Text-driven action generation method and device, equipment, storage medium and program product

Abstract

The invention discloses a text-driven action generation method, a text-driven action generation device, text-driven action generation equipment, a storage medium and a computer program product, and relates to the technical field of artificial intelligence. In the application, firstly, an input text and a null motion sequence are input into an autoregressive motion converter; obtaining a basic motion sequence; then, sequentially processing the basic motion sequence through an encoder and a quantizer in the vector quantization variational auto-encoder to obtain a residual index; thirdly, inputting the input text, the basic motion sequence and the residual index into a non-autoregressive motion converter to obtain a residual motion sequence; and finally, processing the residual motion sequence through a decoder in the vector quantization variational auto-encoder to obtain a target motion sequence.

Description

Technical Field

[0001] The present application relates to the technical field of artificial intelligence, and in particular to a text-driven action generation method, a text-driven action generation apparatus, a text-driven action generation device, a storage medium, and a computer program product. Background Art

[0002] In the fields of digital human video live broadcast, virtual special effects production, and game character driving, human motion synthesis technology can provide strong support for the dynamic performance of 3D characters, effectively improving the realism and smoothness of movement.

[0003] In the current action generation technology, text-driven action generation methods provide an efficient way to generate actions. In practical applications, if the data shows an obvious linear relationship and the computing resources are limited, an autoregressive model may be used to achieve text-driven action generation. If the data is complex and requires parallel computing, or if long sequence data needs to be processed, a non-autoregressive model can be used to achieve text-driven action generation.

[0004] However, the accuracy of current text-driven action generation methods needs to be improved. When using an autoregressive model to achieve text-driven action generation, it assumes that the generation process of the time series is linear, which makes it difficult to capture nonlinear characteristics; when using a non-autoregressive model to achieve text-driven action generation, its model complexity is greater than that of an autoregressive model, and its prediction accuracy is lower than that of an autoregressive model.

[0005] The above contents are only used to assist in understanding the technical solution of the present application and do not constitute an admission that the above contents are prior art. Summary of the invention

[0006] The main purpose of the present application is to provide a text-driven action generation method, a text-driven action generation device, a text-driven action generation equipment, a storage medium and a computer program product, aiming to solve the technical problem of poor accuracy of text-driven action generation.

[0007] To achieve the above objectives, the present application proposes a text-driven action generation method, the text-driven action generation method comprising: Input the input text and the empty motion sequence into the autoregressive motion transformer to obtain the basic motion sequence; The basic motion sequence is processed by an encoder and a quantizer in a vector quantized variational autoencoder in sequence to obtain a residual index; Inputting the input text, the basic motion sequence and the residual index into a non-autoregressive motion transformer to obtain a residual motion sequence; The residual motion sequence is processed by the decoder in the vector quantization variational autoencoder to obtain a target motion sequence.

[0008] In one embodiment, the step of sequentially processing the base motion sequence through an encoder and a quantizer in a vector quantized variational autoencoder to obtain a residual index includes: Adding the inter-frame motion speed loss and the acceleration loss to the reconstruction loss function of the vector quantized variational autoencoder to be trained to obtain a new reconstruction loss function; Based on the new reconstruction loss function, a vector quantized variational autoencoder is trained.

[0009] In one embodiment, before the step of sequentially processing the base motion sequence through an encoder and a quantizer in a vector quantization variational autoencoder to obtain a residual index, the step further includes: During training of the vector quantized variational autoencoder, different motions are mapped to corresponding vector quantization layers, where the number of vector quantization layers is the number of quantizers determined by the uniform sampling quantizer.

[0010] In one embodiment, the step of obtaining a target motion sequence by processing the residual motion sequence through a decoder in the vector quantized variational autoencoder comprises: Adding a sparse attention mask to the residual motion sequence to obtain a new residual motion sequence; The new residual motion sequence is processed by the decoder in the vector quantization variational autoencoder to obtain a target motion sequence.

[0011] In one embodiment, the step of inputting the input text and the empty motion sequence into the autoregressive motion transformer to obtain the basic motion sequence comprises: In the latent space, based on the text embedding conditions extracted from the input text, the distribution of possible next indexes is predicted through a motion autoregressive transformer and sub-motion sequences are generated one by one until a complete sequence is obtained, and the complete sequence is used as the basic motion sequence.

