Vehicle type identification methods, electronic devices and computer storage media

Abstract

This application provides a vehicle type identification method, an electronic device, and a computer storage medium. The method includes: acquiring target audio data; extracting channel features from the target audio data using a depthwise separable convolution module to obtain first feature data; performing dimensionality transformation on the first feature data to obtain second feature data; processing the second feature data using element-wise multiplication to obtain third feature data; extracting channel features from the third feature data using a depthwise separable convolution module to obtain fourth feature data; calculating multiple probability values ​​between the fourth feature data and multiple vehicle labels; and determining the vehicle type based on the multiple probability values. This application can promptly identify the vehicle type based on the audio signal of the preceding driving environment.

Description

Technical Field

[0001] This application relates to the field of vehicle technology, specifically to a vehicle type identification method, electronic device, and computer storage medium. Background Technology

[0002] With the acceleration of urbanization and the continuous development of vehicle technology, more and more private cars, commercial vehicles, and special vehicles are appearing on the road. Among them, special vehicles are those that need to perform emergency tasks, such as ambulances, police cars, military vehicles, rescue vehicles, and fire trucks.

[0003] If the types of nearby vehicles are not promptly identified to determine the presence of emergency vehicles, and other vehicles on the road do not take evasive action towards these emergency vehicles, it becomes difficult for them to travel quickly. This can lead to loss of life or property due to the emergency vehicles' inability to reach their destinations quickly. Summary of the Invention

[0004] In view of the above, embodiments of this application provide a vehicle type identification method, electronic device, and computer storage medium, which can solve the problem in the prior art of not timely identifying the vehicle type of nearby vehicles in order to determine whether there are special vehicles.

[0005] This application provides a vehicle type identification method applied to a first vehicle. The first vehicle is equipped with an audio classification model, which is used to identify the vehicle type of a second vehicle located to one side of the first vehicle. The method includes: acquiring target audio data, which includes sound information collected by the first vehicle under the current driving environment, and the target audio data having multiple channels; extracting channel features from the target audio data to obtain first feature data, wherein the channel dimension of the first feature data is greater than the channel dimension of the target audio data; performing dimensionality transformation on the first feature data to obtain second feature data, wherein the number of channels in the second feature data is the same as the number of channels in the target audio data; processing the second feature data to obtain third feature data, which includes data interaction information between the channels in the second feature data; extracting channel features from the third feature data to obtain fourth feature data, which is used to characterize the sound features of the second vehicle; and determining the vehicle type of the second vehicle based on the fourth feature data.

[0006] The vehicle type identification method of this application extracts channel features from the target audio data, thereby separating the correlation between the spatial and channel dimensions of the target audio data and reducing computational parameters. Then, dimensionality transformation is performed on the first feature data to reduce the complexity of subsequent data computation. Next, data processing is performed on the second feature data to enhance information interaction between different feature channels and improve the model's ability to represent features. Then, channel features are extracted from the third feature data to obtain data that accurately reflects the sound characteristics of the second vehicle. Finally, the vehicle type of the second vehicle is determined based on the fourth feature data. This method can accurately identify the type of the second vehicle in a timely manner.

[0007] In some embodiments, after processing the second feature data to obtain the third feature data, the method further includes: performing fully connected processing on the third feature data to obtain first processed data, wherein the number of channels in the first processed data is the same as the number of channels in the second feature data; and performing channel feature extraction on the third feature data to obtain the fourth feature data includes: performing channel feature extraction on the first processed data to obtain the fourth feature data.

[0008] In some embodiments, after performing fully connected processing on the third feature data to obtain the first processed data, the method further includes: normalizing the first processed data to obtain the second processed data; the step of extracting channel features from the first processed data to obtain the fourth feature data includes: extracting channel features from the second processed data to obtain the fourth feature data.

[0009] In some embodiments, after extracting channel features from the target audio data to obtain first feature data, the method further includes: normalizing the first feature data to obtain third processed data; the step of performing dimensional transformation on the first feature data to obtain second feature data includes: performing dimensional transformation on the third processed data to obtain the second feature data.

[0010] In some embodiments, the audio classification model includes a second activation layer; after performing dimensional transformation on the third processed data to obtain the second feature data, the model further includes: inputting the second feature data into the second activation layer to obtain fourth processed data, wherein each data in the fourth processed data is within a preset numerical range; the step of processing the second feature data to obtain the third feature data includes: processing the fourth processed data to obtain the third feature data.

