A detection method, device, electronic device, storage medium and vehicle

Abstract

This invention discloses a detection method, apparatus, electronic device, storage medium, and vehicle. The method includes: acquiring different frame data corresponding to the current scene; inputting the different frame data into a semantic segmentation network model according to an inter-frame strategy to adaptively obtain the CFAR weights corresponding to each point in the range-Doppler spectrum; wherein, one frame of data corresponds to one range-Doppler spectrum; determining the CFAR weights of different frame data based on the CFAR weights of each point in the range-Doppler spectrum; and calculating the detection result by applying the CFAR weights of the different frame data to a CFAR detection algorithm. This method can adaptively obtain the CFAR weights of each point in the range-Doppler spectrum corresponding to different frame data through a semantic segmentation network model, making the CFAR detection process adaptive to distance and velocity, thereby improving the accuracy of the CFAR detection results.

Description

Technical Field

[0001] The present invention relates to the field of target detection technology, and in particular to a detection method, apparatus, electronic device, storage medium and vehicle. Background Technology

[0002] Constant False Alarm Rate (CFAR) detection technology has always been one of the main methods for target screening in vehicle-mounted millimeter-wave radar. It uses a sliding detection window composed of a target window, a protection window, and a training window to estimate the parameters and model the probability of background clutter. The detection threshold is dynamically adjusted according to the radar clutter data to maximize the target detection probability while keeping the false alarm rate constant.

[0003] Current main CFAR techniques include Cell Averaging CFAR (CA-CFAR), Smallest Of CFAR (SO-CFAR), Greatest Of CFAR (GO-CFAR), and Order Statistical CFAR (OS-CFAR). However, these methods either suffer from large CFAR loss, lack robustness to edge and multi-target environments, have complex processing procedures, or lack adaptability to multi-target environments. This limits the application of radar technology to some extent.

[0004] In summary, the existing millimeter-wave radar CFAR detection process lacks adaptability to range and velocity, and using the same weight for CFAR detection will lead to false alarms or missed detections. Summary of the Invention

[0005] This invention provides a detection method, device, electronic device, storage medium, and vehicle to address the problems of existing millimeter-wave radar CFAR detection processes lacking adaptability to distance and speed, and the use of the same weight for CFAR detection leading to false alarms or missed detections.

[0006] According to one aspect of the present invention, a detection method is provided, comprising:

[0007] Obtain different frame data corresponding to the current scene;

[0008] The different frame data are input into the semantic segmentation network model according to the frame-separated strategy to adaptively obtain the CFAR weights corresponding to each point in the range-Doppler spectrum; wherein, one frame of data corresponds to one range-Doppler spectrum;

[0009] The CFAR weights of different frame data are determined based on the CFAR weights of each point in the range Doppler spectrum.

[0010] The CFAR weights of the different frame data are calculated according to the CFAR detection algorithm to obtain the detection result.

[0011] According to another aspect of the present invention, a detection device is provided, comprising:

[0012] The acquisition module is used to acquire different frame data corresponding to the current scene;

[0013] The input module is used to input the different frame data into the semantic segmentation network model according to the frame-separated strategy, and adaptively obtain the CFAR weights corresponding to each point in the range-Doppler spectrum; wherein, one frame of data corresponds to one range-Doppler spectrum;

[0014] The determination module is used to determine the CFAR weights of different frame data based on the CFAR weights of each point in the distance Doppler spectrum;

[0015] The calculation module is used to calculate the CFAR weights of the different frame data according to the CFAR detection algorithm to obtain the detection result.

[0016] According to another aspect of the present invention, an electronic device is provided, the electronic device comprising: at least one processor; and

[0017] A memory communicatively connected to the at least one processor; wherein,

[0018] The memory stores a computer program that can be executed by the at least one processor, the computer program being executed by the at least one processor to enable the at least one processor to perform the detection method described in any embodiment of the present invention.

[0019] According to another aspect of the present invention, a computer-readable storage medium is provided, the computer-readable storage medium storing computer instructions for causing a processor to execute and implement the detection method described in any embodiment of the present invention.

[0020] According to another aspect of the present invention, a vehicle is provided, the vehicle including the electronic equipment described in the embodiments of the present invention and a plurality of sensors.

[0021] The technical solution of this invention involves acquiring different frame data corresponding to the current scene; inputting the different frame data into a semantic segmentation network model according to an inter-frame strategy to adaptively obtain the CFAR weights corresponding to each point in the range-Doppler spectrum; wherein, one frame of data corresponds to one range-Doppler spectrum; determining the CFAR weights of different frame data based on the CFAR weights of each point in the range-Doppler spectrum; and calculating the detection results by using the CFAR weights of different frame data according to the CFAR detection algorithm. This solves the problem that the existing millimeter-wave radar CFAR detection process lacks adaptability to distance and velocity, and that using the same weight for CFAR detection will lead to false alarms or missed detections. It achieves the beneficial effect of making the CFAR detection process adaptable to distance and velocity, thereby improving the accuracy of CFAR detection results.

