Data transmission method, device and storage medium

Abstract

This invention belongs to the field of data transmission and discloses a data transmission method, device, and storage medium. The method includes: extracting visual features of a target image and projecting the visual features into a latent low-dimensional space to obtain latent low-dimensional features; projecting the latent low-dimensional features into a semantic feature space based on a multi-level semantic encoder to obtain semantic features; performing category determination on the semantic features based on a first semantic knowledge base to obtain semantic errors; determining the target features of the target image based on the semantic errors; and sending the target features to a data receiving device. Because this invention extracts visual features of a target image, determines the target features of the target image based on the semantic errors corresponding to the visual features, and sends the target features to the data receiving device, compared to existing methods that directly send the target image to the data receiving device, the above method of this invention can reduce transmission latency and save transmission bandwidth.

Description

Technical Field

[0001] This invention relates to the field of data transmission technology, and in particular to a data transmission method, device and storage medium. Background Technology

[0002] In typical 6G intelligent scenarios of the future, remote zero-shot semantic understanding presents a widespread challenge. For example, in autonomous driving scenarios, driverless vehicles often rely on wireless communication with distant vehicles or base stations to obtain traffic conditions in advance, thereby optimizing route planning. However, road traffic frequently encounters new situations such as emergencies, accidents, and the introduction of new types of vehicles. Images of these scenarios are often unavailable in advance, invisible during training, and belong to unknown categories. Therefore, autonomous vehicles face the challenge of remote zero-shot image recognition. Traditional recognition methods often rely on large-scale, strongly supervised labeled training datasets or directly sending raw images to address this zero-shot problem. These methods are unsuitable for solving the zero-shot problem and require significant manpower, resources, and time, which contradicts the requirements of low latency, lightweight, and efficient intelligent communication.

[0003] The above content is only used to help understand the technical solution of the present invention and does not represent an admission that the above content is prior art. Summary of the Invention

[0004] The main objective of this invention is to provide a data transmission method, device, and storage medium, which aims to solve the technical problem of the inability to identify remote zero-sample images in the prior art.

[0005] To achieve the above objectives, the present invention provides a data transmission method applied to a data transmission device, the method comprising the following steps:

[0006] Visual features of the target image are extracted and projected into a latent low-dimensional space to obtain latent low-dimensional features;

[0007] Based on a multi-level semantic encoder, the latent low-dimensional features are projected onto the semantic feature space to obtain semantic features;

[0008] Based on the first semantic knowledge base, the semantic features are classified to obtain the semantic error;

[0009] The target features of the target image are determined based on the semantic error, and the target features are sent to the data receiving device.

[0010] Optionally, the step of determining the target features of the target image based on the semantic error and sending the target features to the data receiving device includes:

[0011] When the semantic error is less than a preset threshold, the estimated category information corresponding to the semantic error is used as the target feature of the target image, and the target feature is sent to the data receiving device;

[0012] When the semantic error is greater than or equal to the preset threshold, the semantic feature is used as the target feature of the target image, and the target feature is sent to the data receiving device.

[0013] Optionally, after the step of using the semantic feature as the target feature of the target image and sending the target feature to the data receiving device when the semantic error is greater than or equal to the preset threshold, the method further includes:

[0014] Receive the semantic feature category judgment result fed back by the data receiving device;

[0015] When the semantic error at the receiving end in the semantic feature category judgment result is greater than the preset threshold, the visual features of the target image are sent to the data receiving device.

[0016] Optionally, the estimated category information includes an estimated category identifier and / or an estimated category semantic vector;

[0017] The step of taking the estimated category information corresponding to the semantic error as the target feature of the target image when the semantic error is less than a preset threshold, and sending the target feature to the data receiving device, includes:

[0018] When the semantic error is less than a preset threshold, the estimated category identifier is used as the target feature of the target image, and the target feature is sent to the data receiving device;

[0019] The receiving device sends back the identification results;

[0020] If the estimated category semantic vector corresponding to the estimated category identifier does not exist in the second semantic knowledge base of the data receiving device, the estimated category semantic vector is sent to the data receiving device as the target feature of the target image.