[0011] In some embodiments, determining the vehicle type of the second vehicle based on the fourth feature data includes: calculating multiple probability values ​​between the fourth feature data and multiple vehicle tags, where each vehicle tag corresponds to a vehicle type and each vehicle tag corresponds to a probability value; determining a target probability value based on the multiple probability values; and using the vehicle type corresponding to the target probability value as the vehicle type of the second vehicle.

[0012] In some embodiments, the audio classification model includes a first activation layer; calculating multiple probability values ​​between the fourth feature data and multiple vehicle labels includes: pooling the fourth feature data to obtain fifth feature data; and inputting the fifth feature data into the first activation layer to obtain multiple probability values.

[0013] In some embodiments, acquiring the target audio data includes: acquiring the audio signal of the first vehicle in the current driving environment; performing data processing on the audio signal to obtain initial audio data; reshaping the initial audio data to obtain reshaped audio data; and aggregating the reshaped audio data to obtain the target audio data.

[0014] This application also provides an electronic device, which includes a processor and a memory. The memory is used to store instructions, and the processor is used to call the instructions in the memory to cause the electronic device to execute the above-described vehicle type recognition method.

[0015] This application also provides a computer-readable storage medium that stores computer instructions that, when executed on an electronic device, cause the electronic device to perform the vehicle type identification method described above.

[0016] The aforementioned electronic devices and computer-readable storage media correspond to the aforementioned vehicle type identification method and vehicle type identification method. Therefore, the beneficial effects they can achieve can be referred to the beneficial effects in the corresponding methods provided above, and will not be repeated here. Attached Figure Description

[0017] Figure 1 This is a flowchart of a vehicle type identification method according to an embodiment of this application.

[0018] Figure 2 This is a schematic diagram of the structure of an audio classification model provided in an embodiment of this application.

[0019] Figure 3 for Figure 2 A schematic diagram of the StarBlock module in the diagram.

[0020] Figure 4 This is a flowchart of the training process of an audio classification model provided according to an embodiment of this application.

[0021] Figure 5 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application. Detailed Implementation

[0022] To better understand the above-mentioned objectives, features, and advantages of this application, the application will be described in detail below with reference to the accompanying drawings and specific embodiments. It should be noted that, unless otherwise specified, the embodiments and features described in these embodiments can be combined with each other.

[0023] The following description sets forth many specific details to provide a full understanding of this application. The described embodiments are only some, not all, of the embodiments of this application.

[0024] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs. The terminology used herein in the specification of this application is for the purpose of describing particular embodiments only and is not intended to be limiting of this application.

[0025] It should be further noted that, in this document, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes that element.

[0026] In this application, "at least one" means one or more, and "more than one" means two or more. "And / or" describes the relationship between related objects, indicating that there can be three relationships. For example, A and / or B can mean: A exists alone, A and B exist simultaneously, or B exists alone, where A and B can be singular or plural.

[0027] In the embodiments of this application, the terms "exemplary" or "for example" are used to indicate that something is an example, illustration, or description. Any embodiment or design that is described as "exemplary" or "for example" in the embodiments of this application should not be construed as being more preferred or advantageous than other embodiments or design. Specifically, the use of the terms "exemplary" or "for example" is intended to present the relevant concepts in a specific manner.

[0028] This application provides a vehicle type identification method, an electronic device, and a computer-readable storage medium. The vehicle type identification method is applied to a first vehicle, which may be a commercial vehicle, a private vehicle, an engineering vehicle, etc. This application does not limit the specific type of the first vehicle.

[0029] Specifically, the first vehicle is equipped with an audio classification model, which is used to identify the vehicle type of the second vehicle, which is located to one side of the first vehicle. In this embodiment, the second vehicle is located within a preset acquisition range of the first vehicle. The preset acquisition range can be a circular area with a radius of 10 meters centered on the location of the first vehicle. Alternatively, the preset acquisition range can be a quadrilateral area with a length of 10 meters and a width of 5 meters centered on the location of the first vehicle. This application does not limit the preset acquisition range of the first vehicle; the preset acquisition range of the first vehicle can be a regular or irregular shape.