[0022] It should be understood that the description in this section is not intended to identify key or essential features of the embodiments of the present invention, nor is it intended to limit the scope of the invention. Other features of the invention will become readily apparent from the following description. Attached Figure Description

[0023] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0024] Figure 1 This is a schematic flowchart of a detection method provided in Embodiment 1 of the present invention;

[0025] Figure 2 This is a schematic diagram illustrating multi-vehicle joint information acquisition provided in an embodiment of the present invention;

[0026] Figure 3 This is a schematic flowchart of a detection method provided in Embodiment 2 of the present invention;

[0027] Figure 4 This is a schematic flowchart of a detection method provided in an exemplary embodiment of the present invention;

[0028] Figure 5 This is a schematic diagram of the structure of a detection device provided in Embodiment 3 of the present invention;

[0029] Figure 6 This is a schematic diagram of the electronic device used in the detection method according to an embodiment of the present invention;

[0030] Figure 7 This is a schematic diagram of the structure of a vehicle provided in Embodiment 5 of the present invention. Detailed Implementation

[0031] To enable those skilled in the art to better understand the present invention, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are merely some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention. It should be understood that the various steps described in the method embodiments of the present invention can be performed in different orders and / or in parallel. Furthermore, the method embodiments may include additional steps and / or omit the steps shown. The scope of the present invention is not limited in this respect.

[0032] The term "comprising" and its variations as used herein are open-ended inclusions, meaning "including but not limited to". The term "based on" means "at least partially based on". The term "one embodiment" means "at least one embodiment"; the term "another embodiment" means "at least one additional embodiment"; the term "some embodiments" means "at least some embodiments". Definitions of other terms will be given in the description below.

[0033] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of the invention described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0034] It should be noted that the terms "a" and "a plurality of" used in this invention are illustrative rather than restrictive. Those skilled in the art should understand that, unless otherwise expressly indicated in the context, they should be understood as "one or more".

[0035] The names of the messages or information exchanged between the multiple devices in the embodiments of the present invention are for illustrative purposes only and are not intended to limit the scope of these messages or information.

[0036] It is understood that the data involved in this technical solution (including but not limited to the data itself, the acquisition or use of the data) shall comply with the requirements of relevant laws, regulations and related provisions.

[0037] Example 1

[0038] During vehicle detection, vehicle-mounted millimeter-wave radar encounters numerous diverse scenarios with many uncertainties. These uncertainties primarily include: road vehicle uncertainties (e.g., traffic flow, distance, direction, and vehicle shape); lane uncertainties (e.g., road surface smoothness, slope, width, clutter reflection intensity, and road curvature); weather uncertainties (e.g., sunny, rainy, foggy, and snowy conditions); air humidity and temperature uncertainties; light intensity uncertainties; and other unforeseen events such as pedestrians crossing the road, landslides, and various traffic incidents. These diverse scenarios present significant challenges to radar detection capabilities.

[0039] Figure 1 This is a flowchart illustrating a detection method provided in Embodiment 1 of the present invention. This method is applicable to multi-object detection, particularly suitable for multi-object detection while a vehicle is traveling on a road. The method can be executed by a detection device, which can be implemented in software and / or hardware and is generally integrated into an electronic device. In this embodiment, the electronic device includes, but is not limited to, an onboard controller.

[0040] like Figure 1 As shown, the detection method provided in Embodiment 1 of the present invention includes the following steps:

[0041] S110. Obtain different frame data corresponding to the current scene.

[0042] Here, the current scene can be understood as the vehicle's current driving scenario. Frame data can be the pulses within each frame of the range-Doppler spectrum (RDM). The range-Doppler spectrum can be obtained by processing the raw signal collected by the vehicle-mounted millimeter-wave radar. By processing the multiple cycles of chirp sequences and echo information sent by the radar in both fast and slow time dimensions, the range-Doppler spectrum can be obtained, and thus the range and velocity information of multiple targets can be extracted.

[0043] In this embodiment, the original signal data collected by the vehicle-mounted millimeter-wave radar in the current scene can be obtained. The original signal data can be processed by frequency mixing and two-dimensional Fourier transform to obtain the range Doppler spectrum, and then different frame data can be obtained from the range Doppler spectrum.

