set-top box image semantic vector corresponding to the set-top box image data. In this way, the efficiency of feature extraction processing can be improved to a certain extent, and the corresponding amount of computation can be reduced.

[0070] For example, in another alternative implementation, in order to improve the reliability of feature extraction processing and enable the obtained set-top box image semantic vector to fully and reliably represent the semantic information in the set-top box image data, the above step S120 may further include steps S121, S122, S123, S124 and S125, as detailed below.

[0071] Step S121: The convolution units in the feature extraction model of the pre-trained information recognition network are used to perform convolution processing on multiple channels of the set-top box image data to form convolution vectors of multiple channels of the set-top box image data.

[0072] In this embodiment, the convolutional units in the feature extraction model of a pre-trained information recognition network can be used to perform convolution processing on multiple channels of the set-top box image data, forming multiple channel data convolution vectors corresponding to the set-top box image data. These multiple channels include at least red, green, blue, and grayscale channels, resulting in four channel data convolution vectors. Since there are specific requirements for the color of the printed text, mining each color channel separately allows for the inclusion of more detailed information, thereby improving the reliability of subsequent recognition.

[0073] Step S122: Multiply the channel data convolution vector corresponding to the grayscale channel data and the channel data convolution vector corresponding to the red channel data to obtain the corresponding correlation parameter distribution. Based on the correlation parameter distribution, perform weighted calculation on the channel data convolution vector corresponding to the grayscale channel data to form the corresponding first correlation mining vector.

[0074] In this embodiment, after obtaining the corresponding channel data convolution vectors, the channel data convolution vectors corresponding to the grayscale channel data and the red channel data can be multiplied to obtain the corresponding correlation parameter distribution. Based on this correlation parameter distribution, the channel data convolution vectors corresponding to the grayscale channel data are weighted to form the corresponding first correlation mining vector. It should be noted that the channel data convolution vectors corresponding to the grayscale channel data actually represent the semantic information of the entire printed character. Thus, the semantic information corresponding to the red channel data can be integrated into the semantic information of the entire printed character, allowing for the representation of the overall semantic information while also considering the semantic information of the red channel data, thereby improving the semantic representation capability. Furthermore, the correlation parameter distribution can refer to the dot product between the transpose of the channel data convolution vector corresponding to the grayscale channel data and the channel data convolution vector corresponding to the red channel data. The first correlation mining vector can refer to the result of a weighted summation of the channel data convolution vectors corresponding to the grayscale channel data based on this dot product, thus enabling the fusion of different semantic information.

[0075] Step S123: Multiply the channel data convolution vector corresponding to the grayscale channel data and the channel data convolution vector corresponding to the green channel data to obtain the corresponding correlation parameter distribution. Based on the correlation parameter distribution, perform weighted calculation on the channel data convolution vector corresponding to the grayscale channel data to form the corresponding second correlation mining vector.

[0076] In this embodiment of the application, after obtaining the corresponding channel data convolution vector, the channel data convolution vector corresponding to the grayscale channel data and the channel data convolution vector corresponding to the green channel data can be multiplied to obtain the corresponding correlation parameter distribution. Based on the correlation parameter distribution, the channel data convolution vector corresponding to the grayscale channel data is weighted to form the corresponding second correlation mining vector. The specific processing procedure is as described above.

[0077] Step S124: Multiply the channel data convolution vector corresponding to the grayscale channel data and the channel data convolution vector corresponding to the blue channel data to obtain the corresponding correlation parameter distribution. Based on the correlation parameter distribution, perform weighted calculation on the channel data convolution vector corresponding to the grayscale channel data to form the corresponding third correlation mining vector.

[0078] In this embodiment of the application, after obtaining the corresponding channel data convolution vector, the channel data convolution vector corresponding to the grayscale channel data and the channel data convolution vector corresponding to the blue channel data can be multiplied to obtain the corresponding correlation parameter distribution. Based on the correlation parameter distribution, the channel data convolution vector corresponding to the grayscale channel data is weighted to form the corresponding third correlation mining vector. The specific processing procedure is as described above.

[0079] Step S125: The mean of the channel data convolution vector, the first correlation mining vector, the second correlation mining vector and the third correlation mining vector corresponding to the grayscale image channel data are calculated or spliced ​​to form the corresponding set-top box image semantic vector.