[0030] The first vehicle uses a pre-trained audio classification model to identify the vehicle type of the second vehicle, thereby determining whether the second vehicle is a special vehicle. When the second vehicle is identified as a special vehicle, the driver of the first vehicle is alerted that there is a special vehicle nearby, so as to remind the driver to take evasive action and allow the special vehicle to pass first.

[0031] Special vehicles are those required to perform emergency tasks, including ambulances, police cars, military vehicles, rescue vehicles, fire trucks, breakdown vehicles, engineering rescue vehicles, road sweepers, water trucks, garbage trucks, aerial work platforms, road construction vehicles, traffic command vehicles, hazardous materials transport vehicles, mobile communication vehicles, emergency medical vehicles, power engineering vehicles, water conservancy engineering vehicles, and tunnel inspection vehicles, etc. In other embodiments, the method can also be applied to other equipment besides the first vehicle, and this application is not limited thereto.

[0032] The following section will describe in detail the specific applications and training methods of the audio classification model.

[0033] Figure 1 This is a flowchart illustrating the steps of an embodiment of the vehicle type identification method of this application. Depending on different requirements, the order of the steps in the flowchart can be changed, and some steps can be omitted.

[0034] See Figure 1 As shown, the vehicle type identification method may include the following steps.

[0035] Step 101: Obtain the target audio data.

[0036] In this embodiment, the target audio data includes sound information collected by the first vehicle under the current driving environment, and the target audio data has multiple channels. Specifically, in order to obtain the target audio data, a sound acquisition device can be installed on the first vehicle. During the driver's operation of the first vehicle, the sound acquisition device collects the audio signal of the first vehicle under the current driving environment. The sound acquisition device can be an audio sensor, microphone, etc.

[0037] Specifically, the audio signal is digitized to obtain initial audio data. This initial audio data is then preprocessed to obtain intermediate audio data. Data preprocessing includes data type conversion, data segmentation, and data normalization. For example, the numerical value of each data point in the initial audio data is processed to ensure that each data point's value falls within a preset range. This preset range can be [-1, 1], and can be changed according to the actual data preprocessing requirements.

[0038] Since the initial audio data is quite large, a sliding window approach can be used to process it step by step. The sliding window is set to a length of 0.96 seconds and a step size of 0.48 seconds. In other embodiments, the sliding window can also be set to a length of 1.25 seconds and a step size of 0.65 seconds. The sliding window can be adjusted according to the actual data processing requirements.

[0039] Furthermore, such as Figure 2 As shown, the audio classification model includes a data reshaping layer, a data aggregation layer, a StarBlock module, a pooling layer, and a first activation layer arranged in sequence. Intermediate audio data is input to the data reshaping layer for data reshaping processing to remove interfering data. The reshaped intermediate audio data is then input to the data aggregation layer for data aggregation processing to ensure data consistency. In this embodiment, before inputting the intermediate audio data to the data reshaping layer for data reshaping processing, a short-time Fourier transform can be used to process the intermediate audio data to obtain log-Mel features, which are then used as the intermediate audio data. Data aggregation and data reshaping processing are existing technologies and will not be described in detail here.

[0040] For audio signals, the corresponding data has a high dimensionality. Directly inputting this data into a model for training or inference leads to high complexity and a high risk of overfitting. Therefore, it is necessary to preprocess, reshape, and aggregate the audio signal data before using it as model input. This not only improves the model's accuracy but also reduces the complexity of processing the initial data. In this embodiment, log-Mel features can perform frame-by-frame calculations on sound, capturing the instantaneous dynamic features of the sound source and mapping a frequency response similar to human auditory perception. Therefore, log-Mel features are closer to the characteristics of audio signals. When using the model for training or inference, short-time Fourier transform is used to process the intermediate audio data to obtain log-Mel features. Using these log-Mel features as intermediate audio data better reflects the differences between different sample categories in the audio signal, thus enabling more accurate sound identification and classification.

[0041] Step 102: Extract channel features from the target audio data to obtain the first feature data.

[0042] In some embodiments, the channel dimension of the first feature data is greater than the channel dimension of the target audio data. Please refer to... Figure 3 The StarBlock module includes a depthwise separable convolutional module, a fully connected layer, a normalization layer, and a second activation layer. The depthwise separable convolutional module includes a first depthwise separable convolutional layer and a second depthwise separable convolutional layer.

[0043] Assuming the target audio data has a channel dimension of d, the target audio data is input into the StarBlock module, and the first depthwise separable convolutional layer is used to extract channel features from the target audio data, resulting in the first feature data with a channel dimension of 4d.