[0044] S120. Input the different frame data into the semantic segmentation network model according to the frame-separated strategy, and adaptively obtain the CFAR weights corresponding to each point in the distance Doppler spectrum.

[0045] One frame of data corresponds to one range Doppler spectrum.

[0046] In this embodiment, the frame-skipping strategy may include acquiring frame data once every preset number of frames. The preset number can be set according to the computing power of the electronic device and is not specifically limited here. For example, if there are 50 frames, one frame can be acquired every 9 frames, i.e., the 0th frame, the 10th frame, the 20th frame, ..., the 50th frame can be acquired. These 6 acquired frames are then input into the semantic segmentation network model to calculate the CFAR weights corresponding to each frame.

[0047] In this embodiment, the semantic segmentation network model is a convolutional neural network (CNN) model obtained after model training. Semantic segmentation technology is an important subclass of CNN and has been widely used in image-level autonomous driving in recent years. It is a natural step of coarse-to-fine reasoning, which can provide not only predicted categories but also pixel-level spatial locations, achieving fine-grained reasoning by performing dense prediction and inference of labels for each pixel. The semantic segmentation network model can be DeepLabv3+.

[0048] In this embodiment, semantic segmentation training of the CFAR detection parameters of the range Doppler spectrum is performed based on a large amount of data from different scenarios. The aim is to achieve adaptive acquisition of the CFAR weight parameters of the RDM in different scenarios, rather than fixing them, so as to minimize the impact of various background noises caused by different scenarios.

[0049] It is understandable that after inputting different frame data obtained through the inter-frame strategy into the semantic segmentation network model, the CFAR weights of each point in the range-Doppler spectrum corresponding to different frame data can be output. The CFAR weights of each point in the range-Doppler spectrum obtained by inputting frame data from different scenarios into the semantic segmentation network model are different and not fixed values. S130, Determine the CFAR weights of different frame data based on the CFAR weights of each point in the range-Doppler spectrum.

[0050] In this embodiment, in order to prevent abrupt changes in CFAR weights, when calculating the CFAR weights of the current frame, the calculation results of the previous k calculations are added to the CFAR weights of each point in the distance-Doppler spectrum corresponding to the current frame, and the mean is calculated.

[0051] The calculation results of the first k times can be understood as the CFAR weight calculation results of the first k frames of data. The first k frames of data can be frame data obtained through the inter-frame strategy. For example, if the current frame is the 40th frame of data, the first k frames of data can include the 0th frame of data, the 10th frame of data, the 20th frame of data, and the 30th frame of data.

[0052] It should be noted that the CFAR weights of the first k frames are calculated in the same way as the CFAR weights of the current frame. Since environmental changes are slow, the CFAR weights of the current frame can be carried over to subsequent frames, allowing for further adjustments based on computing power and the required frame rate. For example, the CFAR weights of frame 0 can be carried over to frames 1 through 9, meaning the CFAR weights of frames 1 and 9 are the same as those of frame 0.

[0053] S140. The CFAR weights of the different frame data are calculated according to the CFAR detection algorithm to obtain the detection result.

[0054] The detection results include whether there is a target object, and there are no restrictions on the specific form of the target object.

[0055] In this embodiment, the detection map can be obtained by substituting the CFAR weights of different frame data into the traditional CA-CFAR detection algorithm. The traditional CA-CFAR detection algorithm typically first determines the region of the reference cell Tr of the point to be detected on the detection map; then it calculates the mean of this region, adds the offset to the mean to obtain T; finally, it compares T with the value of the point to be detected. The comparison is performed; if the value of the point to be detected is greater than T, then the point to be detected is determined to be the target point. The specific formula is as follows:

[0056]

[0057] Where N represents the number of points to be detected.

[0058] Finally, the detection result for each point can be described as follows:

[0059]

[0060] Here, CUT represents the unit being detected.

[0061] The detection method provided in Embodiment 1 of this invention first acquires different frame data corresponding to the current scene; then, the different frame data is input into a semantic segmentation network model according to an inter-frame strategy to adaptively obtain the CFAR weights corresponding to each point in the range-Doppler spectrum; wherein, one frame of data corresponds to one range-Doppler spectrum; then, the CFAR weights of different frame data are determined based on the CFAR weights of each point in the range-Doppler spectrum; finally, the CFAR weights of the different frame data are calculated according to the CFAR detection algorithm to obtain the detection result. The above method can adaptively obtain the CFAR weights of each point in the range-Doppler spectrum corresponding to different frame data through the semantic segmentation network model, making the CFAR detection process adaptive to distance and velocity, thereby improving the accuracy of the CFAR detection results.