[0080] In this embodiment of the application, after fusing the first correlation mining vector, the second correlation mining vector, and the third correlation mining vector respectively, the mean calculation or splicing processing can be performed on the channel data convolution vector corresponding to the grayscale image channel data, the first correlation mining vector, the second correlation mining vector, and the third correlation mining vector to form the corresponding set-top box image semantic vector. In this way, the set-top box image semantic vector can focus on more important information, thereby improving the accuracy of semantic information representation.

[0081] It is understood that in step S121 above, the specific method of performing convolution processing on multiple channels of the set-top box image data is not limited and can be selected according to actual needs. For example, in an alternative implementation, in order to enable the obtained channel data convolution vector to pay more attention to and represent the semantic information of characters, thereby improving the reliability of subsequent processing, step S121 above may further include steps S121a, S121b, S121c, S121d and S121e, the specific contents of which are as follows.

[0082] Step S121a: The convolutional units in the feature extraction model of the pre-trained information recognition network are used to perform convolution processing on the red channel data, green channel data and blue channel data of the set-top box image data respectively, to form the channel data convolution vector corresponding to the red channel data, the channel data convolution vector corresponding to the green channel data and the channel data convolution vector corresponding to the blue channel data.

[0083] In this embodiment, the red, green, and blue channel data of the set-top box image data can be convolved by the convolutional units in the feature extraction model of the pre-trained information recognition network, respectively, to form convolutional vectors for the red, green, and blue channel data. For example, considering that different color channel data represent different important information, the convolutional unit can include multiple convolutional sub-units, allowing the red, green, and blue channel data to be convolved separately by these sub-units, thus improving the accuracy of the convolutional processing.

[0084] Step S121b: Binarize the grayscale channel data of the set-top box image data to obtain the binarized image data corresponding to the set-top box image data.

[0085] In this embodiment, the grayscale channel data of the set-top box image data can also be binarized to obtain binarized image data corresponding to the set-top box image data. For example, the pixel values ​​in the binarized image data can be either 0 and 1, or 0 and 255; there are no specific limitations, and the appropriate selection can be made according to actual needs.

[0086] Step S121c: Perform at least one morphological combination operation on the binarized image data to obtain at least one morphological image data corresponding to the binarized image data.

[0087] In this embodiment, after obtaining the binarized image data, at least one morphological combination operation can be performed on the binarized image data to obtain at least one morphological image data corresponding to the binarized image data. Each morphological combination operation includes one dilation operation and one erosion operation. The size of the dilation kernel corresponding to the dilation operation can be larger than the size of the erosion kernel corresponding to the erosion operation; for example, the size of the dilation kernel can be 5*5, and the size of the erosion kernel can be 3*3. Furthermore, when performing multiple morphological combination operations, they can be performed separately. This avoids the problem of excessive image data distortion caused by cascading operations (i.e., the object of a later morphological combination operation is the result of a previous morphological combination operation). Moreover, to ensure that more information can be obtained through multiple morphological combination operations, at least one of the sizes of the dilation kernel and the erosion kernel corresponding to every two morphological combination operations is different.

[0088] Step S121d: The grayscale image channel data is convolved by the convolution unit to form a grayscale image convolution vector corresponding to the grayscale image channel data. Each morphological image data is convolved by the convolution unit to form a morphological image convolution vector corresponding to each morphological image data.

[0089] In this embodiment, the grayscale image channel data can be convolved using the convolution unit to form grayscale convolution vectors corresponding to the grayscale image channel data. Furthermore, each morphological image data can be convolved using the convolution unit to form a morphological image convolution vector corresponding to each morphological image data. It should be noted that since different image data represent different key information, various convolution processes can be implemented using different convolution sub-units within the convolution unit. Moreover, for ease of subsequent processing, the vectors output by each convolution sub-unit can have the same size.

[0090] Step S121e: Perform correlation mining processing on the grayscale convolution vector corresponding to the grayscale channel data and the morphological image convolution vector corresponding to each of the morphological image data, and output the channel data convolution vector corresponding to the grayscale channel data.

[0091] In this embodiment, after obtaining the grayscale image convolution vector and the morphological image convolution vector, correlation mining processing can be performed on the grayscale image convolution vector corresponding to each grayscale image channel data and the morphological image convolution vector corresponding to each morphological image data, outputting the channel data convolution vector corresponding to the grayscale image channel data. That is, the information in the morphological image convolution vector corresponding to each morphological image data can be fused into the grayscale image convolution vector corresponding to the grayscale image channel data.