[0044] Specifically, depthwise separable convolution mainly includes channel-wise convolution and pointwise convolution. Channel-wise convolution applies a convolutional kernel independently to each input channel, with each kernel responsible for only one channel's convolution operation, thus significantly reducing the number of parameters and computational cost. Following channel-wise convolution, pointwise convolution uses a 1x1 convolutional kernel to convolve the output of all channels, mixing and combining features from different channels to generate new feature data, thereby enhancing the model's expressive power.

[0045] Step 103: Perform dimensional transformation on the first feature data to obtain the second feature data.

[0046] In some embodiments, the number of channels in the second feature data is the same as the number of channels in the target audio data. Before performing dimensionality transformation on the first feature data, it is necessary to normalize the first feature data based on a normalization layer to obtain the third processed data. Then, the third processed data is dimensionally transformed to obtain the second feature data. In this embodiment, the normalization layer is a batch normalization layer. The application of the normalization layer stabilizes the model inference process, thereby reducing the occurrence of overfitting. At the same time, two parallel fully connected layers are designed after the first depthwise separable convolutional layer. The third processed data is input into the two parallel fully connected layers for dimensionality transformation to obtain the second feature data. The two parallel fully connected layers can transform the first feature data with a channel dimension of 4d into the second feature data with a channel dimension of d. This not only extracts multi-scale feature data from the third processed data but also ensures that the model has high computational efficiency.

[0047] Furthermore, the second feature data is input into the second activation layer to obtain the fourth processed data, where each data point is within a preset numerical range. In this embodiment, the second activation layer is equipped with a ReLU6 activation function. The ReLU6 activation function sets all outputs of data less than 0 in the second feature data to 0, while limiting all outputs to a preset numerical range, i.e., between [0, 6]. The processing of the second activation layer helps reduce the computational load of the model and prevent overfitting.

[0048] Step 104: Process the second feature data to obtain the third feature data.

[0049] In some embodiments, the third feature data includes data interaction information between the channels in the second feature data. Specifically, please refer to... Figure 3 The fourth processing data is processed using element-wise multiplication to obtain the third feature data. The application of element-wise multiplication can enhance the information interaction ability between different feature channels in the fourth processing data, and also improve the model's ability to represent features.

[0050] Furthermore, the third feature data is processed using a fully connected layer to obtain the first processed data. The number of channels in the first processed data is the same as the number of channels in the second feature data. Processing the third feature data using a fully connected layer allows the model to combine the feature data extracted from the first depthwise separable convolutional layer, two parallel fully connected layers, a normalization layer, the ReLU6 activation function, and element-wise multiplication, resulting in data that fully reflects the channel characteristics.

[0051] Step 105: Extract channel features from the third feature data to obtain the fourth feature data.

[0052] In some embodiments, the fourth feature data is used to characterize the sound features of the second vehicle. Specifically, channel features are extracted from the first processed data based on a second deep separable convolutional layer to obtain the fourth feature data. Specifically, the first processed data is normalized based on a normalization layer to obtain the second processed data. Channel features are extracted from the second processed data based on a second deep separable convolutional layer to obtain the fourth feature data.

[0053] Step 106: Determine the vehicle type of the second vehicle based on the fourth feature data.

[0054] In some embodiments, multiple probability values ​​are calculated between the fourth feature data and multiple vehicle tags, where each vehicle tag corresponds to a vehicle type and each vehicle tag corresponds to a probability value. A target probability value is determined based on the multiple probability values, and the vehicle type corresponding to the target probability value is taken as the vehicle type of the second vehicle.

[0055] In this embodiment, corresponding vehicle tags are set according to vehicle type. For example, when special vehicles include ambulances, police cars, military vehicles, rescue vehicles, and fire trucks, the vehicle tags include ambulance tags, police car tags, military vehicle tags, rescue vehicle tags, and fire truck tags. In other embodiments, if special vehicles also include other types of vehicles, corresponding other vehicle tags are set.

[0056] For specific details, please refer to... Figure 2 Based on the pooling layer, the fourth feature data is pooled to obtain the fifth feature data. The fifth feature data is then input to the first activation layer to obtain multiple probability values. In this embodiment, the fourth feature data is subjected to average pooling to obtain the fifth feature data. Average pooling uses a sliding window to average all values ​​within each window, obtaining the output value for that window. Average pooling effectively suppresses noise and reduces information redundancy. It also preserves the background information of the data well, reducing the loss of background information caused by ignoring local maxima.