[0012] In one embodiment, the step of inputting the input text, the basic motion sequence and the residual index into a non-autoregressive motion transformer to obtain a residual motion sequence comprises: The text embedding conditions extracted from the input text, the basic motion sequence and the residual index as the residual layer indicator of the quantization layer are embedded as the input of the non-autoregressive motion transformer, and the residual motion sequence of the residual quantization layer is predicted by the non-autoregressive motion transformer.

[0013] In addition, to achieve the above-mentioned purpose, the present application also proposes a text-driven action generation device, the text-driven action generation device comprising: An autoregressive module, used for inputting input text and an empty motion sequence into an autoregressive motion transformer to obtain a basic motion sequence; An encoding and quantization module, used for sequentially processing the basic motion sequence through an encoder and a quantizer in a vector quantization variational autoencoder to obtain a residual index; A non-autoregressive module, used for inputting the input text, the basic motion sequence and the residual index into a non-autoregressive motion transformer to obtain a residual motion sequence; The decoding module is used to process the residual motion sequence through the decoder in the vector quantization variational autoencoder to obtain a target motion sequence.

[0014] In addition, to achieve the above-mentioned purpose, the present application also proposes a text-driven action generation device, which includes: a memory, a processor, and a computer program stored in the memory and executable on the processor, and the computer program is configured to implement the steps of the text-driven action generation method described above.

[0015] In addition, to achieve the above-mentioned purpose, the present application also proposes a storage medium, which is a computer-readable storage medium, and a computer program is stored on the storage medium. When the computer program is executed by a processor, the steps of the text-driven action generation method described above are implemented.

[0016] In addition, to achieve the above-mentioned purpose, the present application also provides a computer program product, which includes a computer program, and when the computer program is executed by a processor, the steps of the text-driven action generation method described above are implemented.

[0017] One or more technical solutions proposed in this application have at least the following technical effects: In the present application, first, the input text and the empty motion sequence are input into the autoregressive motion transformer to obtain the basic motion sequence; then, the basic motion sequence is processed by the encoder and quantizer in the vector quantized variational autoencoder in turn to obtain the residual index; then, the input text, the basic motion sequence and the residual index are input into the non-autoregressive motion transformer to obtain the residual motion sequence; finally, the residual motion sequence is processed by the decoder in the vector quantized variational autoencoder to obtain the target motion sequence. Thus, the accuracy of text-driven action generation is further improved by combining the advantages of the autoregressive motion transformer using the historical data of the motion sequence itself to predict the future sequence, the vector quantized variational autoencoder capturing and representing richer and more complex data distributions in the discrete latent space, and the non-autoregressive motion transformer reducing the error accumulation problem caused by sequential dependence. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the present application and, together with the description, serve to explain the principles of the present application.

[0019] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, for ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0020] Figure 1 A flowchart diagram of the first embodiment of the text-driven action generation method of the present application is provided; Figure 2 A schematic diagram of double-layer motion modeling generation provided in Example 1 of the text-driven motion generation method of the present application; Figure 3 A schematic diagram of residual vector motion quantization provided in Embodiment 1 of the text-driven motion generation method of the present application; Figure 4 A schematic diagram of sliding focus sparse attention provided for Example 1 of the text-driven action generation method of this application; Figure 5 This is a schematic diagram of the module structure of the text-driven action generation device according to an embodiment of the present application; Figure 6 Schematic diagram of the device structure of the hardware operating environment involved in the text-driven action generation method in the embodiment of the present application.

[0021] The purpose, features and advantages of this application will be further described in conjunction with the embodiments and with reference to the accompanying drawings. DETAILED DESCRIPTION

[0022] It should be understood that the specific embodiments described herein are only used to explain the technical solutions of the present application and are not used to limit the present application.

[0023] In order to better understand the technical solution of the present application, a detailed description will be given below in conjunction with the accompanying drawings and specific implementation methods.

[0024] It should be noted that the execution subject of this embodiment can be a computing service device with data processing, network communication and program running functions, such as a tablet computer, a personal computer, a mobile phone, etc., or an electronic device capable of realizing the above functions, a text-driven action generation device, etc. The following takes a text-driven action generation device as an example to illustrate this embodiment and the following embodiments.

