Multi-modal data processing method and device, equipment and storage medium

Abstract

The embodiment of the invention provides a multi-modal data processing method and device, equipment and a storage medium. Obtaining multi-modal data; wherein the multi-modal data comprises visual data and text data; inputting the visual data into a plurality of visual encoders for feature extraction to obtain a plurality of visual features; wherein the plurality of visual encoders are different from one another; and inputting the plurality of visual features and the text data into a multi-modal large language model, and outputting a target text. According to the multi-modal data processing method provided by the embodiment of the invention, after feature extraction is performed on the visual data through a plurality of different visual encoders, the extracted visual features are input into the multi-modal large language model for processing, so that the data processing precision of the multi-modal language model can be improved.

Description

Technical Field

[0001] The embodiments of the present disclosure relate to the field of computer technology, and in particular to a method, apparatus, device and storage medium for processing multimodal data. Background Art

[0002] A multi-modal large language model (LLM) refers to a language model that generates text based on multi-modal data (mainly visual data and text data). Compared with language models, multi-modal language models can support more diverse inputs, so they have many applications in content understanding, content summarization, and content creation. Among them, there are certain difficulties in processing visual data. In related technologies, due to the lack of sufficient recognition and processing of visual data, the output results of multi-modal language large models are relatively low. Summary of the invention

[0003] The embodiments of the present disclosure provide a method, apparatus, device and storage medium for processing multimodal data. Feature extraction of visual data is performed by multiple different visual encoders, which can improve the data processing accuracy of a multimodal language model.

[0004] In a first aspect, an embodiment of the present disclosure provides a method for processing multimodal data, including:

[0005] Acquire multimodal data; wherein the multimodal data includes visual data and text data;

[0006] Inputting the visual data into a plurality of visual encoders for feature extraction to obtain a plurality of visual features; wherein the plurality of visual encoders are different from each other;

[0007] The multiple visual features and the text data are input into a multimodal large language model, and a target text is output.

[0008] In a second aspect, the present disclosure also provides a multimodal data processing device, including:

[0009] A multimodal data acquisition module, used to acquire multimodal data; wherein the multimodal data includes visual data and text data;

[0010] A visual feature acquisition module, used for inputting the visual data into a plurality of visual encoders for feature extraction to obtain a plurality of visual features; wherein the plurality of visual encoders are different from each other;

[0011] The target text determination module is used to input the multiple visual features and the text data into the multimodal large language model and output the target text.

[0012] In a third aspect, an embodiment of the present disclosure further provides an electronic device, the electronic device comprising:

[0013] one or more processors;

[0014] a storage device for storing one or more programs,

[0015] When the one or more programs are executed by the one or more processors, the one or more processors implement the method for processing multimodal data as described in the embodiment of the present disclosure.

[0016] In a fourth aspect, an embodiment of the present disclosure further provides a storage medium comprising computer executable instructions, which, when executed by a computer processor, are used to execute the method for processing multimodal data as described in the embodiment of the present disclosure.

[0017] The disclosed embodiments disclose a method, device, equipment and storage medium for processing multimodal data, which obtains multimodal data; wherein the multimodal data includes visual data and text data; the visual data is input into a plurality of visual encoders for feature extraction to obtain a plurality of visual features; wherein the plurality of visual encoders are different from each other; the plurality of visual features and text data are input into a multimodal large language model to output a target text. The method for processing multimodal data provided by the disclosed embodiments extracts features from visual data using a plurality of different visual encoders, and then inputs the extracted visual features into a multimodal large language model for processing, which can improve the data processing accuracy of the multimodal language model. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] The above and other features, advantages and aspects of the embodiments of the present disclosure will become more apparent with reference to the following detailed description in conjunction with the accompanying drawings. Throughout the accompanying drawings, the same or similar reference numerals represent the same or similar elements. It should be understood that the drawings are schematic and the originals and elements are not necessarily drawn to scale.

[0019] Figure 1 It is a flowchart of a method for processing multimodal data provided by an embodiment of the present disclosure;

[0020] Figure 2 is a schematic diagram of a method for processing multimodal data provided by an embodiment of the present disclosure;

[0021] Figure 3 is a schematic diagram of a method for processing multimodal data provided by an embodiment of the present disclosure;

[0022] Figure 4 This is a schematic diagram of a method for processing multimodal data provided by an embodiment of the present disclosure.