[0057] The first activation layer uses a Softmax function. The Softmax function calculates the probability value between the fifth feature data and each vehicle label. The sum of all probability values ​​is 1.

[0058] In some embodiments, the highest probability value among multiple probability values ​​is used as the target probability value. The vehicle type corresponding to the vehicle tag that matches the target probability value is the vehicle type of the second vehicle.

[0059] In this embodiment, it can also detect whether the target probability value is greater than a preset probability threshold. If the target probability value is detected to be greater than the preset probability threshold, then the vehicle type corresponding to the vehicle tag matching the target probability value is the vehicle type of the second vehicle. The preset probability threshold is 0.95 or 0.9. The specific value of the preset probability threshold can be set according to the actual detection requirements.

[0060] It should be noted that two or more StarBlock modules can be set between the data aggregation layer and the pooling layer. The more StarBlock modules are set, the more key features can be extracted, and the higher the accuracy of the calculated probability values.

[0061] In this embodiment, when the second vehicle is determined to be a special vehicle, the first vehicle can issue a warning message to other vehicles, alerting them that a special vehicle is passing by and needs to give way to ensure that the special vehicle can reach its destination in a timely manner. The first vehicle can issue the warning message through sound or a screen display. This application does not limit the method by which the first vehicle issues the warning message.

[0062] Compared with the prior art, this embodiment has at least the following advantages:

[0063] By utilizing a first deep convolutional layer to process the target audio data, the correlation between the spatial and channel dimensions of the target audio data can be separated, reducing the number of parameters required for convolution calculations. Then, the first feature data undergoes dimensionality transformation to reduce the complexity of subsequent data computations. Next, element-wise multiplication is used to process the second feature data to enhance information interaction between different feature channels and improve the model's ability to represent features. Following this, a second deep separable convolutional layer is used to process the third feature data, obtaining data that accurately reflects the sound characteristics of the second vehicle. Finally, the vehicle type of the second vehicle is determined by calculating multiple probability values. This method can promptly identify the vehicle type of the second vehicle, and when the second vehicle is identified as a special vehicle, it alerts other vehicles to give way, enabling the special vehicle to complete its emergency mission in a timely manner, thereby avoiding or reducing casualties or property damage.

[0064] refer to Figure 4 The diagram shown is a flowchart illustrating the training process of the audio classification model provided in this embodiment. The specific steps include:

[0065] Step 201: Obtain training audio data.

[0066] In some embodiments, the training audio data includes sound information collected by the first vehicle under historical driving conditions.

[0067] For example, alarm sound samples and road sound samples from a specific type of vehicle can be collected. Then, an audio synthesis editor is used to edit these samples. Specifically, the time-series characteristics of the alarm and road sound samples are adjusted, including changing the pitch and volume, to ensure they can adapt to diverse application scenarios. Subsequently, these processed alarm and road sound samples are synthesized to obtain a synthesized sound. After synthesis, the realism and consistency of the synthesized sound are verified through manual evaluation and professional audio analysis tools. This ensures that the synthesized sound performs well in the subsequent training of audio classification models.

[0068] Next, the synthesized sound undergoes data processing to obtain training audio data. This data processing includes data preprocessing, data reshaping, and data aggregation. For details, please refer to step 101; to avoid repetition, it will not be elaborated here.

[0069] Step 202: Extract channel features from the training audio data to obtain the first training feature data.

[0070] In some embodiments, the channel dimension of the first training feature data is greater than the channel dimension of the training audio data, and the training audio data has multiple channels.

[0071] Step 203: Perform dimensionality transformation on the first training feature data to obtain the second training feature data.

[0072] In some embodiments, the number of channels in the second training feature data is the same as the number of channels in the training audio data.

[0073] Step 204: Process the second training feature data to obtain the third training feature data.

[0074] In some embodiments, the third training feature data is used to characterize the data interaction information between the channels in the second training feature data.

[0075] Step 205: Extract channel features from the third training feature data to obtain the fourth training feature data.

[0076] The specific content of steps 202 to 205 is the same as that of steps 102 to 105, and will not be repeated here to avoid repetition.