[0062] Based on the above embodiments, modified embodiments of the above embodiments are proposed. It should be noted that, in order to keep the description brief, only the differences from the above embodiments are described in the modified embodiments.

[0063] In one embodiment, the parameters of the semantic segmentation network model are obtained after training based on the dataset and sample labels, and the feature map size of the last layer of the semantic segmentation network model is the same as the size of the distance Doppler spectrum.

[0064] Furthermore, the methods for obtaining the dataset include:

[0065] Acquire raw signal data collected by vehicle-mounted millimeter-wave radar in various scenarios;

[0066] The original signal data is processed by frequency mixing and two-dimensional Fourier transform to obtain the range Doppler spectrum;

[0067] A dataset is constructed based on the matrix data of the range-Doppler spectrum.

[0068] In this embodiment, when the vehicle is traveling in different scenarios such as elevated roads, plains, roads, or congested sections, it can receive raw signal data collected by the millimeter-wave radar from the onboard millimeter-wave radar. The millimeter-wave radar is mounted on the vehicle body, and the number of millimeter-wave radars is not limited here.

[0069] The specific process of mixing the original signal and obtaining the range Doppler spectrum through two-dimensional Fourier transform will not be elaborated here.

[0070] Understandably, matrix data from multiple RDMs can be grouped into a single dataset, meaning the dataset includes matrix data from multiple range-Doppler spectra.

[0071] Furthermore, the methods for obtaining the sample labels include:

[0072] The dataset was processed using the CFAR detection algorithm to obtain detection results corresponding to different matrix data.

[0073] Vehicle information is acquired in various scenarios. The vehicle information is obtained by combining multiple vehicles through sensors installed on the vehicle body. The vehicle information includes road information, location information and speed information.

[0074] The detection results corresponding to the different matrix data are corrected based on the vehicle information to obtain the correct detection results, and the correct detection results are used as sample labels.

[0075] In this embodiment, the matrix data in the dataset is calculated using the traditional CFAR detection algorithm to obtain detection results for different scenarios. These detection results can include whether a target object exists in the current scene.

[0076] It should be noted that the detection results obtained by the traditional CFAR detection algorithm are not necessarily accurate. Therefore, the detection results can be further corrected by combining vehicle information obtained from multiple vehicles to obtain more accurate detection results.

[0077] Vehicle information can include road information, location information, and speed information. Road information can be collected by cameras mounted on the vehicle, location information by radar sensors mounted on the vehicle, and speed information by speed sensors mounted on the vehicle. Multi-vehicle collaboration can be understood as combining vehicle information collected from multiple vehicles to obtain a single information set. Figure 2 This is a schematic diagram illustrating multi-vehicle joint information acquisition provided in an embodiment of the present invention, as shown below. Figure 2 As shown, it can combine multiple vehicles to obtain road information, precise coordinates of multiple vehicles, and speed information.

[0078] In this embodiment, the detection results can be corrected by combining the vehicle information set obtained from multiple vehicles, and the corrected detection results can be used as sample labels for model training.

[0079] Furthermore, the training process of the semantic segmentation network model includes:

[0080] Input the dataset and sample labels into the semantic segmentation network model;

[0081] The CFAR weights of each point in the Doppler spectrum are obtained by normalizing the data in the dataset using the semantic segmentation network model.

[0082] The CFAR weights of each point in the Doppler spectrum are calculated using the CFAR detection algorithm to obtain the detection result of each matrix data in the dataset;

[0083] Calculate the loss value based on the detection result of each matrix data and the sample label;

[0084] The model parameters are adjusted based on the loss value, and the semantic segmentation network model with adjusted parameters is trained again until the semantic segmentation network model is fitted.

[0085] In this embodiment, after inputting the dataset and sample labels into the semantic segmentation network model, the model can normalize the data in the dataset to obtain the CFAR weights for each point on the Doppler spectrum. At this point, the model parameters of the semantic segmentation network model are the initial values.

[0086] Specifically, after inputting the dataset into the semantic segmentation network model, the CFAR weights of each point in the Doppler spectrum can be output. Substituting these CFAR weights into the traditional CFAR detection algorithm yields the detection results for each matrix data in the dataset. The loss function of the model is calculated based on the label (sample label) of each matrix data in the dataset and the detection results. The model parameters are then modified through backpropagation based on the loss function until the model is well-fitted. At this point, the parameters of the semantic segmentation network model are different from the parameter values ​​of the semantic segmentation network model before model training.

[0087] Among them, when the detection results obtained by substituting the CFAR weights of each point of the Doppler spectrum output by the model into the CFAR detection algorithm are mostly consistent with the sample labels, the characterization is considered to be well-fitted.