[0025] Based on this, the embodiment of the present application provides a text-driven action generation method, referring to Figure 1 , Figure 1 This is a flow chart of the first embodiment of the text-driven action generation method of the present application.

[0026] In this embodiment, the text-driven action generation method includes steps S10 to S40: Step S10, inputting the input text and the empty motion sequence into the autoregressive motion transformer to obtain a basic motion sequence; In one embodiment, referring to Figure 2 , input the input text and the empty motion sequence into the autoregressive motion transformer to obtain the basic motion sequence. Among them, autoregressive movement is a prediction method based on historical data. It assumes that the current value depends on the past value and is usually used to predict periodic time series data. In the autoregressive motion model, the current value of the time series is a linear combination of its own past values, plus an error term.

[0027] In a feasible implementation manner, the step S10 includes: In the latent space, based on the text embedding conditions extracted from the input text, the motion autoregressive transformer is used to predict the distribution of possible next indexes and generate sub-motion sequences one by one until a complete sequence is obtained, which is used as the base motion sequence.

[0028] In one embodiment, in the latent space, each text embedding is obtained by extracting information from the prompt text query, i.e., the input text, using CLIP. Given several previous indexes, based on the text embedding conditions, the motion autoregressive transformer predicts the distribution of the possible next index. Then, the sub-motion sequences are generated one by one in an autoregressive manner, and finally the complete sequence is represented to obtain the basic motion sequence.

[0029] Step S20, the basic motion sequence is processed by the encoder and the quantizer in the vector quantization variational autoencoder in sequence to obtain a residual index; In one embodiment, referring to Figure 3 ,In residual vector motion quantization, the residual vector quantization motion encoder is used to fit the ,discrete distribution of human motion space to latent space, and then the decoder is used to restore ,the original human motion.

[0030] In one embodiment, residual vectorized motion quantization uses VQ-VAE (Vector-Quantized Variational Autoencoder), which achieves data compression and generation by encoding input data into a discrete latent space and then decoding it back to the original data space. VQ-VAE learns the complex distribution of data through the discrete latent space and can generate high-quality samples when generating data. It includes the following execution processes: 1. Encoding process: Map the input data to a continuous latent space through the encoder, and encode the input data (such as an image) into a latent representation. 2. Quantization process: Discretize the continuous latent space representation into a set of fixed vectors, which are called codebooks. That is, in the vector quantization layer, the continuous latent representation output by the encoder is mapped to a predefined discrete latent vector space (corresponding to the embedding space of the graph) to achieve vector quantization. 3. Decoding process: Accept the quantized discrete latent representation and generate reconstructed input data, so that the discrete latent representation is converted back to the original data space through the decoder.

[0031] That is, the residual VQ-VAE consists of a motion encoder, a decoder, and a discrete codebook. The encoder maps the motion sequence to a latent vector space, performs vector quantization based on the codebook, and the decoder establishes a projection mapping between the motion sequence and the latent representation obtained by the codebook.

[0032] Among them, in VQ-VAE, residual layers are usually used in encoders and decoders to help the model better learn the feature representation of the data. The calculation method of the residual layer usually involves processing the input data through a series of convolutional layers, and then adding the processed results to the original input data to form a residual connection. This connection method can help the model better learn the feature representation of the data, especially when dealing with deep networks, it can effectively alleviate the gradient disappearance problem. Through residual connections, the model can more easily learn the local and global features of the data, thereby improving the performance of the model. In addition, the residual layer can also help the model converge faster during training and improve training efficiency.

[0033] In a feasible implementation manner, before step S20, the process includes: Adding the inter-frame motion speed loss and the acceleration loss to the reconstruction loss function of the vector quantized variational autoencoder to be trained to obtain a new reconstruction loss function; Based on the new reconstruction loss function, a vector quantized variational autoencoder is trained.

[0034] Current motion quantization methods often lead to unintentional jitter in synthesized motion, assuming that such jitter originates from sudden changes in velocity and acceleration between frames. For motion sequences, traditional representations only fit and focus on the positions of joints, but the actual motion is related to both its velocity and acceleration. To this end, based on the residual quantization design, we introduce a continuity-guided loss to alleviate the inter-frame discontinuity in motion sequences to further improve the problem caused by jitter.