[0077] Step 206: Calculate multiple training probability values ​​between the fourth training feature data and multiple vehicle labels, where each vehicle label corresponds to a vehicle type.

[0078] In some embodiments, the fourth training feature data is pooled to obtain the fifth training feature data. The fifth training feature data is then input into the first activation layer to obtain multiple training probability values. The first activation layer has a Softmax function. The Softmax function calculates the probability value between the fifth training feature data and each vehicle label. The sum of all probability values ​​is 1.

[0079] Step 207: Obtain the labeled probability data.

[0080] In some embodiments, the labeled probability data represents the correlation between the training audio data and each vehicle label. In this embodiment, the training audio data is manually labeled to obtain the labeled probability data. For example, if the training audio data corresponds to the sound emitted by an ambulance, the probability value between the training audio data and the ambulance label is marked as 1, and the probability value between the training audio data and other vehicle labels is marked as 0. The probability values ​​between the training audio data and all vehicle labels are used as the labeled probability data.

[0081] Step 208: Based on multiple training probability values ​​and labeled probability data, calculate the loss value of the loss function corresponding to the audio classification model.

[0082] In this embodiment, the cross-entropy loss function is used to calculate the loss value of the loss function corresponding to the audio classification model, thereby measuring the difference between multiple training probability values ​​and labeled probability data.

[0083] Step 209: Update the model parameters of the audio classification model based on the loss value until the audio classification model converges.

[0084] The gradient of the loss function with respect to the model parameters is calculated by backpropagation algorithm, and then the model parameters are updated using optimization algorithm (such as Adam) to minimize the loss function value, thereby improving the prediction accuracy of the model.

[0085] In some embodiments, the He initialization method is also used to initialize model parameters. The He initialization method is a method for initializing neural network weights, designed to address the problem of neuron "death" that may occur during the initialization of the ReLU activation function. He initialization initializes weights reasonably based on the number of activation units in the previous layer, ensuring that the input and output variances of each layer are consistent, thereby accelerating training speed and improving model stability.

[0086] Combining the Adam optimization algorithm, loss function, and He initialization method can effectively optimize neural network models. The Adam algorithm accelerates convergence and improves training stability by dynamically adjusting the learning rate; the loss function, as the optimization objective, guides the adjustment of model parameters and improves prediction accuracy; and the He initialization method ensures the rationality of network initialization and reduces difficulties in the early stages of training. These methods work together to enable the model to learn and generalize better.

[0087] It should be noted that the training audio data can include multiple initial training data sets. Each initial training data set corresponds to a vehicle type. Each initial training data set is then input into the audio classification model, allowing the model to learn the sound characteristics emitted by vehicles of different types. This improves the accuracy of the audio classification model in identifying the vehicle type of a second vehicle.

[0088] Compared with the prior art, this embodiment has at least the following advantages:

[0089] On the one hand, the model parameters are updated using a loss function and the Adam optimization algorithm to minimize the loss function value, thereby improving the model's prediction accuracy. On the other hand, the He initialization method is used to address the potential neuron "death" problem during ReLU activation function initialization, ensuring consistent input and output variance at each layer, thus accelerating training and improving model stability. This ultimately enhances the training performance of the audio classification model.

[0090] This application also provides a vehicle type recognition system. A vehicle type recognition method is applied to the vehicle type recognition system. The vehicle type recognition system is deployed within a first vehicle. The vehicle type recognition system includes a sound acquisition device, a data transmission component, a signal processing unit, and a control unit. The control unit is communicatively connected to the sound acquisition device, the data transmission component, and the signal processing unit.

[0091] In this embodiment, the sound acquisition device is an external microphone array responsible for capturing ambient sound and performing preliminary processing using pulse density modulation (PDM) technology to optimize the quality of the audio signal. These signals are then transmitted from the sound acquisition device to the data transmission component via a main line, where the data transmission component transmits the signals to ensure accurate data transmission and synchronization.

[0092] The data transmission component includes multiple repeaters that forward signals in the mainline network, ensuring stable signal transmission between nodes. Simultaneously, the data transmission component utilizes Time Division Multiplexing (TDM) technology during data transmission. The application of TDM technology further improves transmission efficiency, allowing multiple signals to be transmitted sequentially on the same channel without collisions.