[0023] Figure 5is a structural schematic diagram of a multimodal data processing device provided by an embodiment of the present disclosure;

[0024] Figure 6 It is a structural schematic diagram of an electronic device provided by an embodiment of the present disclosure. DETAILED DESCRIPTION

[0025] Embodiments of the present disclosure will be described in more detail below with reference to the accompanying drawings. Although certain embodiments of the present disclosure are shown in the accompanying drawings, it should be understood that the present disclosure can be implemented in various forms and should not be construed as being limited to the embodiments described herein, which are instead provided for a more thorough and complete understanding of the present disclosure. It should be understood that the drawings and embodiments of the present disclosure are only for exemplary purposes and are not intended to limit the scope of protection of the present disclosure.

[0026] It should be understood that the various steps described in the method embodiments of the present disclosure may be performed in different orders and / or in parallel. In addition, the method embodiments may include additional steps and / or omit the steps shown. The scope of the present disclosure is not limited in this respect.

[0027] The term "including" and its variations used herein are open inclusions, i.e., "including but not limited to". The term "based on" means "based at least in part on". The term "one embodiment" means "at least one embodiment"; the term "another embodiment" means "at least one additional embodiment"; the term "some embodiments" means "at least some embodiments". The relevant definitions of other terms will be given in the following description.

[0028] It should be noted that the concepts such as "first" and "second" mentioned in the present disclosure are only used to distinguish different devices, modules or units, and are not used to limit the order or interdependence of the functions performed by these devices, modules or units.

[0029] It should be noted that the modifications of "one" and "plurality" mentioned in the present disclosure are illustrative rather than restrictive, and those skilled in the art should understand that unless otherwise clearly indicated in the context, it should be understood as "one or more".

[0030] The names of the messages or information exchanged between multiple devices in the embodiments of the present disclosure are only used for illustrative purposes and are not used to limit the scope of these messages or information.

[0031] It is understandable that before using the technical solutions disclosed in the embodiments of the present disclosure, the types, scope of use, usage scenarios, etc. of the personal information involved in the present disclosure should be informed to the user and the user's authorization should be obtained in an appropriate manner in accordance with relevant laws and regulations.

[0032] For example, in response to receiving an active request from a user, a prompt message is sent to the user to clearly prompt the user that the operation requested to be performed will require obtaining and using the user's personal information. Thus, the user can autonomously choose whether to provide personal information to software or hardware such as an electronic device, application, server, or storage medium that performs the operation of the technical solution of the present disclosure according to the prompt message.

[0033] As an optional but non-limiting implementation, in response to receiving an active request from the user, the prompt information may be sent to the user in the form of a pop-up window, in which the prompt information may be presented in text form. In addition, the pop-up window may also carry a selection control for the user to choose "agree" or "disagree" to provide personal information to the electronic device.

[0034] It is understandable that the above notification and the process of obtaining user authorization are merely illustrative and do not constitute a limitation on the implementation of the present disclosure. Other methods that meet the relevant laws and regulations may also be applied to the implementation of the present disclosure.

[0035] It is understandable that the data involved in this technical solution (including but not limited to the data itself, the acquisition or use of the data) shall comply with the requirements of relevant laws, regulations and relevant provisions.

[0036] Figure 1 A flowchart of a method for processing multimodal data provided in an embodiment of the present disclosure is provided. The embodiment of the present disclosure is applicable to a situation where multimodal data is input into a large language model for processing. The method can be executed by a multimodal data processing device, which can be implemented in the form of software and / or hardware, and optionally, by an electronic device, which can be a mobile terminal, a PC or a server, etc.

[0037] like Figure 1 As shown, the method includes:

[0038] S110, acquiring multimodal data.

[0039] Among them, the multimodal data includes visual data and text data. The visual data includes static image data and / or dynamic image data, the static image data can be a picture, and the dynamic image data can be a dynamic picture or a video. The text data can be a question-and-answer or query text. In this embodiment, when the user wants to query information through the multimodal large language model, the visual data and text data can be entered in the input box corresponding to the multimodal large language model.

[0040] S120, inputting the visual data into a plurality of visual encoders for feature extraction to obtain a plurality of visual features.