[0088] In this embodiment, the parameters of the semantic segmentation network model are trained using data from multiple scenarios. This enables the semantic segmentation network model to adaptively output the CFAR weights of each point in the distance-Doppler spectrum corresponding to the frame data of different scenarios, avoiding the CFAR detection compatibility problem of different scenarios and making CFAR detection applicable to the detection of different scenarios.

[0089] Example 2

[0090] Figure 3 This is a schematic flowchart of a detection method provided in Embodiment 2 of the present invention. Embodiment 2 is an optimization based on the above embodiments. For details not covered in this embodiment, please refer to Embodiment 1.

[0091] like Figure 3 As shown, the detection method provided in Embodiment 2 of the present invention includes the following steps:

[0092] S210. Obtain different frame data corresponding to the current scene.

[0093] S220. From the different frame data, obtain frame data once at a preset number of frame data intervals, input the obtained current frame data into the semantic segmentation network model, and adaptively obtain the CFAR weights corresponding to each point in the distance Doppler spectrum.

[0094] One frame of data corresponds to one range Doppler spectrum.

[0095] S230. For the current frame data, the CFAR weight of the current frame data at the corresponding point in the Doppler spectrum is added to the CFAR weight calculation results of the previous k calculations, and the average value is calculated to obtain the CFAR weight of the current frame data; the CFAR weight of the current frame data is used as the CFAR weight of the preset number of frames.

[0096] The CFAR weight calculation results of the first k times include the CFAR weight calculation results of k frames of data, and the interval between two adjacent frames of data in the k frames of data is the preset number of frames of data.

[0097] It is understandable that the current frame data can be the data of the currently acquired frame, for example, the current frame data can be the 30th frame data.

[0098] In step S230, the parameters of the semantic segmentation network model have been trained, which can be understood as the semantic segmentation network model having completed model training.

[0099] For example, if the preset number is 9, then the frame data is obtained once every 9 frames, that is, the CFAR weight is recalculated only once every 9 frames. This can be understood as the CFAR weight being calculated only for the 0th frame, the 10th frame, the 20th frame, and so on, while the CFAR weight is not calculated for other frames.

[0100] The calculation process of the CFAR weights for the current frame data is illustrated as an example. If the current frame data is the 30th frame data, then the 30th frame data is input into the semantic segmentation network model, and the CFAR weights of the corresponding points in the distance-Doppler spectrum of the current frame data are output. The CFAR weights of the current frame data at the corresponding points in the distance-Doppler spectrum are then added together with the CFAR weights calculated for the 20th, 10th, and 0th frames, and the average value is taken to obtain the CFAR weight calculation result for the current frame data. The specific calculation formula is as follows:

[0101]

[0102] Where i represents the i-th frame of data; This represents the CFAR weight of the i-th frame data input into the semantic segmentation network model, which corresponds to the point in the distance-Doppler spectrum of the current frame data. This represents the CFAR weight calculation results for the first k iterations.

[0103] It should be noted that the CFAR weight calculation result of the 30th frame data can be used as the weight calculation result of the 31st to 39th frames data.

[0104] In this embodiment, when a change in road conditions is detected, the CFAR weights of the frame data need to be recalculated. Whether the road conditions have changed depends on the combined results of various onboard sensors, such as video and images captured by cameras, signal waveforms captured by radar, and GPS positioning information.

[0105] S240. The CFAR weights of the different frame data are calculated according to the CFAR detection algorithm to obtain the detection result.

[0106] The second embodiment of the present invention provides a detection method that specifies the frame-interval strategy and the process of determining the CFAR weight of the current frame data. The frame-interval strategy used in this method can greatly reduce the amount of computation; the way this method calculates the CFAR weight of the current frame data can greatly improve the accuracy of the calculation results.

[0107] Based on the technical solutions of the above embodiments, this invention provides a specific implementation method.

[0108] As one specific implementation method of this embodiment. Figure 4 This is a schematic flowchart of a detection method provided in an example embodiment of the present invention, such as... Figure 4 As shown, the training of the semantic segmentation network model parameters in offline mode includes: obtaining coarse data by using CFAR detection on the acquired dataset, i.e., obtaining detection results corresponding to different matrix data by using the CFAR detection algorithm on the dataset; obtaining precise raw data through multi-vehicle and multi-machine joint processing, i.e., obtaining vehicle information obtained by multi-vehicle joint processing through sensors installed on the vehicle body in various scenarios; manually adjusting the two results to generate sample labels, i.e., correcting the detection results corresponding to the different matrix data according to the vehicle information to obtain the correct detection results, and using the correct detection results as sample labels; inputting the dataset and sample labels into the semantic segmentation network model, adjusting the feature of the last layer of the model to be consistent with the RDM size, iterating the model training until it is fitted, and outputting the weights, i.e., the CFAR weights of each point of the Doppler spectrum.