[0021] Furthermore, to achieve the above objectives, the present invention also provides a data transmission method applied to a data receiving device, the method comprising the following steps:

[0022] Target features of the target image sent by the receiving data transmitting device;

[0023] The target image is identified by classifying the target features based on a second semantic knowledge base and / or a multi-level semantic decoder.

[0024] Optionally, the step of classifying the target image based on the target features using a second semantic knowledge base and / or a multi-level semantic decoder includes:

[0025] When the target feature is an estimated category identifier, it is determined whether the estimated category semantic vector corresponding to the estimated category identifier exists in the second semantic knowledge base based on the second semantic knowledge base;

[0026] If it exists, the estimated category semantic vector is obtained from the second semantic knowledge base according to the estimated category identifier, and the target image is identified according to the estimated category semantic vector.

[0027] Optionally, the step of classifying the target image based on the target features using a second semantic knowledge base and / or a multi-level semantic decoder includes:

[0028] When the target feature is a potential low-dimensional feature, the potential low-dimensional feature is projected onto the semantic feature space according to the multi-level semantic decoder to obtain the target semantic feature.

[0029] Based on the second semantic knowledge base, the target semantic features are classified to obtain estimated category information;

[0030] The target image is identified based on the estimated category information.

[0031] Furthermore, to achieve the above objectives, the present invention also proposes a data transmission device, the device comprising: a memory, a processor, and a data transmission program stored in the memory and executable on the processor, the data transmission program being configured to implement the steps of the data transmission method as described above for the data transmission device.

[0032] Furthermore, to achieve the above objectives, the present invention also proposes a data receiving device, characterized in that the device includes: a memory, a processor, and a data transmission program stored in the memory and executable on the processor, the data transmission program being configured to implement the steps of the data transmission method as described above for the data receiving device.

[0033] Furthermore, to achieve the above objectives, the present invention also proposes a storage medium storing a data transmission program, which, when executed by a processor, implements the steps of the data transmission method described above.

[0034] This invention extracts visual features from a target image and projects these visual features into a latent low-dimensional space to obtain latent low-dimensional features. Based on a multi-level semantic encoder, these latent low-dimensional features are projected into a semantic feature space to obtain semantic features. A category decision is made on these semantic features based on a first semantic knowledge base to obtain semantic errors. The target features of the target image are determined based on these semantic errors, and then sent to a data receiving device. Because this invention extracts visual features from a target image, determines the target features based on the semantic errors corresponding to these visual features, and then sends these target features to the data receiving device, compared to existing methods that directly send the target image to the data receiving device, this invention reduces data transmission latency and saves transmission bandwidth. Attached Figure Description

[0035] Figure 1 This is a schematic diagram of the structure of the data transmission device in the hardware operating environment involved in the embodiments of the present invention;

[0036] Figure 2 This is a flowchart illustrating the first embodiment of the data transmission method of the present invention;

[0037] Figure 3 This is a flowchart illustrating the second embodiment of the data transmission method of the present invention;

[0038] Figure 4 This is a flowchart illustrating the third embodiment of the data transmission method of the present invention;

[0039] Figure 5 This is a structural block diagram of the first embodiment of the data transmission device of the present invention;

[0040] Figure 6 This is a structural block diagram of a second embodiment of the data transmission device of the present invention.

[0041] The realization of the objective, functional features and advantages of the present invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0042] It should be understood that the specific embodiments described herein are for illustrative purposes only and are not intended to limit the scope of the invention.

[0043] Reference Figure 1 , Figure 1 This is a schematic diagram of the data transmission device structure of the hardware operating environment involved in the embodiments of the present invention.