[0041] Among them, the multiple vision encoders are different from each other. The vision encoder can be pre-trained and is a neural network structure that can be used to encode visual data. The encoding process can be understood as a feature extraction process. Multiple vision encoders can be understood as multiple vision encoders with different structures, different distributions of data samples used in training, or different tasks.

[0042] Among them, the different structures of multiple visual encoders can be understood as the different structures of the neural networks that constitute the visual encoders, for example: the network layers included are of different types, different numbers or different orders, etc. The different distributions of data samples used in the training of multiple visual encoders can be understood as: when training each visual encoder, the data sources of the data samples used are different. Exemplarily, when training visual encoder a, the visual sample data comes from a Chinese database; when training visual encoder b, the visual sample data comes from a foreign database. The different tasks of multiple visual encoders can be understood as: after each visual encoder extracts features from the visual data, the tasks or functions of the extracted features are different, for example: they can be used for tasks such as target recognition, image segmentation or style conversion.

[0043] S130, inputting multiple visual features and text data into a multimodal large language model, and outputting a target text.

[0044] Among them, the visual features can be represented by a matrix. The multimodal large language model can be built using an open source neural network structure, and the structure of the multimodal large language model is not limited here. In this embodiment, after obtaining the visual features extracted by each visual encoder, multiple visual features and text data are input into the multimodal large language model, and the multimodal large language model performs a series of processing on the multiple visual features and text data to output the target text.

[0045] In this application scenario, the scale of the visual features output by each visual encoder may be different from the scale required by the multimodal large language model. Therefore, the visual features need to be linearly transformed first and then input into the multimodal large language model.

[0046] Optionally, a plurality of visual features and text data are input into a multimodal large language model, and a method of outputting a target text may be: inputting the plurality of visual features into corresponding visual converters for linear transformation respectively; inputting the plurality of visual features and text data after linear transformation into the multimodal large language model, and outputting the target text.

[0047] Among them, the vision adapter corresponds to the vision encoder one by one. For example, Figure 2 is a schematic diagram of a method for processing multimodal data in this embodiment. Figure 2As shown, the visual data are respectively input into visual encoder 1, visual encoder 2, ..., visual encoder N, and visual feature 1, visual feature 2, ..., visual feature N are output; then visual feature 1 is input into visual converter 1, visual feature 2 is input into visual converter 2, ..., and visual feature N is input into visual converter N to obtain visual feature 1, visual feature 2, ..., visual feature N after linear transformation; finally, visual feature 1, visual feature 2, ..., visual feature N after linear transformation and text data are input into a multimodal large language model to output the target text.

[0048] In this embodiment, as the number of visual encoders increases, the number of visual features input into the multimodal large language model will also increase, which will increase the data processing complexity of the multimodal large language model. Therefore, the visual features or visual encoders can be screened to reduce the amount of visual features input into the multimodal large language model, thereby reducing the complexity of data processing and improving efficiency.

[0049] Optionally, before inputting the visual data into multiple visual encoders for feature extraction, it also includes: inputting the visual data into a first gating module to determine at least one target visual encoder; correspondingly, the method of inputting the visual data into multiple visual encoders for feature extraction can be: inputting the visual data into at least one target visual encoder and outputting at least one visual feature.

[0050] Among them, the first gating module can be a pre-trained neural network model, which is used to screen visual encoders based on visual data. The output of the first gating module is the probability of selecting each visual encoder, and the visual encoder with a probability exceeding a certain value or a certain number of visual encoders with the largest probability is determined as the target visual encoder. Optionally, the output of the first gating module can be in the form of "visual encoder identification + probability". Correspondingly, after obtaining at least one visual feature, the at least one visual feature is respectively input into the corresponding visual converter for linear transformation, and finally the at least one visual feature after linear transformation and the text data are input into the multimodal large language model to output the target text. Exemplarily, Figure 3 is a schematic diagram of a method for processing multimodal data in this embodiment, such as Figure 3 As shown, after the visual data is input into the first gating module, the visual encoder 1 and the visual encoder 2 are determined as the target time encoders, the visual data are input into the visual encoder 1 and the visual encoder 2 respectively, and the visual feature 1 and the visual feature 2 are output; then the visual feature 1 is input into the visual converter 1, and the visual feature 2 is input into the visual converter 2; finally, the linearly transformed visual feature 1, the visual feature 2 and the text data are input into the multimodal large language model, and the target text is output.