[0092] It is understood that in step S121e above, the specific method for performing correlation mining processing on the grayscale convolution vectors corresponding to the grayscale channel data and the morphological image convolution vectors corresponding to each morphological image data is not limited and can be selected according to actual needs. For example, in an alternative implementation, in order to enable the formed channel data convolution vectors to have higher accuracy in semantic representation, step S121e above can further include the following:

[0093] First, for each morphological image data corresponding to the morphological image convolution vector, the morphological image convolution vector is multiplied with the restoration image convolution vector corresponding to the grayscale image channel data to obtain the corresponding correlation parameter distribution. Then, based on the correlation parameter distribution, the restoration image convolution vector corresponding to the grayscale image channel data is weighted and calculated to output a morphological image correlation mining vector corresponding to the grayscale image channel data. The specific processing procedure is as described above.

[0094] Secondly, the mean calculation or splicing process can be performed on each morphological image correlation mining vector corresponding to the grayscale image channel data and the grayscale image convolution vector corresponding to the grayscale image channel data to form the channel data convolution vector corresponding to the grayscale image channel data.

[0095] It should be noted that for step S130, the specific method of performing feature recognition processing on the semantic vector of the set-top box image in conjunction with the reference set-top box image data is not limited and can be selected according to actual needs.

[0096] For example, in an alternative implementation, feature extraction can be performed on the reference set-top box image data to obtain a corresponding semantic vector. Then, the vector distance between this semantic vector and the semantic vector of the set-top box image can be calculated. Based on this vector distance, it can be determined whether the data is qualified. If the vector distance is greater than a preset distance, it can be determined as unqualified; conversely, if the vector distance is less than or equal to the preset distance, it can be determined as qualified. This improves the efficiency of feature recognition processing.

[0097] For example, in another alternative implementation, in order to ensure the reliability of feature recognition processing, the above step S130 may further include steps S131, S132, S133 and S134, the specific contents of each step are as follows.

[0098] Step S131: Using the feature extraction model, perform feature extraction processing on the reference set-top box image data and output the reference image semantic vector corresponding to the reference set-top box image data.

[0099] In this embodiment of the application, the feature extraction model can be used to perform feature extraction processing on the reference set-top box image data and output the reference image semantic vector corresponding to the reference set-top box image data. The specific processing process can refer to the processing process of the set-top box image data described above, and will not be repeated here.

[0100] Step S132: Through the embedding unit included in the feature recognition model in the information recognition network, feature embedding processing is performed on the set-top box description text corresponding to qualified set-top box related information in the reference set-top box image data, and the description text embedding vector corresponding to the set-top box description text is output.

[0101] In this embodiment of the application, the embedding unit included in the feature recognition model in the information recognition network can also be used to perform feature embedding processing on the set-top box description text corresponding to qualified set-top box related information in the reference set-top box image data, and output the description text embedding vector corresponding to the set-top box description text. For example, each word in the set-top box description text can be embedded using a word embedding model to obtain the corresponding word embedding vector. Then, the mean of each word embedding vector can be calculated or concatenated to obtain the corresponding description text embedding vector.

[0102] Step S133: Using the correlation mining unit included in the feature recognition model, correlation mining is performed on the semantic vector of the reference image based on the embedding vector of the descriptive text, and the reference image mining vector corresponding to the semantic vector of the reference image is output.

[0103] In this embodiment, after obtaining the descriptive text embedding vector and the reference image semantic vector, the relevance mining unit included in the feature recognition model can perform relevance mining on the reference image semantic vector based on the descriptive text embedding vector, and output a reference image mining vector corresponding to the reference image semantic vector. In this way, the semantic information in the descriptive text embedding vector can be fused into the reference image semantic vector, thereby forming a reference image mining vector that fully represents the information content itself.

[0104] Step S134: Based on the vector distance between the reference image mining vector and the set-top box image semantic vector, determine the target set-top box quality data corresponding to the target IPTV set-top box, so as to complete the feature recognition processing of the set-top box image semantic vector.

[0105] In this embodiment, after obtaining the reference image mining vector, the target set-top box quality data corresponding to the target IPTV set-top box can be determined based on the vector distance (such as cosine distance) between the reference image mining vector and the set-top box image semantic vector, thereby completing the feature recognition processing of the set-top box image semantic vector. For example, if the vector distance is greater than a preset distance, it can be determined as unqualified; if the vector distance is less than or equal to the preset distance, it can be determined as qualified. Alternatively, in other embodiments, the reference image mining vector and the set-top box image semantic vector can be concatenated, and then convolution, pooling, and fully connected processing can be performed. Finally, the result of the fully connected processing is classified and output to obtain the probability of being qualified and the probability of being unqualified. The classification output can be implemented using functions such as softmax.