[0044] like Figure 1As shown, the data transmission device may include: a processor 1001, such as a central processing unit (CPU), a communication bus 1002, a user interface 1003, a network interface 1004, and a memory 1005. The communication bus 1002 is used to enable communication between these components. The user interface 1003 may include a display screen or an input unit such as a keyboard; optionally, the user interface 1003 may also include a standard wired interface or a wireless interface. The network interface 1004 may optionally include a standard wired interface or a wireless interface (such as a Wireless-Fidelity (Wi-Fi) interface). The memory 1005 may be high-speed random access memory (RAM) or stable non-volatile memory (NVM), such as a disk drive. The memory 1005 may also optionally be a storage device independent of the aforementioned processor 1001.

[0045] Those skilled in the art will understand that Figure 1 The structure shown does not constitute a limitation on the data transmission device and may include more or fewer components than shown, or combine certain components, or have different component arrangements.

[0046] like Figure 1 As shown, the memory 1005, which serves as a storage medium, may include an operating system, a network communication module, a user interface module, and a data transmission program.

[0047] exist Figure 1 In the data transmission device shown, the network interface 1004 is mainly used for data communication with the network server; the user interface 1003 is mainly used for data interaction with the user; the processor 1001 and the memory 1005 in the data transmission device of the present invention can be set in the data transmission device, and the data transmission device calls the data transmission program stored in the memory 1005 through the processor 1001 and executes the data transmission method provided in the embodiment of the present invention.

[0048] Based on the aforementioned data transmission device, this embodiment of the invention provides a data transmission method, referring to... Figure 2 , Figure 2 This is a flowchart illustrating the first embodiment of the data transmission method of the present invention.

[0049] In this embodiment, the data transmission method includes the following steps:

[0050] Step S10: Extract the visual features of the target image and project the visual features into a latent low-dimensional space to obtain latent low-dimensional features.

[0051] It should be noted that the executing entity in this embodiment can be a computing service device with data processing, network communication, and program execution functions, such as a mobile phone, tablet computer, or personal computer, or an electronic device or data transmission device capable of performing the above functions. The following description uses the data transmission device as an example to illustrate this embodiment and the subsequent embodiments.

[0052] It should be noted that the target image can be an image to be sent to a data receiving device. The data receiving device can be a device that receives information sent by the data sending device. The visual features can be visual features of the target image extracted by the data sending device based on a pre-trained deep convolutional neural network (CNN), such as GoogleNet or ResNet. Projecting the visual features into a latent low-dimensional space to obtain latent low-dimensional features can be achieved by the data sending device projecting the visual features into a latent low-dimensional space based on a local visual autoencoder, as shown in the following formula:

[0053]

[0054] in, Used to characterize the potential low-dimensional features, t is used to characterize the data transmitting device, m is used to characterize the target image, and P is used to characterize the potential low-dimensional features. t,v V is used to characterize the local visual autoencoder. m R is used to characterize the visual features of the target image, R is used to characterize the set of real numbers, and k is used to characterize the dimension of the potential low-dimensional features.

[0055] Step S20: Based on a multi-level semantic encoder, project the potential low-dimensional features onto the semantic feature space to obtain semantic features.

[0056] It should be noted that the data transmitting device includes a multi-level semantic encoder, which includes a local visual autoencoder and a local semantic autoencoder. The process of projecting the latent low-dimensional features onto the semantic feature space based on the multi-level semantic encoder to obtain semantic features can also be based on the local semantic autoencoder projecting the latent low-dimensional features onto the semantic feature space, as shown in the following formula:

[0057]

[0058] in, Used to characterize the semantic features. The data is used to characterize the potential low-dimensional features, where t represents the data transmitting device and m represents the target image. Used to characterize the local semantic autoencoder.

[0059] Step S30: Based on the first semantic knowledge base, perform category determination on the semantic features to obtain semantic error.