[0035] Currently, the model is trained by minimizing the reconstruction error, embedding error, and constraint loss: In this step, the model is trained by minimizing the reconstruction error, which measures the difference between the original data and the decoder output, and the embedding error, which measures the difference between the latent vector and the closest prototype vector.

[0036] Among them, the decoder maps the discrete latent vector back to the high-dimensional continuous data space to generate the output of the model: In this step, the discrete latent vector is mapped back to the high-dimensional continuous data space through the decoder network to generate the output of the model.

[0037] During training, the loss function of VQ-VAE includes reconstruction loss (encouraging the decoded data to be close to the original input) and regularization term (encouraging the latent representation to conform to a predefined distribution, usually a standard normal distribution). By optimizing these loss functions, the model can learn effective data representations and be able to reconstruct the original data from these representations.

[0038] In this embodiment, the loss function of the encoder is optimized during the model training process. A motion continuity guidance mechanism is added to the motion reconstruction loss function, and motion speed loss and acceleration loss are used to alleviate the jitter in the motion generation process.

[0039] Furthermore, based on the guidance of motion continuity, a more continuous reconstruction effect is obtained. In motion processing, speed loss and acceleration loss are added to the reconstruction loss to prevent jitter in the motion generation process. The specific calculation formula is:

[0040] in, is an adjustable coefficient, represents the reconstructed motion sequence, represents the real motion sequence, is the speed representation of the motion sequence, is the acceleration representation of the motion sequence.

[0041] In a feasible implementation manner, before step S20, the step further includes: During training of the vector quantized variational autoencoder, different motions are mapped to corresponding vector quantization layers, where the number of vector quantization layers is the number of quantizers determined by the uniform sampling quantizer.

[0042] During training, for each input instance, multiple residual quantizers are uniformly sampled, where the number of quantizers is obtained by uniform sampling. During the training process, only The information of the layers is used to map the motion of different granularities (different motion types and motion amplitudes) to different vector quantization layers, for example, some quantization layers are responsible for small-amplitude motions, and other quantization layers are responsible for large-amplitude motions. By strengthening this hierarchical motion structure, the utilization of common motion patterns in the basic codebook can be increased, thereby aligning with the two-layer generative modeling module.

[0043] Step S30, inputting the input text, the basic motion sequence and the residual index into a non-autoregressive motion transformer to obtain a residual motion sequence; In one embodiment, referring to Figure 2 , the input text, the base motion sequence and the residual index are input into the non-autoregressive motion transformer to obtain the residual motion sequence. Among them, non-autoregressive movement is a prediction method that does not rely on historical data. It usually uses other variables or features to predict future values. In a non-autoregressive model, the current value of the time series is a function of other variables or features, rather than a function of its own past values.

[0044] Among them, the token from the residual quantization layer is modeled using a non-autoregressive motion transformer to obtain the residual layer indicator, that is, Figure 2 The residual index in . During training, the number of quantizer layers is randomly chosen.

[0045] In a feasible implementation manner, the step S30 includes: The text embedding condition, the basic motion sequence and the residual index as the residual layer indicator of the quantization layer extracted from the input text are embedded as the input of the non-autoregressive motion transformer, and the residual motion sequence of the residual quantization layer is predicted by the non-autoregressive motion transformer.

[0046] In one embodiment, referring to Figure 2, all quantization layers and previous tokens are embedded and aggregated as word unit input. Input text, residual layer indicator and base motion sequence embedding are used as input, wherein the non-autoregressive motion transformer is trained to predict the motion sequence of the residual quantization layer, so that the residual motion sequence of the residual quantization layer is predicted by the non-autoregressive motion transformer.

[0047] Step S40, the residual motion sequence is processed by the decoder in the vector quantization variational autoencoder to obtain a target motion sequence.

[0048] In one embodiment, referring to Figure 2 , the residual motion sequence obtained by the non-autoregressive motion transformer is processed by the decoder in the vector quantized variational autoencoder to obtain the target motion sequence.