[0106] It is understood that the specific method for relevance mining of the semantic vector of the reference image in step S133 above is not limited and can be selected according to actual needs. For example, in an alternative implementation, in order to capture more detailed information through relevance mining, step S133 above may further include steps S133a and S133b, the specific contents of each step of which are described below.

[0107] Step S133a: Using the correlation mining unit included in the feature recognition model, the semantic vector of the reference image is subjected to multiple depths of correlation mining based on the description text embedding vector, and multiple depths of correlation mining vectors corresponding to the semantic vector of the reference image are output.

[0108] In this embodiment of the application, the relevance mining unit included in the feature recognition model can perform relevance mining at multiple depths on the semantic vector of the reference image based on the embedding vector of the descriptive text, and output multiple deep relevance mining vectors corresponding to the semantic vector of the reference image. In this way, semantic information at different depths can be captured, thereby improving the semantic representation capability.

[0109] Step S133b involves averaging or concatenating multiple deep-related mining vectors corresponding to the semantic vector of the reference image to output the mining vector of the reference image corresponding to the semantic vector of the reference image.

[0110] In this embodiment of the application, after obtaining multiple deep-related mining vectors, the multiple deep-related mining vectors corresponding to the reference image semantic vector can be averaged or concatenated to output the reference image mining vector corresponding to the reference image semantic vector.

[0111] It is understood that in step S133a above, the specific method of performing multiple depths of relevance mining on the semantic vector of the reference image is not limited and can be selected according to actual needs. For example, in an alternative implementation, in order to capture more related details in the process of fusing the semantic information in the embedded vector of the descriptive text into the semantic vector of the reference image, step S133a above can further include the following:

[0112] First, the descriptive text embedding vector and the reference image semantic vector can be loaded into the relevance mining unit included in the feature recognition model. The specific processing procedure is as follows: Figure 4 As shown;

[0113] Secondly, in the third depth of relevance mining, the description text embedding vector and the reference image semantic vector are multiplied to obtain the corresponding relevance parameter distribution. Based on the relevance parameter distribution, the reference image semantic vector is weighted and calculated to output the first depth mining vector. The specific processing procedure is as described above.

[0114] Then, in the second depth of relevance mining, the descriptive text embedding vector and the first depth mining vector are convolved to form a second depth convolved descriptive text embedding vector and a convolved reference image semantic vector (the size after convolution can be the same). The convolved descriptive text embedding vector and the convolved reference image semantic vector are multiplied to obtain the corresponding relevance parameter distribution. The convolved reference image semantic vector is weighted based on the relevance parameter distribution to output the second depth mining vector.

[0115] Subsequently, in the relevance mining at each depth after the second depth, the convolutional descriptive text embedding vector and the depth mining vector of the previous depth are convolved to form the convolutional descriptive text embedding vector and the convolutional reference image semantic vector of the current depth. Then, the convolutional descriptive text embedding vector and the convolutional reference image semantic vector are multiplied to obtain the corresponding relevance parameter distribution. Finally, the convolutional reference image semantic vector is weighted based on the relevance parameter distribution to output the current depth mining vector.

[0116] Finally, deconvolution is performed on the depth mining vectors of the second depth and each subsequent depth to obtain multiple vectors with the same size as the first depth mining vector. These multiple vectors and the first depth mining vector are then used as multiple depth correlation mining vectors corresponding to the semantic vector of the reference image.

[0117] Finally, it should be noted that in the above implementation, when concatenating the vectors, pooling can be performed on the concatenated vectors to achieve compression.

[0118] Combination Figure 4 This application also provides an IPTV set-top box quality testing device applicable to the aforementioned electronic devices. The IPTV set-top box quality testing device may include a set-top box image acquisition module, a feature extraction module, and a feature recognition module.

[0119] The set-top box image acquisition module is used to acquire set-top box image data obtained by image acquisition of a target IPTV set-top box, wherein the target IPTV set-top box has set-top box related information formed by a printing process on its body. In this embodiment of the application, the set-top box image acquisition module can be used to perform... Figure 2 The relevant content regarding the set-top box image acquisition module in step S110 can be found in the preceding description of step S110.