[0109] The semantic segmentation network model and offline-trained model parameters are downloaded online. Multiple frames of data are acquired according to the initial frame-interval condition (i.e., the frame-interval strategy). The average weight of CFAR across multiple frames is calculated by adding the CFAR weight of the current frame data at the corresponding point in the Doppler spectrum to the average of the CFAR weights calculated in the previous k iterations. The CFAR detection result of the current frame is then calculated based on this weight. The system then determines whether the vehicle should continue driving. If so, it checks whether the road conditions have changed or whether the frame interval has reached its maximum. If so, multiple frames are reacquired, and the average weight of CFAR across multiple frames is calculated. If not, the CFAR detection result of the current frame is calculated based on the weight.

[0110] Example 3

[0111] Figure 5 This is a schematic diagram of a detection device provided in Embodiment 3 of the present invention. The device is applicable to the detection of multiple targets, especially to the detection of multiple targets when a vehicle is driving on the road. The device can be implemented by software and / or hardware and is generally integrated into an electronic device.

[0112] like Figure 5 As shown, the device includes: an acquisition module 110, an input module 120, a determination module 130, and a calculation module 140.

[0113] The acquisition module 110 is used to acquire different frame data corresponding to the current scene;

[0114] Input module 120 is used to input the different frame data into the semantic segmentation network model according to the frame-separated strategy, and adaptively obtain the CFAR weights corresponding to each point in the range Doppler spectrum; wherein, one frame of data corresponds to one range Doppler spectrum;

[0115] The determination module 130 is used to determine the CFAR weights of different frame data based on the CFAR weights of each point in the distance Doppler spectrum;

[0116] The calculation module 140 is used to calculate the CFAR weights of the different frame data according to the CFAR detection algorithm to obtain the detection result.

[0117] In this embodiment, the device first acquires different frame data corresponding to the current scene through the acquisition module 110; then, the input module 120 inputs the different frame data into the semantic segmentation network model according to the frame-by-frame strategy to adaptively obtain the CFAR weights corresponding to each point in the range-Doppler spectrum; wherein, one frame of data corresponds to one range-Doppler spectrum; then, the determination module 130 determines the CFAR weights of different frame data based on the CFAR weights of each point in the range-Doppler spectrum; finally, the calculation module 140 calculates the CFAR weights of the different frame data according to the CFAR detection algorithm to obtain the detection result.

[0118] This embodiment provides a detection device that enables the CFAR detection process to be adaptive to distance and speed, thereby improving the accuracy of CFAR detection results.

[0119] Furthermore, the parameters of the semantic segmentation network model are obtained after training based on the dataset and sample labels, and the dataset is obtained in the following ways:

[0120] Acquire raw signal data collected by vehicle-mounted millimeter-wave radar in various scenarios;

[0121] The original signal data is processed by frequency mixing and two-dimensional Fourier transform to obtain the range Doppler spectrum;

[0122] A dataset is constructed based on the matrix data of the range-Doppler spectrum.

[0123] Based on the above optimizations, the methods for obtaining the sample labels include:

[0124] The dataset was processed using the CFAR detection algorithm to obtain detection results corresponding to different matrix data.

[0125] Vehicle information is acquired in various scenarios. The vehicle information is obtained by combining multiple vehicles through sensors installed on the vehicle body. The vehicle information includes road information, location information and speed information.

[0126] The detection results corresponding to the different matrix data are corrected based on the vehicle information to obtain the correct detection results, and the correct detection results are used as sample labels.

[0127] Furthermore, the training process of the semantic segmentation network model includes:

[0128] Input the dataset and sample labels into the semantic segmentation network model;

[0129] The CFAR weights of each point in the Doppler spectrum are obtained by normalizing the data in the dataset using the semantic segmentation network model.

[0130] The CFAR weights of each point in the Doppler spectrum are calculated using the CFAR detection algorithm to obtain the detection result of each matrix data in the dataset;

[0131] Calculate the loss value based on the detection result of each matrix data and the sample label;

[0132] The model parameters are adjusted based on the loss value, and the semantic segmentation network model with adjusted parameters is trained again until the semantic segmentation network model is fitted.

[0133] Furthermore, in the semantic segmentation network model, the feature map size of the last layer of the network is the same as the size of the distance Doppler spectrum.

[0134] Furthermore, the input module 120 is specifically used to: acquire frame data once at a preset number of frame data intervals from the different frame data, and input the acquired current frame data into the semantic segmentation network model.