[0049] In a feasible implementation manner, the step S40 includes: Add a sparse attention mask to the residual motion sequence to obtain a new residual motion sequence; The new residual motion sequence is processed by the decoder in the vector quantized variational autoencoder to obtain the target motion sequence.

[0050] In this embodiment, a transformer in the form of a mask decoder is designed in the self-attention mechanism, feature embedding is performed for text conditions, and as a conditional feature, it is required to be continuously visible in the attention mask, so as to realize continuous monitoring of the generation of motion sequences in the transformer. A sparse attention mask is added to the input motion sequence, and the local visibility of the motion subsequence is controlled when passing through the self-attention technology, reducing mutual interference in the process of generating long sequence actions. For the autoregressive transformer, the motion subsequence can only see the current information, and for the non-autoregressive transformer, it can additionally see the future local information.

[0051] Reference Figure 4 , based on the fixed conditional sparse attention, the mask design in the attention calculation process is shown; the blue in the figure represents the fixed range of the conditional tag, and the red represents the sliding window range of the adjacent motion tag. It uniquely sets the text tag to be globally visible while assigning the sliding window to each tag. This configuration not only enhances the model's ability to reason about long sequences, but also significantly improves computational efficiency.

[0052] The sliding focus sparse attention mechanism restricts the motion tokens to two receptive fields: a broader one that accesses all text conditions and a narrower one that focuses on the local context. During the training phase, the text tokens are utilized as reference points for the autoregressive transformer, while for the non-autoregressive transformer, a combination of non-autoregressive identifiers (NAR IDs) and text tokens is utilized.

[0053] For the autoregressive transformer, its mask takes the form of peaks, allowing its sliding window to only look at prior information; for the non-autoregressive transformer, it has a bidirectional context view, able to look at future information.

[0054] The text-driven motion generation method of the present application is applicable to the fields of digital human driving, virtual character animation, robot control, etc. In one application scenario, in order to overcome the jitter and cumulative error problems existing in the existing motion generation technology and achieve smoother, more natural and accurate motion generation, a transformer-based two-layer generation model architecture is provided. In the latent space of motion residual vectorization, a two-layer generation modeling structure composed of an autoregressive motion transformer and a non-autoregressive motion transformer is used to reconstruct the discrete distribution of the residual space based on sliding focus sparse attention.

[0055] Specifically, in the latent space, each text embedding uses CLIP to extract information from the prompt text query, i.e., the input text. Given several previous indexes, the motion autoregressive transformer predicts the distribution of possible next indexes based on text conditions. Then, the sub-motion sequences are generated one by one autoregressively, and finally the complete sequence is represented to obtain the basic motion sequence. The tokens from the residual quantization layer are modeled using a non-autoregressive transformer to obtain a residual motion sequence. During training, the number of quantizer layers is randomly selected. Subsequently, all the tokens of this quantization layer and before are embedded and aggregated as word unit inputs. Among them, using text conditions, residual layer indicators, and motion sequence embeddings as inputs, the non-autoregressive transformer is trained to predict the residual motion sequence of the residual quantization layer. Finally, a continuous motion guidance strategy is used to reduce the jitter of the generation process, and a decoder consisting of a 6-layer residual convolutional neural network is passed to obtain the reconstructed target motion sequence.

[0056] In this way, the quality of motion reconstruction and the fidelity of generated motion are improved through motion continuity guidance strategy and hierarchical quantitative screening strategy. The conditional control and local understanding capabilities of motion converters are enhanced through two-layer generative modeling, thereby achieving fine-grained motion generation driven by text description. This method reduces the jitter phenomenon and cumulative error in motion synthesis and improves the generation quality of continuous motion sequences.

[0057] It should be noted that the above examples are only used to understand the present application and do not constitute a limitation on the text-driven action generation method of the present application. More simple transformations based on this technical concept are all within the scope of protection of the present application.

[0058] This application also provides a text-driven action generation device, please refer to Figure 5 , the text-driven action generation device comprises: An autoregressive module 10 is used to input the input text and the empty motion sequence into the autoregressive motion transformer to obtain a basic motion sequence; The coding and quantization module 20 is used to process the basic motion sequence in sequence through the encoder and the quantizer in the vector quantization variational autoencoder to obtain a residual index; A non-autoregressive module 30, for inputting the input text, the basic motion sequence and the residual index into a non-autoregressive motion transformer to obtain a residual motion sequence; The decoding module 40 is used to process the residual motion sequence through the decoder in the vector quantization variational autoencoder to obtain the target motion sequence.