[0120] The feature extraction module is used to perform feature extraction processing on the set-top box image data using a feature extraction model in a pre-trained information recognition network, and output a set-top box image semantic vector corresponding to the set-top box image data. The information recognition network is a neural network. In this embodiment, the feature extraction module can be used to perform... Figure 2 The details of step S120 shown above, and the relevant content regarding the feature extraction module, can be found in the preceding description of step S120.

[0121] The feature recognition module is used to perform feature recognition processing on the semantic vector of the set-top box image using the feature recognition model in the information recognition network and in combination with reference set-top box image data, and output target set-top box quality data corresponding to the target IPTV set-top box. The reference set-top box image data contains qualified set-top box related information, and the target set-top box quality data reflects whether the set-top box related information on the target IPTV set-top box is qualified. In this embodiment, the feature recognition module can be used to perform... Figure 2 The relevant content regarding the feature recognition module in step S130 shown can be found in the previous description of step S130.

[0122] In this embodiment of the application, corresponding to the above-described IPTV set-top box quality testing method applied to the electronic device, a computer-readable storage medium is also provided. This computer-readable storage medium stores a computer program, which, when executed, performs each step of the IPTV set-top box quality testing method. The steps executed by the aforementioned computer program are not described in detail here, but can be found in the preceding explanation of the IPTV set-top box quality testing method.

[0123] In summary, the IPTV set-top box quality detection method, apparatus, equipment, and medium provided in this application first acquire set-top box image data obtained by image acquisition of the target IPTV set-top box; second, through a feature extraction model in a pre-trained information recognition network, feature extraction processing is performed on the set-top box image data to output the set-top box image semantic vector corresponding to the set-top box image data; then, through a feature recognition model in the information recognition network, combined with reference set-top box image data, feature recognition processing is performed on the set-top box image semantic vector to output the target set-top box quality data corresponding to the target IPTV set-top box. Based on the above, on the one hand, the powerful learning ability of neural networks can be utilized to perform reliable feature recognition processing, thereby obtaining reliable target set-top box quality data; on the other hand, since reference set-top box image data is used as an aid during the feature recognition process, the reliability of feature recognition processing is improved. Therefore, it can effectively ensure that the target set-top box quality data can truly and reliably reflect the actual situation of the set-top box, thereby improving the problem of relatively low reliability of set-top box quality detection in the prior art.

[0124] In the several embodiments provided in this application, it should be understood that the disclosed apparatus and methods can also be implemented in other ways. The apparatus and method embodiments described above are merely illustrative. For example, the flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of apparatus, methods, and computer program products according to various embodiments of this application. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing 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 blocks may occur in a different order than those marked in the drawings. For example, two consecutive blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in a block diagram and / or flowchart, and combinations of blocks in block diagrams and / or flowcharts, can be implemented using a dedicated hardware-based system that performs the specified function or action, or using a combination of dedicated hardware and computer instructions.

[0125] In addition, the functional modules in the various embodiments of this application can be integrated together to form an independent part, or each module can exist independently, or two or more modules can be integrated to form an independent part.

[0126] If the aforementioned functions are implemented as software functional modules and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, electronic device, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks. It should be noted that, in this document, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. In the absence of further restrictions, an element defined by the phrase "comprising a..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0127] The above description is merely a preferred embodiment of this application and is not intended to limit this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the protection scope of this application.