[0051] Optionally, based on the above embodiments, the method also includes the following steps: obtaining first multimodal sample data; and training a first gating module, multiple visual converters and a multimodal large language model based on the first multimodal sample data.

[0052] The first multimodal sample data includes first visual sample data and first text sample data. Figure 3 For example, the process of training the first gating module, multiple visual converters and the multimodal large language model based on the first multimodal sample data can be: first, the first visual sample data is input into the first gating module to determine at least one target visual encoder; then the first visual sample data is input into at least one target visual encoder to output at least one visual sample feature; then the at least one visual sample feature is respectively input into the corresponding visual converter for linear transformation, and then the at least one visual sample feature after linear transformation and the first text sample data are input into the multimodal large language model to output the first target text; finally, the loss function is determined based on the first target text to adjust the parameters in the first gating module, the multiple visual converters and the multimodal large language model based on the loss function.

[0053] Optionally, after obtaining multiple visual features, the following steps are also included: inputting the multiple visual features into a second gating module to determine at least one target visual feature; correspondingly, inputting the multiple visual features and text data into a multimodal large language model, and the method of outputting the target text can be: inputting at least one target visual feature and text data into the multimodal large language model, and outputting the target text.

[0054] Among them, the second gating module can be a pre-trained neural network model, which is used for screening based on visual features. The output of the second gating module is the probability of selecting each visual feature, and the visual features with probabilities exceeding a certain value or a certain number of visual features with the largest probability are determined as target visual features. Optionally, the output of the second gating module can be in the form of "visual feature identification + probability". Correspondingly, after determining at least one target visual feature, the at least one target visual feature is respectively input into the corresponding visual converter for linear transformation, and the at least one target visual feature after linear transformation and the text data are input into the multimodal large language model to output the target text. Exemplarily, Figure 4 is a schematic diagram of a method for processing multimodal data in this embodiment, such as Figure 4As shown, the visual data are respectively input into visual encoder 1, visual encoder 2, ..., visual encoder N, and visual feature 1, visual feature 2, ..., visual feature N are output; then visual feature 1, visual feature 2, ..., visual feature N are input into the second gating module for screening, and visual feature 1 and visual feature 2 are determined to be target visual features, and then visual feature 1 is input into visual converter 1, and visual feature 2 is input into visual converter 2; finally, the linearly transformed visual feature 1, visual feature 2 and text data are input into the multimodal large language model, and the target text is output.

[0055] Optionally, based on the above embodiments, the method also includes the following steps: obtaining second multimodal sample data; and training a second gating module, multiple visual converters and a multimodal large language model based on the second multimodal sample data.

[0056] The second multimodal sample data includes second visual sample data and second text sample data. Figure 4 For example, the process of training the second gating module, multiple visual converters and the multimodal large language model based on the second multimodal sample data can be: first, the second visual sample data is input into multiple visual encoders for feature extraction to obtain multiple visual sample features; then the multiple visual sample features are input into the second gating module to determine at least one target visual sample feature; then the at least one target visual sample feature is respectively input into the corresponding visual converter for linear transformation, and the at least one visual sample feature after linear transformation and the second text sample data are input into the multimodal large language model to output the second target text; finally, the loss function is determined based on the second target text to adjust the parameters in the first gating module, the multiple visual converters and the multimodal large language model based on the loss function.

[0057] The technical solution of the disclosed embodiment obtains multimodal data; wherein the multimodal data includes visual data and text data; the visual data is input into a plurality of visual encoders for feature extraction to obtain a plurality of visual features; wherein the plurality of visual encoders are different from each other; the plurality of visual features and text data are input into a multimodal large language model to output a target text. The method for processing multimodal data provided by the disclosed embodiment extracts features from visual data through a plurality of different visual encoders, and then inputs the extracted visual features into a multimodal large language model for processing, which can improve the data processing accuracy of the multimodal language model.

[0058] Figure 5 A schematic diagram of the structure of a multimodal data processing device provided by an embodiment of the present disclosure is shown in FIG. Figure 5 As shown, the device comprises:

[0059] A multimodal data acquisition module 510, used to acquire multimodal data; wherein the multimodal data includes visual data and text data;

[0060] A visual feature acquisition module 520 is used to input visual data into a plurality of visual encoders for feature extraction to obtain a plurality of visual features; wherein the plurality of visual encoders are different from each other;

[0061] The target text determination module 530 is used to input multiple visual features and text data into the multimodal large language model and output the target text.