[0059] In one embodiment, the encoding and quantization module 20 is further configured to: Before the step of sequentially processing the basic motion sequence through the encoder and the quantizer in the vector quantized variational autoencoder to obtain the residual index, Adding the inter-frame motion speed loss and the acceleration loss to the reconstruction loss function of the vector quantized variational autoencoder to be trained to obtain a new reconstruction loss function; Based on the new reconstruction loss function, a vector quantized variational autoencoder is trained.

[0060] In one embodiment, the encoding and quantization module 20 is further configured to: Before the step of sequentially processing the basic motion sequence through the encoder and the quantizer in the vector quantized variational autoencoder to obtain the residual index, During training of the vector quantized variational autoencoder, different motions are mapped to corresponding vector quantization layers, where the number of vector quantization layers is the number of quantizers determined by the uniform sampling quantizer.

[0061] In one embodiment, the decoding module 40 is further configured to: Add a sparse attention mask to the residual motion sequence to obtain a new residual motion sequence; The new residual motion sequence is processed by the decoder in the vector quantized variational autoencoder to obtain the target motion sequence.

[0062] In one embodiment, the autoregressive module 10 is further used for: In the latent space, based on the text embedding conditions extracted from the input text, the motion autoregressive transformer is used to predict the distribution of possible next indexes and generate sub-motion sequences one by one until a complete sequence is obtained, which is used as the base motion sequence.

[0063] In one embodiment, the non-autoregressive module 30 is further configured to: The text embedding condition, the basic motion sequence and the residual index as the residual layer indicator of the quantization layer extracted from the input text are embedded as the input of the non-autoregressive motion transformer, and the residual motion sequence of the residual quantization layer is predicted by the non-autoregressive motion transformer.

[0064] The text-driven action generation device provided by the present application adopts the text-driven action generation method in the above-mentioned embodiment, which can solve the technical problem of poor accuracy of text-driven action generation. Compared with the prior art, the beneficial effects of the text-driven action generation device provided by the present application are the same as the beneficial effects of the text-driven action generation method provided by the above-mentioned embodiment, and the other technical features in the text-driven action generation device are the same as the features disclosed in the above-mentioned embodiment method, which will not be repeated here.

[0065] The present application provides a text-driven action generation device, which includes: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute the text-driven action generation method in the above-mentioned embodiment one.

[0066] Reference below Figure 6 , which shows a schematic diagram of the structure of a text-driven action generation device suitable for implementing the embodiment of the present application. The text-driven action generation device in the embodiment of the present application may include but is not limited to mobile terminals such as mobile phones, laptop computers, digital broadcast receivers, PDAs (Personal Digital Assistants), PADs (Portable Application Descriptions), PMPs (Portable Media Players), vehicle-mounted terminals (such as vehicle-mounted navigation terminals), etc., and fixed terminals such as digital TVs, desktop computers, etc. Figure 6 The text-driven action generation device shown is merely an example and should not bring any limitation to the functions and scope of use of the embodiments of the present application.

[0067] like Figure 6As shown, the text-driven action generation device may include a processing device 1001 (e.g., a central processing unit, a graphics processor, etc.), which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM: Read Only Memory) 1002 or a program loaded from a storage device 1003 to a random access memory (RAM: Random Access Memory) 1004. In RAM1004, various programs and data required for the operation of the text-driven action generation device are also stored. The processing device 1001, ROM1002, and RAM1004 are connected to each other through a bus 1005. An input / output (I / O) interface 1006 is also connected to the bus. Generally, the following systems can be connected to the I / O interface 1006: an input device 1007 including, for example, a touch screen, a touchpad, a keyboard, a mouse, an image sensor, a microphone, an accelerometer, a gyroscope, etc.; an output device 1008 including, for example, a liquid crystal display (LCD: Liquid Crystal Display), a speaker, a vibrator, etc.; a storage device 1003 including, for example, a magnetic tape, a hard disk, etc.; and a communication device 1009. Communication device 1009 can allow text-driven action generation device to communicate with other devices wirelessly or by wire to exchange data. Although the figure shows a text-driven action generation device with various systems, it should be understood that it is not required to implement or have all the systems shown. More or fewer systems can be implemented or have alternatively.