[0060] It should be noted that the first semantic knowledge base may be a knowledge base maintained in the data transmitting device, used to store the semantic vectors of each sample image. The category determination of the semantic features based on the first semantic knowledge base may be performed using the following formula:

[0061]

[0062] in, The semantic vector identifier, hereinafter referred to as the estimated category, is the semantic vector with the smallest error between the semantic vector in the first semantic knowledge base and the semantic feature in the category decision result. arg is used to represent obtaining the semantic vector identifier, min is used to represent taking the minimum value, and S... c B is used to represent semantic vectors in the first semantic knowledge base. t Used to represent the first semantic knowledge base This is used to calculate the error between the semantic vectors in the first semantic knowledge base and the semantic features. The semantic error obtained through category decision can be:

[0063]

[0064] in, The semantic vector used to characterize the semantic feature with the smallest error in the first semantic knowledge base.

[0065] Step S40: Determine the target features of the target image based on the semantic error, and send the target features to the data receiving device.

[0066] It should be noted that determining the target feature of the target image based on the semantic error can refer to determining which dimension of the target image's features are to be sent to the data receiving device based on the semantic error. The target feature may include at least one of the following: the estimated category of the target image, the semantic vector in the first semantic knowledge base with the smallest error to the semantic feature, the semantic feature of the target image, a latent low-dimensional feature, and a visual feature. Specifically, when the semantic error is less than a preset error threshold, the estimated category may be used as the target feature, and the target feature may be sent to the data receiving device.

[0067] This embodiment extracts visual features from the target image and projects these visual features into a latent low-dimensional space to obtain latent low-dimensional features. Based on a multi-level semantic encoder, these latent low-dimensional features are projected into a semantic feature space to obtain semantic features. A category decision is made on the semantic features based on a first semantic knowledge base to obtain semantic errors. The target features of the target image are determined based on the semantic errors, and then the target features are sent to the data receiving device. Since this embodiment extracts visual features from the target image, determines the target features based on the semantic errors corresponding to the visual features, and then sends the target features to the data receiving device, compared to the existing method of directly sending the target image to the data receiving device, the above method in this embodiment can reduce data transmission latency and save transmission bandwidth.

[0068] refer to Figure 3 , Figure 3 This is a flowchart illustrating the second embodiment of the data transmission method of the present invention.

[0069] Based on the first embodiment described above, in this embodiment, step S40 includes:

[0070] Step S401: When the semantic error is less than a preset threshold, the estimated category information corresponding to the semantic error is used as the target feature of the target image, and the target feature is sent to the data receiving device.

[0071] It should be noted that when the semantic error is less than a preset threshold, it indicates that the error between the semantic vector of the target image and the semantic vector corresponding to the estimated category is small, and the target image can be basically identified through the semantic vector corresponding to the estimated category. In this case, the estimated category information corresponding to the estimated category can be directly used as the target feature of the target image, and the target feature can be sent to the data receiving device. The estimated category information corresponding to the semantic error can be the identification information of the estimated category or the semantic vector corresponding to the estimated category.

[0072] Furthermore, to save network bandwidth and improve data transmission efficiency, the estimated category information includes an estimated category identifier and / or an estimated category semantic vector; the step of using the estimated category information corresponding to the semantic error as the target feature of the target image when the semantic error is less than a preset threshold, and sending the target feature to the data receiving device, includes:

[0073] When the semantic error is less than a preset threshold, the estimated category identifier is used as the target feature of the target image, and the target feature is sent to the data receiving device; the recognition result fed back by the data receiving device is received; when the recognition result is that the estimated category semantic vector corresponding to the estimated category identifier does not exist in the second semantic knowledge base of the data receiving device, the estimated category semantic vector is used as the target feature of the target image and sent to the data receiving device.