[0062] Optionally, the target text determination module 530 is further configured to:

[0063] Inputting multiple visual features into corresponding visual converters respectively for linear transformation;

[0064] Multiple visual features and text data after linear transformation are input into a multimodal large language model to output the target text.

[0065] Optionally, a visual encoder screening module is also included, for:

[0066] Inputting visual data into a first gating module to determine at least one target visual encoder;

[0067] Accordingly, the visual feature acquisition module 520 is further used for:

[0068] Visual data is input into at least one target visual encoder, which outputs at least one visual feature.

[0069] Optionally, the method further includes: a first training module, for:

[0070] Acquire first multimodal sample data; wherein the first multimodal sample data includes first visual sample data and first text sample data;

[0071] A first gating module, a plurality of visual converters, and a multimodal large language model are trained based on the first multimodal sample data.

[0072] Optionally, it also includes: a visual feature screening module, used to:

[0073] inputting the plurality of visual features into a second gating module to determine at least one target visual feature;

[0074] Accordingly, the target text determination module 530 is further used for:

[0075] At least one target visual feature and text data are input into a multimodal large language model, and a target text is output.

[0076] Optionally, the method further includes: a second training module for:

[0077] Acquire second multimodal sample data; wherein the second multimodal sample data includes second visual sample data and second text sample data;

[0078] The second gating module, the plurality of visual converters and the multimodal large language model are trained based on the second multimodal sample data.

[0079] Optionally, the visual data includes static image data and / or dynamic image data.

[0080] The multimodal data processing device provided in the embodiments of the present disclosure can execute the multimodal data processing method provided in any embodiment of the present disclosure, and has the corresponding functional modules and beneficial effects of the execution method.

[0081] It is worth noting that the various units and modules included in the above-mentioned device are only divided according to functional logic, but are not limited to the above-mentioned division, as long as the corresponding functions can be achieved; in addition, the specific names of the functional units are only for the convenience of distinguishing each other, and are not used to limit the protection scope of the embodiments of the present disclosure.

[0082] Figure 6 This is a schematic diagram of the structure of an electronic device provided by an embodiment of the present disclosure. Figure 6 , which shows an electronic device (eg, Figure 6 The terminal device in the embodiment of the present disclosure may include but is not limited to mobile terminals such as mobile phones, notebook computers, digital broadcast receivers, PDAs (personal digital assistants), PADs (tablet computers), PMPs (portable multimedia players), vehicle-mounted terminals (such as vehicle-mounted navigation terminals), etc., and fixed terminals such as digital TVs, desktop computers, etc. Figure 6 The electronic device shown is only an example and should not bring any limitation to the functions and scope of use of the embodiments of the present disclosure.

[0083] like Figure 6 As shown, the electronic device 500 may include a processing device (e.g., a central processing unit, a graphics processing unit, etc.) 501, which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) 502 or a program loaded from a storage device 508 into a random access memory (RAM) 503. In the RAM 503, various programs and data required for the operation of the electronic device 500 are also stored. The processing device 501, the ROM 502, and the RAM 503 are connected to each other via a bus 504. An edit / output (I / O) interface 505 is also connected to the bus 504.

[0084] Typically, the following devices may be connected to the I / O interface 505: an input device 506 including, for example, a touch screen, a touch pad, a keyboard, a mouse, a camera, a microphone, an accelerometer, a gyroscope, etc.; an output device 507 including, for example, a liquid crystal display (LCD), a speaker, a vibrator, etc.; a storage device 508 including, for example, a magnetic tape, a hard disk, etc.; and a communication device 509. The communication device 509 may allow the electronic device 500 to communicate with other devices wirelessly or by wire to exchange data. Although Figure 6 The electronic device 500 is shown with various devices, but it should be understood that it is not required to implement or possess all the devices shown. More or fewer devices may be implemented or possessed instead.

[0085] In particular, according to an embodiment of the present disclosure, the process described above with reference to the flowchart can be implemented as a computer software program. For example, an embodiment of the present disclosure includes a computer program product, which includes a computer program carried on a non-transitory computer-readable medium, and the computer program contains program code for executing the method shown in the flowchart. In such an embodiment, the computer program can be downloaded and installed from a network through a communication device 509, or installed from a storage device 508, or installed from a ROM 502. When the computer program is executed by the processing device 501, the above-mentioned functions defined in the method of the embodiment of the present disclosure are executed.