Claims

1. A method for quality inspection of IPTV set-top boxes, characterized in that, include: The set-top box image data is obtained by image acquisition of the target IPTV set-top box, wherein the set-top box has set-top box related information formed by printing process on the box body; The set-top box image data is processed by a feature extraction model in a pre-trained information recognition network to output the set-top box image semantic vector corresponding to the set-top box image data. The information recognition network is a neural network. The feature extraction model is used to extract features from the reference set-top box image data, outputting a reference image semantic vector corresponding to the reference set-top box image data. The embedding unit included in the feature recognition model of the information recognition network is used to embed features into the set-top box description text corresponding to qualified set-top box related information in the reference set-top box image data, outputting a description text embedding vector corresponding to the set-top box description text. The description text embedding vector and the reference image semantic vector are loaded into the relevance mining unit included in the feature recognition model. In the first depth of relevance mining, the description text embedding vector and the reference image semantic vector are multiplied to obtain a... The relevance parameter distribution is obtained, and the reference image semantic vector is weighted based on the relevance parameter distribution to output the first depth mining vector; in the relevance mining at the second depth, the descriptive text embedding vector and the first depth mining vector are convolved to form the second depth convolved descriptive text embedding vector and the convolved reference image semantic vector, and the convolved descriptive text embedding vector and the convolved reference image semantic vector are multiplied to obtain the corresponding relevance parameter distribution, and the convolved reference image semantic vector is weighted based on the relevance parameter distribution to output the second depth mining vector; the relevance mining at each depth after the second depth is performed. In this process, the convolutional descriptive text embedding vector and the depth mining vector from the previous depth are convolved to form the convolutional descriptive text embedding vector and the convolutional reference image semantic vector for the current depth. The convolutional descriptive text embedding vector and the convolutional reference image semantic vector are then multiplied to obtain the corresponding relevance parameter distribution. Based on this relevance parameter distribution, the convolutional reference image semantic vector is weighted to output the current depth mining vector. Deconvolution is then performed on the depth mining vectors for each depth from the second depth onwards to obtain multiple vectors of the same size as the first depth mining vector. These multiple vectors and the first depth mining vector are then used as the parameters... The system considers multiple deep correlation mining vectors corresponding to the semantic vector of the reference image; it performs mean calculation or concatenation processing on the multiple deep correlation mining vectors corresponding to the semantic vector of the reference image to output the reference image mining vector; based on the vector distance between the reference image mining vector and the semantic vector of the set-top box image, it determines the target set-top box quality data corresponding to the target IPTV set-top box, so as to complete the feature recognition processing of the semantic vector of the set-top box image. The reference set-top box image data contains qualified set-top box related information, and the target set-top box quality data is used to reflect whether the set-top box related information on the box body of the target IPTV set-top box is qualified.

2. The method for quality inspection of IPTV set-top boxes according to claim 1, characterized in that, The step of using a feature extraction model in a pre-trained information recognition network to extract features from the set-top box image data and output the corresponding set-top box image semantic vector includes: The convolutional units in the feature extraction model of the pre-trained information recognition network are used to perform convolution processing on multiple channels of the set-top box image data to form convolutional vectors for multiple channels of the set-top box image data. The multiple channels of the data include at least red channel data, green channel data, blue channel data, and grayscale channel data. Multiply the channel data convolution vector corresponding to the grayscale channel data and the channel data convolution vector corresponding to the red channel data to obtain the corresponding correlation parameter distribution. Based on the correlation parameter distribution, perform weighted calculation on the channel data convolution vector corresponding to the grayscale channel data to form the corresponding first correlation mining vector. Multiply the channel data convolution vector corresponding to the grayscale channel data and the channel data convolution vector corresponding to the green channel data to obtain the corresponding correlation parameter distribution. Based on the correlation parameter distribution, perform weighted calculation on the channel data convolution vector corresponding to the grayscale channel data to form the corresponding second correlation mining vector. Multiply the channel data convolution vector corresponding to the grayscale channel data and the channel data convolution vector corresponding to the blue channel data to obtain the corresponding correlation parameter distribution. Based on the correlation parameter distribution, perform weighted calculation on the channel data convolution vector corresponding to the grayscale channel data to form the corresponding third correlation mining vector. The mean value is calculated or spliced ​​on the channel data convolution vector, the first correlation mining vector, the second correlation mining vector and the third correlation mining vector corresponding to the grayscale image channel data to form the corresponding set-top box image semantic vector.

3. The method for quality inspection of IPTV set-top boxes according to claim 2, characterized in that, The step of using convolutional units in the feature extraction model of the pre-trained information recognition network to perform convolution processing on multiple channels of the set-top box image data to form convolutional vectors for the multiple channels of the set-top box image data includes: The convolutional units in the feature extraction model of the pre-trained information recognition network are used to perform convolution processing on the red channel data, green channel data, and blue channel data of the set-top box image data, respectively, to form the channel data convolution vectors corresponding to the red channel data, the channel data convolution vectors corresponding to the green channel data, and the channel data convolution vectors corresponding to the blue channel data. The grayscale channel data of the set-top box image data is binarized to obtain the binarized image data corresponding to the set-top box image data; Perform at least one morphological combination operation on the binarized image data to obtain at least one morphological image data corresponding to the binarized image data, wherein each morphological combination operation includes one dilation operation and one erosion operation. The grayscale image channel data is convolved by the convolution unit to form a grayscale image convolution vector corresponding to the grayscale image channel data. Each morphological image data is convolved by the convolution unit to form a morphological image convolution vector corresponding to each morphological image data. Correlation mining is performed on the grayscale convolution vector corresponding to the grayscale channel data and the morphological image convolution vector corresponding to each of the morphological image data to output the channel data convolution vector corresponding to the grayscale channel data.