[0068] In particular, according to the embodiments disclosed in the present application, the process described above with reference to the flowchart can be implemented as a computer software program. For example, the embodiments disclosed in the present application include a computer program product, which includes a computer program carried on a computer-readable medium, and the computer program includes a 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, or installed from a storage device 1003, or installed from a ROM 1002. When the computer program is executed by the processing device 1001, the above-mentioned functions defined in the method of the embodiment disclosed in the present application are executed.

[0069] The text-driven action generation device provided by the present application adopts the text-driven action generation method in the above embodiment, which can solve the technical problem of poor accuracy of text-driven action generation. Compared with the prior art, the beneficial effects of the text-driven action generation device provided by the present application are the same as the beneficial effects of the text-driven action generation method provided by the above embodiment, and other technical features in the text-driven action generation device are the same as the features disclosed in the method of the previous embodiment, which will not be repeated here.

[0070] It should be understood that the various parts disclosed in this application can be implemented by hardware, software, firmware or a combination thereof. In the description of the above embodiments, specific features, structures, materials or characteristics can be combined in any one or more embodiments or examples in a suitable manner.

[0071] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art who is familiar with the present technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.

[0072] The present application provides a computer-readable storage medium having computer-readable program instructions (ie, computer program) stored thereon, wherein the computer-readable program instructions are used to execute the text-driven action generation method in the above-mentioned embodiment.

[0073] The computer-readable storage medium provided in the present application may be, for example, a USB flash drive, but is not limited to electrical, magnetic, optical, electromagnetic, infrared, or semiconductor systems or devices, 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: Random Access Memory), a read-only memory (ROM: Read Only Memory), an erasable programmable read-only memory (EPROM: Erasable Programmable Read Only Memory or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM: CD-Read Only Memory), an optical storage device, a magnetic storage device, or any suitable combination of the above. In this embodiment, the computer-readable storage medium may be any tangible medium containing or storing a program, which may be used by or in combination with an instruction execution system or device. The program code contained on the computer-readable storage medium may be transmitted using any appropriate medium, including but not limited to: wires, optical cables, RF (Radio Frequency: Radio Frequency), etc., or any suitable combination of the above.

[0074] The computer-readable storage medium may be included in the text-driven action generation device; or may exist independently without being assembled into the text-driven action generation device.

[0075] The above-mentioned computer-readable storage medium carries one or more programs. When the above-mentioned one or more programs are executed by a text-driven action generation device, the text-driven action generation device: inputs the input text and the empty motion sequence into the autoregressive motion transformer to obtain a basic motion sequence; processes the basic motion sequence in sequence through the encoder and the quantizer in the vector quantization variational autoencoder to obtain a residual index; inputs the input text, the basic motion sequence and the residual index into the non-autoregressive motion transformer to obtain a residual motion sequence; processes the residual motion sequence through the decoder in the vector quantization variational autoencoder to obtain a target motion sequence.

[0076] Computer program code for performing the operations of the present application may be written in one or more programming languages ​​or a combination thereof, including 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 the case of 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., via the Internet using an Internet service provider).

[0077] 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 application. 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 alternative implementations, 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.

[0078] The modules involved in the embodiments described in this application may be implemented by software or hardware, wherein the name of the module does not constitute a limitation on the unit itself in some cases.

[0079] The readable storage medium provided by the present application is a computer-readable storage medium, which stores computer-readable program instructions (i.e., computer programs) for executing the above-mentioned text-driven action generation method, and can solve the technical problem of poor accuracy of text-driven action generation. Compared with the prior art, the beneficial effects of the computer-readable storage medium provided by the present application are the same as the beneficial effects of the text-driven action generation method provided by the above-mentioned embodiment, and will not be repeated here.

[0080] The present application also provides a computer program product, including a computer program, which implements the steps of the above-mentioned text-driven action generation method when executed by a processor.

[0081] The computer program product provided by the present application can solve the technical problem of poor accuracy of text-driven action generation. Compared with the prior art, the beneficial effects of the computer program product provided by the present application are the same as the beneficial effects of the text-driven action generation method provided by the above embodiment, which will not be repeated here.