[0074] It should be noted that, in this embodiment, to improve data transmission efficiency and minimize the amount of data sent, the estimated category identifier is first used as the target feature of the target image, and this target feature is sent to the data receiving device. The data receiving device then searches its local second semantic knowledge base for the semantic vector corresponding to the estimated category identifier, thus completing the transmission of the target image. In this case, the transmission dimension is 1. However, the data receiving device may not have stored the semantic vector corresponding to the estimated category identifier in its local second semantic knowledge base. In this situation, transmitting only the estimated category identifier is insufficient for data transmission. If the data receiving device reports that the estimated category semantic vector corresponding to the estimated category identifier does not exist in its second semantic knowledge base, the data sending device will send the estimated category semantic vector as the target feature of the target image to the data receiving device. This allows the data receiving device to directly identify the target image based on the estimated category semantic vector sent by the data sending device, completing the transmission of the target image. In this case, the dimension to be transmitted is the semantic feature dimension.

[0075] Step S402: When the semantic error is greater than or equal to the preset threshold, the semantic feature is used as the target feature of the target image, and the target feature is sent to the data receiving device.

[0076] It should be noted that when the semantic error is greater than or equal to a preset threshold, it indicates that the semantic vector of the target image has a large error compared with the semantic vector corresponding to the estimated category, and the target image cannot be identified by the semantic vector corresponding to the estimated category. In this case, the data transmitting device uses the semantic feature as the target feature of the target image and sends the target feature to the data receiving device.

[0077] Furthermore, to avoid the data receiving device failing to recognize or making a large error in recognizing the target image based on the semantic features, step S402 further includes: receiving the semantic feature category judgment result fed back by the data receiving device; and when the receiving end semantic error in the semantic feature category judgment result is greater than the preset threshold, sending the visual features of the target image to the data receiving device.

[0078] It should be noted that when the data receiving device receives the semantic features, it will perform a category determination on the semantic features based on the second semantic knowledge base on the data receiving device's local machine. Based on the category determination result, it will determine whether the target image can be identified through the semantic features. The data receiving device can refer to the following formula to perform the semantic feature category determination:

[0079]

[0080] in, This is used to characterize the semantic vector whose error with the semantic feature is minimized in the semantic feature category decision result of the data receiving device, from the semantic vector in the second semantic knowledge base. `arg` is used to characterize the acquisition of the semantic vector identifier, `min` is used to characterize taking the minimum value, and `S` is used to characterize the minimum value. c B is used to represent semantic vectors in the second semantic knowledge base. r Used to represent a second semantic knowledge base This is used to calculate the error between the semantic vectors in the second semantic knowledge base and the semantic features. The semantic error obtained through semantic feature category determination can be:

[0081]

[0082] in, The semantic vector used to characterize the semantic feature with the smallest semantic error in the second semantic knowledge base.

[0083] It should be noted that when the semantic error at the receiving end in the semantic feature category judgment result of the data receiving device exceeds the preset threshold, it indicates that the data receiving device cannot accurately identify the target image based on the semantic features. In this case, it is necessary to send the visual features of the target image to the data receiving device so that the data receiving device can identify the target image based on its visual features.

[0084] Furthermore, if the semantic error at the receiving end in the semantic feature category decision result is greater than the preset threshold, the latent low-dimensional features of the target image can also be sent to the data receiving device. This allows the data receiving device to identify the target image based on the latent low-dimensional features of the target image.

[0085] In this embodiment, when the semantic error is less than a preset threshold, the estimated category information corresponding to the semantic error is used as the target feature of the target image, and the target feature is sent to the data receiving device; when the semantic error is greater than or equal to the preset threshold, the semantic feature is used as the target feature of the target image, and the target feature is sent to the data receiving device. This embodiment can determine which dimension of the target image's features are sent to the data receiving device based on the error between the semantic vector in the first semantic knowledge base and the semantic feature, thereby improving data transmission efficiency.

[0086] Based on the aforementioned data transmission device, this embodiment of the invention provides a data transmission method, referring to... Figure 4 , Figure 4 This is a flowchart illustrating the third embodiment of the data transmission method of the present invention.