[0086] The names of the messages or information exchanged between multiple devices in the embodiments of the present disclosure are only used for illustrative purposes and are not used to limit the scope of these messages or information.

[0087] The electronic device provided in the embodiment of the present disclosure and the method for processing multimodal data provided in the above embodiment belong to the same inventive concept. For technical details not fully described in this embodiment, reference can be made to the above embodiment, and this embodiment has the same beneficial effects as the above embodiment.

[0088] An embodiment of the present disclosure provides a computer storage medium on which a computer program is stored. When the program is executed by a processor, the method for processing multimodal data provided by the above embodiment is implemented.

[0089] It should be noted that the computer-readable medium disclosed above may be a computer-readable signal medium or a computer-readable storage medium or any combination of the above two. The computer-readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or device, or any combination of the above. More specific examples of computer-readable storage media may include, but are not limited to: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In the present disclosure, a computer-readable storage medium may be any tangible medium containing or storing a program that may be used by or in combination with an instruction execution system, device or device. In the present disclosure, a computer-readable signal medium may include a data signal propagated in a baseband or as part of a carrier wave, in which a computer-readable program code is carried. This propagated data signal may take a variety of forms, including but not limited to an electromagnetic signal, an optical signal, or any suitable combination of the above. The computer readable signal medium may also be any computer readable medium other than a computer readable storage medium, which may send, propagate or transmit a program for use by or in conjunction with an instruction execution system, apparatus or device. The program code contained on the computer readable medium may be transmitted using any suitable medium, including but not limited to: wires, optical cables, RF (radio frequency), etc., or any suitable combination of the above.

[0090] In some embodiments, the client and the server may communicate using any currently known or future developed network protocol such as HTTP (HyperText Transfer Protocol), and may be interconnected with any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include a local area network ("LAN"), a wide area network ("WAN"), an internet (e.g., the Internet), and a peer-to-peer network (e.g., an ad hoc peer-to-peer network), as well as any currently known or future developed network.

[0091] The computer-readable medium may be included in the electronic device, or may exist independently without being incorporated into the electronic device.

[0092] The computer-readable medium carries one or more programs. When the one or more programs are executed by the electronic device, the electronic device:

[0093] The computer-readable medium carries one or more programs. When the one or more programs are executed by the electronic device, the electronic device: obtains multimodal data; wherein the multimodal data includes visual data and text data; inputs the visual data into multiple visual encoders for feature extraction to obtain multiple visual features; wherein the multiple visual encoders are different from each other; inputs the multiple visual features and the text data into a multimodal large language model to output a target text.

[0094] Computer program code for performing the operations of the present disclosure may be written in one or more programming languages ​​or a combination thereof, including, but not limited to, object-oriented programming languages, such as Java, Smalltalk, C++, and conventional procedural programming languages, such as "C" or similar programming languages. The program code may be executed entirely on the user's computer, partially on the user's computer, as a separate software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving a remote computer, the remote computer may be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computer (e.g., through the Internet using an Internet service provider).

[0095] The flow chart and block diagram in the accompanying drawings illustrate the possible architecture, function and operation of the system, method and computer program product according to various embodiments of the present disclosure. In this regard, each square box in the flow chart or block diagram can represent a module, a program segment or a part of a code, and the module, the program segment or a part of the code contains one or more executable instructions for realizing the specified logical function. It should also be noted that in some implementations as replacements, the functions marked in the square box can also occur in a sequence different from that marked in the accompanying drawings. For example, two square boxes represented in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each square box in the block diagram and / or flow chart, and the combination of the square boxes in the block diagram and / or flow chart can be implemented with a dedicated hardware-based system that performs a specified function or operation, or can be implemented with a combination of dedicated hardware and computer instructions.

[0096] The units involved in the embodiments described in the present disclosure may be implemented by software or hardware. The name of a unit does not limit the unit itself in some cases. For example, the first acquisition unit may also be described as a "unit for acquiring at least two Internet Protocol addresses".

[0097] The functions described above herein may be performed at least in part by one or more hardware logic components. For example, without limitation, exemplary types of hardware logic components that may be used include: field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), systems on chips (SOCs), complex programmable logic devices (CPLDs), and the like.