[0135] Based on the above optimization, the determining module 130 is specifically used to: for the current frame data, add the CFAR weight of the current frame data at the corresponding point in the Doppler spectrum to the CFAR weight calculation results of the previous k times, and calculate the average value to obtain the CFAR weight of the current frame data; wherein, the CFAR weight calculation results of the previous k times include the CFAR weight calculation results of k frames, and the interval between two adjacent frames in the k frames is the preset number of frames; and use the CFAR weight of the current frame data as the CFAR weight of the preset number of frames.

[0136] The above-described detection device can execute the detection method provided in any embodiment of the present invention, and has the corresponding functional modules and beneficial effects for executing the method.

[0137] Example 4

[0138] Figure 6 A schematic diagram of an electronic device 10, which can be used to implement embodiments of the present invention, is shown. The electronic device is intended to represent, for example, a control device for an in-vehicle controller. The components shown herein, their connections and relationships, and their functions are merely examples and are not intended to limit the implementation of the invention described and / or claimed herein.

[0139] like Figure 6As shown, the electronic device 10 includes at least one processor 11 and a memory, such as a read-only memory (ROM) 12 or a random access memory (RAM) 13, communicatively connected to the at least one processor 11. The memory stores computer programs executable by the at least one processor. The processor 11 can perform various appropriate actions and processes based on the computer program stored in the ROM 12 or loaded from storage unit 18 into the RAM 13. The RAM 13 can also store various programs and data required for the operation of the electronic device 10. The processor 11, ROM 12, and RAM 13 are interconnected via a bus 14. An input / output (I / O) interface 15 is also connected to the bus 14.

[0140] Multiple components in electronic device 10 are connected to I / O interface 15, including: input unit 16, such as keyboard, mouse, etc.; output unit 17, such as various types of displays, speakers, etc.; storage unit 18, such as disk, optical disk, etc.; and communication unit 19, such as network card, modem, wireless transceiver, etc. Communication unit 19 allows electronic device 10 to exchange information / data with other devices through computer networks such as the Internet and / or various telecommunications networks.

[0141] Processor 11 can be a variety of general-purpose and / or special-purpose processing components with processing and computing capabilities. Some examples of processor 11 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various special-purpose artificial intelligence (AI) computing chips, various processors running machine learning model algorithms, a digital signal processor (DSP), and any suitable processor, controller, microcontroller, etc. Processor 11 performs the various methods and processes described above, such as detection methods.

[0142] In some embodiments, the detection method may be implemented as a computer program tangibly contained in a computer-readable storage medium, such as storage unit 18. In some embodiments, part or all of the computer program may be loaded and / or installed on electronic device 10 via ROM 12 and / or communication unit 19. When the computer program is loaded into RAM 13 and executed by processor 11, one or more steps of the detection method described above may be performed. Alternatively, in other embodiments, processor 11 may be configured to perform the detection method by any other suitable means (e.g., by means of firmware).

[0143] Various embodiments of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field-programmable gate arrays (FPGAs), application-specific integrated circuits (ASICs), application-specific standard products (ASSPs), systems-on-a-chip (SoCs), payload-programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments may include implementations in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which may be a dedicated or general-purpose programmable processor, capable of receiving data and instructions from a storage system, at least one input device, and at least one output device, and transmitting data and instructions to the storage system, the at least one input device, and the at least one output device.

[0144] Computer programs used to implement the methods of the present invention may be written in any combination of one or more programming languages. These computer programs may be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device, such that when executed by the processor, the computer programs cause the functions / operations specified in the flowcharts and / or block diagrams to be performed. The computer programs may be executed entirely on a machine, partially on a machine, or as a standalone software package, partially on a machine and partially on a remote machine, or entirely on a remote machine or server.

[0145] In the context of this invention, a computer-readable storage medium can be a tangible medium that may contain or store a computer program for use by or in conjunction with an instruction execution system, apparatus, or device. A computer-readable storage medium may include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination thereof. Alternatively, a computer-readable storage medium may be a machine-readable signal medium. More specific examples of machine-readable storage media include electrical connections based on one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fibers, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof.

[0146] To provide interaction with a user, the systems and techniques described herein can be implemented on an electronic device having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and pointing device (e.g., a mouse or trackball) through which the user provides input to the electronic device. Other types of devices can also be used to provide interaction with the user; for example, feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including sound input, voice input, or tactile input).