[0087] In this embodiment, the data transmission method includes the following steps:

[0088] Step S50: Receive the target features of the target image sent by the data transmitting device.

[0089] It should be noted that the executing entity in this embodiment can be a computing service device with data processing, network communication, and program execution functions, such as a mobile phone, tablet computer, or personal computer, or an electronic device or data receiving device capable of performing the above functions. The following description uses the data receiving device as an example to illustrate this embodiment and the subsequent embodiments.

[0090] It should be noted that the target image can be an image that the data transmitting device needs to send to the data receiving device, and the target features can include at least one of the following: the estimated category of the target image calculated by the data transmitting device based on the semantic features of the target image, the semantic vector with the smallest error with the semantic features in the first semantic knowledge base, the semantic features of the target image, potential low-dimensional features, and visual features.

[0091] Step S60: Based on the second semantic knowledge base and / or multi-level semantic decoder, classify the target features to identify the target image.

[0092] It should be noted that the classification decision of the target features based on the second semantic knowledge base and / or multi-level semantic decoder, and the identification of the target image, can be based on the type of the target features, and the classification decision and semantic recognition of the target features based on the second semantic knowledge base and / or multi-level semantic decoder, thereby obtaining the target image.

[0093] Furthermore, in order to improve transmission efficiency, step S60 may include: when the target feature is an estimated category identifier, determining whether the estimated category semantic vector corresponding to the estimated category identifier exists in the second semantic knowledge base based on the second semantic knowledge base;

[0094] If it exists, the estimated category semantic vector is obtained from the second semantic knowledge base according to the estimated category identifier, and the target image is identified according to the estimated category semantic vector.

[0095] It should be noted that, to improve data transmission efficiency, the data transmitting device sends an estimated category identifier to the data receiving device. This allows the data receiving device to extract the semantic vector corresponding to the estimated category identifier from the second semantic knowledge base, and then obtain the target image based on the semantic vector. If the estimated category identifier is not present in the second semantic knowledge base of the data receiving device, it can report that the semantic vector corresponding to the estimated category identifier does not exist in the second semantic knowledge base. This allows the data transmitting device to directly send the semantic vector corresponding to the estimated category identifier, i.e., the estimated category semantic vector, to the data receiving device, which then directly identifies the target image based on the estimated category semantic vector.

[0096] Furthermore, when the target feature is a potential low-dimensional feature, the potential low-dimensional feature is projected onto the semantic feature space according to the multi-level semantic decoder to obtain the target semantic feature; the target semantic feature is classified based on the second semantic knowledge base to obtain estimated category information; and the target image is identified according to the estimated category information.

[0097] It should be noted that the multi-level semantic decoder may include a local semantic autoencoder and a local visual autoencoder of the data receiving device, used to project potential low-dimensional features onto a semantic feature space, as shown in the following formula:

[0098]

[0099] in, Used to characterize the target semantic features. The data receiver is used to characterize the potential low-dimensional features, where r represents the data receiving device and m represents the target image. Used to characterize the local semantic autoencoder.

[0100] The step of performing category determination on the target semantic features based on the second semantic knowledge base to obtain estimated category information can refer to the step in the above embodiment where the data receiving device performs category determination on the semantic features based on the second semantic knowledge base local to the data receiving device when receiving the semantic features. This embodiment will not repeat the steps here.

[0101] It should be noted that when the target feature is a semantic feature, the steps of the data receiving device in the above embodiment, which perform category determination on the semantic feature based on the second semantic knowledge base of the data receiving device local to the data receiving device, can be referred to.

[0102] Furthermore, when the target feature is a visual feature, the visual feature can be projected to a latent low-dimensional feature by the local visual autoencoder of the data receiving device, and then processed according to the steps described above when the target feature is a latent low-dimensional feature.