[0093] The signal processing unit includes a signal modulation and demodulation module. This module utilizes Quadrature Amplitude Modulation (QAM) technology to modulate and demodulate the signal, achieving high-speed and stable data transmission. The data processed by the signal processing unit is then transmitted to the control unit. A vehicle type recognition method deployed within the control unit processes the data input to the signal processing unit, thereby analyzing and identifying the audio signal acquired from the first vehicle.

[0094] Specifically, the vehicle type recognition method will be deployed in the control unit. This includes:

[0095] Quantization-Aware Training (QAT) is used to obtain the quantization parameters needed to convert the audio classification model from floating-point numbers to low-precision integer representations. Then, the Qualcomm Neural Network (QNN) model conversion tool is used to convert the trained audio classification model into a file format deployable in vehicles. Simultaneously, the quantization parameters are injected into an interpretable JSON file of the converted audio classification model, which is then compiled into a QNN-loadable binary file for inference. Finally, it is deployed to the vehicle's NPU backend, achieving a performance improvement of more than 10 times compared to traditional CPU inference solutions.

[0096] Please refer to Figure 5 This is a schematic diagram of the hardware structure of the electronic device 1000 provided in an embodiment of this application. Figure 5 As shown, the electronic device 1000 may include a processor 1001 and a memory 1002. The memory 1002 is used to store one or more computer programs 1003. The one or more computer programs 1003 are configured to be executed by the processor 1001. The one or more computer programs 1003 include instructions that can be used to implement the methods described above in the electronic device 1000.

[0097] It is understood that the structure illustrated in this embodiment does not constitute a specific limitation on the electronic device 1000. In other embodiments, the electronic device 1000 may include more or fewer components than illustrated, or combine some components, or split some components, or have different component arrangements.

[0098] Processor 1001 may include one or more processing units, such as: application processor (AP), modem, graphics processing unit (GPU), image signal processor (ISP), controller, video codec, digital signal processor (DSP), baseband processor, and / or neural network processing unit (NPU), etc. Different processing units may be independent devices or integrated into one or more processors.

[0099] The processor 1001 may also include a memory for storing instructions and data. In some embodiments, the memory in the processor 1001 is a cache memory. This memory can store instructions or data that the processor 1001 has just used or that are used repeatedly. If the processor 1001 needs to use the instruction or data again, it can retrieve it directly from this memory. This avoids repeated accesses, reduces the waiting time of the processor 1001, and thus improves the efficiency of the system.

[0100] In some embodiments, the processor 1001 may include one or more interfaces. Interfaces may include an inter-integrated circuit (I2C) interface, an inter-integrated circuit sound (I2S) interface, a pulse code modulation (PCM) interface, a universal asynchronous receiver / transmitter (UART) interface, a mobile industry processor interface (MIPI), a general-purpose input / output (GPIO) interface, a SIM interface, and / or a USB interface, etc.

[0101] In some embodiments, the processor 1001 is used to execute acceleration schemes such as Single Instruction Multiple Data (SIMD) and Very Long Instruction Word (VLIW).

[0102] In some embodiments, memory 1002 may include high-speed random access memory, and may also include non-volatile memory, such as hard disk, memory, plug-in hard disk, smart media card (SMC), secure digital (SD) card, flash card, at least one disk storage device, flash memory device, or other volatile solid-state storage device.

[0103] This embodiment also provides a storage medium storing computer instructions. When the instructions are executed on an electronic device, the electronic device performs the aforementioned method steps to implement the methods described in the above embodiments.

[0104] In this embodiment, the electronic device and storage medium are used to execute the corresponding methods provided above. Therefore, the beneficial effects they can achieve can be referred to the beneficial effects of the corresponding methods provided above, and will not be repeated here.

[0105] In practical applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the device can be divided into different functional modules to complete all or part of the functions described above.

[0106] In the several embodiments provided in this application, the disclosed apparatus and methods can be implemented in other ways. For example, the apparatus embodiments described above are illustrative. For instance, the division of modules or units is a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another device, or some features may be ignored or not executed. Furthermore, the mutual coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between devices or units may be electrical, mechanical, or other forms.

[0107] The unit described as a separate component may or may not be physically separate. The component shown as a unit can be one physical unit or multiple physical units, that is, it can be located in one place or distributed in multiple different places. Some or all of the units can be selected to achieve the purpose of the solution in this embodiment according to actual needs.