[0082] The above descriptions are only some embodiments of the present application, and are not intended to limit the patent scope of the present application. All equivalent structural changes made using the contents of the present application specification and drawings under the technical concept of the present application, or direct / indirect applications in other related technical fields are included in the patent protection scope of the present application.

Claims

1. A text-driven action generation method, characterized in that: The text-driven action generation method comprises: Input the input text and the empty motion sequence into the autoregressive motion transformer to obtain the basic motion sequence; The basic motion sequence is processed by an encoder and a quantizer in a vector quantized variational autoencoder in sequence to obtain a residual index; Inputting the input text, the basic motion sequence and the residual index into a non-autoregressive motion transformer to obtain a residual motion sequence; The residual motion sequence is processed by the decoder in the vector quantization variational autoencoder to obtain a target motion sequence.

2. The text-driven action generation method according to claim 1, characterized in that: The step of sequentially processing the basic motion sequence through an encoder and a quantizer in a vector quantized variational autoencoder to obtain a residual index includes: Adding the inter-frame motion speed loss and the acceleration loss to the reconstruction loss function of the vector quantized variational autoencoder to be trained to obtain a new reconstruction loss function; Based on the new reconstruction loss function, a vector quantized variational autoencoder is trained.

3. The text-driven action generation method according to claim 1, characterized in that: Before the step of sequentially processing the basic motion sequence through an encoder and a quantizer in a vector quantized variational autoencoder to obtain a residual index, the step further includes: During training of the vector quantized variational autoencoder, different motions are mapped to corresponding vector quantization layers, where the number of vector quantization layers is the number of quantizers determined by the uniform sampling quantizer.

4. The text-driven action generation method according to claim 1, characterized in that: The step of obtaining a target motion sequence by processing the residual motion sequence through a decoder in the vector quantized variational autoencoder comprises: Adding a sparse attention mask to the residual motion sequence to obtain a new residual motion sequence; The new residual motion sequence is processed by the decoder in the vector quantization variational autoencoder to obtain a target motion sequence.

5. The text-driven action generation method according to claim 1, characterized in that: The step of inputting the input text and the empty motion sequence into the autoregressive motion transformer to obtain the basic motion sequence comprises: In the latent space, based on the text embedding conditions extracted from the input text, the distribution of possible next indexes is predicted through a motion autoregressive transformer and sub-motion sequences are generated one by one until a complete sequence is obtained, and the complete sequence is used as the basic motion sequence.

6. The text-driven action generation method according to claim 1, characterized in that: The step of inputting the input text, the basic motion sequence and the residual index into a non-autoregressive motion transformer to obtain a residual motion sequence comprises: The text embedding conditions extracted from the input text, the basic motion sequence and the residual index as the residual layer indicator of the quantization layer are embedded as the input of the non-autoregressive motion transformer, and the residual motion sequence of the residual quantization layer is predicted by the non-autoregressive motion transformer.

7. A text-driven action generation device, characterized in that: The text-driven action generation device comprises: An autoregressive module, used for inputting input text and an empty motion sequence into an autoregressive motion transformer to obtain a basic motion sequence; An encoding and quantization module, used for sequentially processing the basic motion sequence through an encoder and a quantizer in a vector quantization variational autoencoder to obtain a residual index; A non-autoregressive module, used for inputting the input text, the basic motion sequence and the residual index into a non-autoregressive motion transformer to obtain a residual motion sequence; The decoding module is used to process the residual motion sequence through the decoder in the vector quantization variational autoencoder to obtain a target motion sequence.

8. A text-driven action generation device, characterized in that: The device comprises: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the computer program is configured to implement the steps of the text-driven action generation method according to any one of claims 1 to 6.

9. A storage medium, characterized in that: The storage medium is a computer-readable storage medium, and a computer program is stored on the storage medium. When the computer program is executed by a processor, the steps of the text-driven action generation method according to any one of claims 1 to 6 are implemented.

10. A computer program product, characterized in that The computer program product comprises a computer program, and when the computer program is executed by a processor, the steps of the text-driven action generation method according to any one of claims 1 to 6 are implemented.

Images