[0103] This embodiment receives target features of a target image sent by a data transmitting device; it then performs category determination on the target features based on a second semantic knowledge base and / or a multi-level semantic decoder to identify the target image. This improves data transmission efficiency and solves the problem of remote zero-sample data transmission.

[0104] Reference Figure 5 , Figure 5 This is a structural block diagram of the first embodiment of the data transmission device of the present invention.

[0105] like Figure 5 As shown, the data transmission device proposed in this embodiment of the invention includes:

[0106] The extraction module 10 is used to extract the visual features of the target image and project the visual features into a latent low-dimensional space to obtain latent low-dimensional features;

[0107] Projection module 20 is used to project the latent low-dimensional features onto the semantic feature space based on a multi-level semantic encoder to obtain semantic features;

[0108] The category decision module 30 is used to make a category decision on the semantic features based on the first semantic knowledge base to obtain the semantic error;

[0109] The sending module 40 is used to determine the target features of the target image based on the semantic error, and send the target features to the data receiving device.

[0110] This embodiment extracts visual features from the target image and projects these visual features into a latent low-dimensional space to obtain latent low-dimensional features. Based on a multi-level semantic encoder, these latent low-dimensional features are projected into a semantic feature space to obtain semantic features. A category decision is made on the semantic features based on a first semantic knowledge base to obtain semantic errors. The target features of the target image are determined based on the semantic errors, and then the target features are sent to the data receiving device. Since this embodiment extracts visual features from the target image, determines the target features based on the semantic errors corresponding to the visual features, and then sends the target features to the data receiving device, compared to the existing method of directly sending the target image to the data receiving device, the above method in this embodiment can reduce data transmission latency and save transmission bandwidth.

[0111] Reference Figure 6 , Figure 6 This is a structural block diagram of a second embodiment of the data transmission device of the present invention.

[0112] like Figure 6 As shown, the data transmission device proposed in this embodiment of the invention includes:

[0113] The receiving module 50 is used to receive the target features of the target image sent by the data transmitting device;

[0114] The recognition module 60 is used to make a category judgment on the target features based on the second semantic knowledge base and / or a multi-level semantic decoder, and to recognize the target image.

[0115] This embodiment receives target features of a target image sent by a data transmitting device; it then performs category determination on the target features based on a second semantic knowledge base and / or a multi-level semantic decoder to identify the target image. This improves data transmission efficiency and solves the problem of remote zero-sample data transmission.

[0116] It should be noted that the workflow described above is merely illustrative and does not limit the scope of protection of this invention. In practical applications, those skilled in the art can select some or all of the workflow to achieve the purpose of this embodiment according to actual needs, and no restrictions are imposed here.

[0117] In addition, for technical details not described in detail in this embodiment, please refer to the data transmission method provided in any embodiment of the present invention, which will not be repeated here.

[0118] Other embodiments or specific implementations of the data transmission device of the present invention can be referred to the above-described method embodiments, and will not be repeated here.

[0119] Furthermore, embodiments of the present invention also propose a storage medium storing a data transmission program, which, when executed by a processor, implements the steps of the data transmission method described above.

[0120] 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 system 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 system. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or system that includes that element.

[0121] The sequence numbers of the above embodiments of the present invention are for descriptive purposes only and do not represent the superiority or inferiority of the embodiments.

[0122] Through the above description of the embodiments, those skilled in the art can clearly understand that the methods of the above embodiments can be implemented by means of software plus necessary general-purpose hardware platforms. Of course, they can also be implemented by hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product is stored in a storage medium (such as read-only memory / random access memory, magnetic disk, optical disk) and includes several instructions to cause a terminal device (which may be a mobile phone, computer, server, air conditioner, or network device, etc.) to execute the methods described in the various embodiments of the present invention.

[0123] The above are merely preferred embodiments of the present invention and do not limit the scope of the patent. Any equivalent structural or procedural transformations made based on the description and drawings of the present invention, or direct or indirect applications in other related technical fields, are similarly included within the scope of patent protection of the present invention.