[0108] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0109] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a readable storage medium. Based on this understanding, the technical solutions of the embodiments of this application, essentially or in other words, the parts that contribute to the prior art, or all or part of the technical solutions, can be embodied in the form of a software product. This software product is stored in a storage medium and includes several instructions to cause a device (which may be a microcontroller, chip, etc.) or processor to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0110] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any changes or substitutions within the technical scope disclosed in this application should be covered within the scope of protection of this application.

Claims

1. A vehicle type recognition method characterized by, The method is applied to a first vehicle, the first vehicle is provided with an audio classification model, the audio classification model comprises a StarBlock module and a first activation layer, the StarBlock module comprises a depth separable convolution module, the depth separable convolution module comprises a first depth separable convolution layer and a second depth separable convolution layer, the audio classification model is used to identify a vehicle type of a second vehicle, the second vehicle is located on one side of the first vehicle; the method comprises: obtaining target audio data, the target audio data comprises sound information collected by the first vehicle in a current driving environment, and the target audio data has multiple channels; performing channel feature extraction on the target audio data based on the first depth separable convolution layer to obtain first feature data, wherein the channel dimension of the first feature data is greater than the channel dimension of the target audio data; performing dimension conversion on the first feature data to obtain second feature data, wherein the number of channels of the second feature data is the same as the number of channels of the target audio data; performing data processing on the second feature data by using an element integration method to obtain third feature data, wherein the third feature data comprises data interaction information between channels in the second feature data; performing channel feature extraction on the third feature data based on the second depth separable convolution layer to obtain fourth feature data, wherein the fourth feature data is used to represent sound features of the second vehicle; based on the fourth feature data, determining the vehicle type of the second vehicle, comprising: performing pooling processing on the fourth feature data to obtain fifth feature data; inputting the fifth feature data into the first activation layer to obtain a plurality of probability values between the fourth feature data and a plurality of vehicle labels, each vehicle label corresponds to a vehicle type, and each vehicle label corresponds to a probability value; determining a target probability value according to a plurality of probability values, and taking the vehicle type corresponding to the target probability value as the vehicle type of the second vehicle.

2. The vehicle type identification method according to claim 1, characterized by, After the data processing on the second feature data to obtain the third feature data, the method further comprises: performing full connection processing on the third feature data to obtain first processing data, wherein the number of channels of the first processing data is the same as the number of channels of the second feature data; the channel feature extraction on the third feature data to obtain the fourth feature data comprises: performing channel feature extraction on the first processing data to obtain the fourth feature data.

3. The vehicle type identification method according to claim 2, characterized by, After the full connection processing on the third feature data to obtain the first processing data, the method further comprises: performing normalization processing on the first processing data to obtain second processing data; the channel feature extraction on the first processing data to obtain the fourth feature data comprises: performing channel feature extraction on the second processing data to obtain the fourth feature data.

4. The vehicle type identification method according to claim 1, characterized by, After the channel feature extraction on the target audio data to obtain the first feature data, the method further comprises: performing normalization processing on the first feature data to obtain third processing data; The dimension conversion on the first feature data obtains second feature data, and the dimension conversion on the third processing data obtains the second feature data. The dimension conversion on the third processing data obtains the second feature data.

5. The vehicle type identification method according to claim 4, characterized by The audio classification model further comprises a second activation layer, and after the dimension conversion on the third processing data obtains the second feature data, the method further comprises: The second feature data is input into the second activation layer to obtain fourth processing data, and each data in the fourth processing data is within a preset numerical range. The data processing on the second feature data obtains third feature data, and the data processing on the fourth processing data obtains the third feature data. The target audio data is obtained by:

6. The vehicle type identification method according to claim 1, characterized by, Collecting an audio signal of the first vehicle in the current driving environment; Data processing is performed on the audio signal to obtain initial audio data; Data remodeling is performed on the initial audio data to obtain audio remodeling data; Data aggregation is performed on the audio remodeling data to obtain the target audio data. The memory is configured to store instructions, and the processor is configured to invoke the instructions in the memory to enable the electronic device to perform the vehicle type identification method according to any one of claims 1 to 6. 7.An electronic device comprising a processor and a memory, wherein The computer readable storage medium stores computer instructions, and when the computer instructions run on the electronic device, the electronic device is enabled to perform the vehicle type identification method according to any one of claims 1 to 6.

8. A computer-readable storage medium, characterized in that, ​

Images