[0098] In the context of the present disclosure, a machine-readable medium may be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, device, or equipment. A machine-readable medium may be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium may include, but is not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, device, or equipment, or any suitable combination of the foregoing. A more specific example of a machine-readable storage medium may include an electrical connection based on one or more lines, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0099] The above description is only a preferred embodiment of the present disclosure and an explanation of the technical principles used. Those skilled in the art should understand that the scope of disclosure involved in the present disclosure is not limited to the technical solutions formed by a specific combination of the above technical features, but should also cover other technical solutions formed by any combination of the above technical features or their equivalent features without departing from the above disclosed concept. For example, the above features are replaced with the technical features with similar functions disclosed in the present disclosure (but not limited to) by each other to form a technical solution.

[0100] In addition, although each operation is described in a specific order, this should not be understood as requiring these operations to be performed in the specific order shown or in a sequential order. Under certain circumstances, multitasking and parallel processing may be advantageous. Similarly, although some specific implementation details are included in the above discussion, these should not be interpreted as limiting the scope of the present disclosure. Some features described in the context of a separate embodiment can also be implemented in a single embodiment in combination. On the contrary, the various features described in the context of a single embodiment can also be implemented in multiple embodiments individually or in any suitable sub-combination mode.

[0101] Although the subject matter has been described in language specific to structural features and / or methodological logical actions, it should be understood that the subject matter defined in the appended claims is not necessarily limited to the specific features or actions described above. On the contrary, the specific features and actions described above are merely example forms of implementing the claims.

Claims

1. A method for processing multimodal data, characterized in that: include: Acquire multimodal data; wherein the multimodal data includes visual data and text data; Inputting the visual data into a plurality of visual encoders for feature extraction to obtain a plurality of visual features; wherein the plurality of visual encoders are different from each other; The multiple visual features and the text data are input into a multimodal large language model, and a target text is output.

2. The method according to claim 1, characterized in that Inputting the plurality of visual features and the text data into a multimodal large language model and outputting a target text comprises: Inputting the plurality of visual features into corresponding visual converters respectively for linear transformation; The linearly transformed multiple visual features and the text data are input into a multimodal large language model to output a target text.

3. The method according to claim 1 or 2, characterized in that: Before the visual data is input into a plurality of visual encoders for feature extraction, the method further includes: inputting the visual data into a first gating module to determine at least one target visual encoder; Accordingly, the visual data is input into a plurality of visual encoders for feature extraction, including: The visual data is input into the at least one target visual encoder, and at least one visual feature is output.

4. The method according to claim 3, characterized in that Also includes: Acquire first multimodal sample data; wherein the first multimodal sample data includes first visual sample data and first text sample data; The first gating module, the plurality of visual converters, and the multimodal large language model are trained based on the first multimodal sample data.

5. The method according to claim 1 or 2, characterized in that: After obtaining multiple visual features, it also includes: inputting the plurality of visual features into a second gating module to determine at least one target visual feature; Accordingly, the plurality of visual features and the text data are input into a multimodal large language model, and a target text is output, including: The at least one target visual feature and the text data are input into a multimodal large language model, and a target text is output.

6. The method according to claim 5, characterized in that Also includes: Acquire second multimodal sample data; wherein the second multimodal sample data includes second visual sample data and second text sample data; The second gating module, the plurality of visual converters, and the multimodal large language model are trained based on the second multimodal sample data.

7. The method according to claim 1, characterized in that The visual data includes static image data and / or dynamic image data.

8. A multimodal data processing device, characterized in that: include: A multimodal data acquisition module, used to acquire multimodal data; wherein the multimodal data includes visual data and text data; A visual feature acquisition module, used for inputting the visual data into a plurality of visual encoders for feature extraction to obtain a plurality of visual features; wherein the plurality of visual encoders are different from each other; The target text determination module is used to input the multiple visual features and the text data into the multimodal large language model and output the target text.

9. An electronic device, characterized in that: The electronic device comprises: one or more processors; a storage device for storing one or more programs, When the one or more programs are executed by the one or more processors, the one or more processors implement the multimodal data processing method as described in any one of claims 1 to 7.

10. A storage medium comprising computer executable instructions, wherein the computer executable instructions are used to execute the multimodal data processing method according to any one of claims 1 to 7 when executed by a computer processor.

Images