Claims

1. A data transmission method, characterized in that, Applied to a data transmission device, the data transmission method includes the following steps: Visual features of the target image are extracted and projected into a latent low-dimensional space to obtain latent low-dimensional features; Based on a multi-level semantic encoder, the latent low-dimensional features are projected onto the semantic feature space to obtain semantic features; Based on the first semantic knowledge base, the semantic features are classified to obtain the semantic error; The target features of the target image are determined based on the semantic error, and the target features are sent to the data receiving device; The step of determining the target features of the target image based on the semantic error and sending the target features to the data receiving device includes: When the semantic error is less than a preset threshold, the estimated category information corresponding to the semantic error is used as the target feature of the target image, and the target feature is sent to the data receiving device; When the semantic error is greater than or equal to the preset threshold, the semantic feature is used as the target feature of the target image, and the target feature is sent to the data receiving device.

2. The data transmission method as described in claim 1, characterized in that, After the step of using the semantic feature as the target feature of the target image and sending the target feature to the data receiving device when the semantic error is greater than or equal to the preset threshold, the method further includes: Receive the semantic feature category judgment result fed back by the data receiving device; When the semantic error at the receiving end in the semantic feature category judgment result is greater than the preset threshold, the visual features of the target image are sent to the data receiving device.

3. The data transmission method as described in claim 1, characterized in that, The estimated category information includes an estimated category identifier and / or an estimated category semantic vector; The step of taking the estimated category information corresponding to the semantic error as the target feature of the target image when the semantic error is less than a preset threshold, and sending the target feature to the data receiving device, includes: When the semantic error is less than a preset threshold, the estimated category identifier is used as the target feature of the target image, and the target feature is sent to the data receiving device; The receiving device sends back the identification results; If the estimated category semantic vector corresponding to the estimated category identifier does not exist in the second semantic knowledge base of the data receiving device, the estimated category semantic vector is sent to the data receiving device as the target feature of the target image.

4. A data transmission method, applied to the data transmission method according to any one of claims 1-3, characterized in that, Applied to a data receiving device, the data transmission method includes the following steps: Target features of the target image sent by the receiving data transmitting device; The target image is identified by classifying the target features based on a second semantic knowledge base and / or a multi-level semantic decoder. The step of classifying the target image based on the target features using a second semantic knowledge base and / or a multi-level semantic decoder, and identifying the target image, includes: When the target feature is an estimated category identifier, it is determined whether the estimated category semantic vector corresponding to the estimated category identifier exists in the second semantic knowledge base based on the second semantic knowledge base; If it exists, the estimated category semantic vector is obtained from the second semantic knowledge base according to the estimated category identifier, and the target image is identified according to the estimated category semantic vector.

5. The data transmission method as described in claim 4, characterized in that, The step of classifying the target image based on the target features using a second semantic knowledge base and / or a multi-level semantic decoder, and identifying the target image, includes: When the target feature is a potential low-dimensional feature, the potential low-dimensional feature is projected onto the semantic feature space according to the multi-level semantic decoder to obtain the target semantic feature. Based on the second semantic knowledge base, the target semantic features are classified to obtain estimated category information; The target image is identified based on the estimated category information.

6. A data transmission device, characterized in that, The device includes: a memory, a processor, and a data transfer program stored in the memory and executable on the processor, the data transfer program being configured to implement the steps of the data transfer method as described in any one of claims 1 to 3.

7. A data receiving device, characterized in that, The device includes: a memory, a processor, and a data transfer program stored in the memory and executable on the processor, the data transfer program being configured to implement the steps of the data transfer method as described in any one of claims 4 to 5.

8. A storage medium, characterized in that, The storage medium stores a data transmission program, which, when executed by a processor, implements the steps of the data transmission method as described in any one of claims 1 to 3 or 4 to 5.

Images