4. The IPTV set-top box quality inspection method according to claim 3, characterized in that, The step of performing correlation mining processing on the grayscale convolution vector corresponding to the grayscale channel data and the morphological image convolution vector corresponding to each of the morphological image data, and outputting the channel data convolution vector corresponding to the grayscale channel data, includes: For each morphological image data corresponding to a morphological image convolution vector, the morphological image convolution vector is multiplied by the reconstruction image convolution vector corresponding to the grayscale image channel data to obtain the corresponding correlation parameter distribution. Based on the correlation parameter distribution, the reconstruction image convolution vector corresponding to the grayscale image channel data is weighted and calculated to output a morphological image correlation mining vector corresponding to the grayscale image channel data. The mean value is calculated or concatenated for each morphological image correlation mining vector corresponding to the grayscale image channel data and the grayscale image convolution vector corresponding to the grayscale image channel data to form the channel data convolution vector corresponding to the grayscale image channel data.

5. A quality testing device for IPTV set-top boxes, characterized in that, include: The set-top box image acquisition module is used to acquire set-top box image data obtained by image acquisition of a target IPTV set-top box, wherein the target IPTV set-top box has set-top box related information formed by printing process on the box body; The feature extraction module is used to perform feature extraction processing on the set-top box image data through a feature extraction model in a pre-trained information recognition network, and output the set-top box image semantic vector corresponding to the set-top box image data, wherein the information recognition network belongs to a neural network; The feature recognition module is used to perform feature extraction processing on the reference set-top box image data through the feature extraction model, and output the reference image semantic vector corresponding to the reference set-top box image data; through the embedding unit included in the feature recognition model in the information recognition network, it performs feature embedding processing on the set-top box description text corresponding to qualified set-top box related information in the reference set-top box image data, and outputs the description text embedding vector corresponding to the set-top box description text; it loads the description text embedding vector and the reference image semantic vector into the relevance mining unit included in the feature recognition model; in the first depth of relevance mining, it loads the description text embedding vector and the reference image semantic vector into the relevance mining unit. Multiplying the quantities yields the corresponding relevance parameter distribution. Based on this relevance parameter distribution, a weighted calculation is performed on the reference image semantic vector to output the first depth mining vector. In the second depth relevance mining, the descriptive text embedding vector and the first depth mining vector are convolved to form a second-depth convolutional descriptive text embedding vector and a convolutional reference image semantic vector. Multiplying these two vectors yields the corresponding relevance parameter distribution. Based on this relevance parameter distribution, a weighted calculation is performed on the convolutional reference image semantic vector to output the second depth mining vector. For each depth after the second depth... In relevance mining, the convolutional descriptive text embedding vector and the depth mining vector from the previous depth are convolved to form the convolutional descriptive text embedding vector and the convolutional reference image semantic vector for the current depth. The convolutional descriptive text embedding vector and the convolutional reference image semantic vector are then multiplied to obtain the corresponding relevance parameter distribution. Based on this relevance parameter distribution, the convolutional reference image semantic vector is weighted to output the current depth mining vector. Deconvolution is then performed on the depth mining vectors for each depth from the second depth onwards to obtain multiple vectors of the same size as the first depth mining vector. These multiple vectors and the first depth mining vector are then used as the... The process involves: generating multiple deep correlation mining vectors corresponding to the semantic vector of the reference image; averaging or concatenating these deep correlation mining vectors to output the reference image mining vector; determining the target set-top box quality data based on the vector distance between the reference image mining vector and the set-top box image semantic vector, thereby completing the feature recognition processing of the set-top box image semantic vector. The reference set-top box image data contains qualified set-top box related information, and the target set-top box quality data reflects whether the set-top box related information on the target IPTV set-top box is qualified.

6. An electronic device, characterized in that, include: Memory, used to store computer programs; A processor connected to the memory is used to execute the computer program stored in the memory to implement the IPTV set-top box quality testing method according to any one of claims 1-4.

7. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed, performs the IPTV set-top box quality testing method according to any one of claims 1-4.

Images