[0147] The systems and technologies described herein can be implemented in computing systems that include backend components (e.g., as data servers), or middleware components (e.g., application servers), or frontend components (e.g., user computers with graphical user interfaces or web browsers through which users can interact with implementations of the systems and technologies described herein), or any combination of such backend, middleware, or frontend components. The components of the system can be interconnected via digital data communication of any form or medium (e.g., communication networks). Examples of communication networks include local area networks (LANs), wide area networks (WANs), blockchain networks, and the Internet.

[0148] A computing system can include clients and servers. Clients and servers are generally located far apart and typically interact through communication networks. The client-server relationship is created by computer programs running on the respective computers and having a client-server relationship with each other. The server can be a cloud server, also known as a cloud computing server or cloud host, which is a hosting product within the cloud computing service system to address the shortcomings of traditional physical hosts and VPS services, such as high management difficulty and weak business scalability.

[0149] Example 5

[0150] Figure 7 This is a structural schematic diagram of a vehicle provided in Embodiment 5 of the present invention, as shown below. Figure 7 As shown, the vehicle includes the electronic equipment provided in Embodiment 4 of the present invention, enabling the vehicle to perform the detection method provided in any embodiment of the present invention.

[0151] The vehicle includes multiple onboard sensors, such as cameras, radar sensors, and speed sensors.

[0152] The above-mentioned vehicles can perform the detection method provided in any embodiment of the present invention and have the corresponding beneficial effects of performing the method.

[0153] It should be understood that the various forms of processes shown above can be used, with steps reordered, added, or deleted. For example, the steps described in this invention can be executed in parallel, sequentially, or in different orders, as long as the desired result of the technical solution of this invention can be achieved, and this is not limited herein.

[0154] The specific embodiments described above do not constitute a limitation on the scope of protection of this invention. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this invention should be included within the scope of protection of this invention.

Claims

1. A method of detection, characterized in that, The method includes: Obtain different frame data corresponding to the current scene; The different frame data are input into the semantic segmentation network model according to the frame-separated strategy to adaptively obtain the CFAR weights corresponding to each point in the range-Doppler spectrum; wherein, one frame of data corresponds to one range-Doppler spectrum; The CFAR weights of different frame data are determined based on the CFAR weights of each point in the range Doppler spectrum. The CFAR weights of the different frame data are calculated according to the CFAR detection algorithm to obtain the detection result; The step of determining the CFAR weights of different frame data based on the CFAR weights of each point in the distance Doppler spectrum includes: For the current frame data, the CFAR weight of the current frame data at the corresponding point in the Doppler spectrum is added to the CFAR weight calculation results of the previous k times, and the average value is calculated to obtain the CFAR weight of the current frame data; wherein, the CFAR weight calculation results of the previous k times include the CFAR weight calculation results of k frames, and the interval between two adjacent frames in the k frames is a preset number of frames; the CFAR weight of the current frame data is used as the CFAR weight of the preset number of frames.

2. The method of claim 1, wherein, The parameters of the semantic segmentation network model are obtained after training based on the dataset and sample labels. The dataset is obtained in the following ways: Acquire raw signal data collected by vehicle-mounted millimeter-wave radar in various scenarios; The original signal data is processed by frequency mixing and two-dimensional Fourier transform to obtain the range Doppler spectrum; A dataset is constructed based on the matrix data of the range-Doppler spectrum.

3. The method according to claim 2, characterized in that, The methods for obtaining the sample labels include: The dataset was processed using the CFAR detection algorithm to obtain detection results corresponding to different matrix data. Vehicle information is acquired in various scenarios. The vehicle information is obtained by combining multiple vehicles through sensors installed on the vehicle body. The vehicle information includes road information, location information and speed information. The detection results corresponding to the different matrix data are corrected based on the vehicle information to obtain the correct detection results, and the correct detection results are used as sample labels.

4. The method according to claim 1, characterized in that, The training process of the semantic segmentation network model includes: Input the dataset and sample labels into the semantic segmentation network model; The CFAR weights of each point in the Doppler spectrum are obtained by normalizing the data in the dataset using the semantic segmentation network model. The CFAR weights of each point in the Doppler spectrum are calculated using the CFAR detection algorithm to obtain the detection result of each matrix data in the dataset; Calculate the loss value based on the detection result of each matrix data and the sample label; The model parameters are adjusted based on the loss value, and the semantic segmentation network model with adjusted parameters is trained again until the semantic segmentation network model is fitted.

5. The method according to claim 2, characterized in that, In the semantic segmentation network model, the feature map size of the last layer of the network is the same as the size of the distance Doppler spectrum.

6. The method according to claim 1, characterized in that, The different frame data are input into the semantic segmentation network model according to the frame-interval strategy, including: From the different frame data, frame data is obtained once every preset number of frame data intervals, and the obtained current frame data is input into the semantic segmentation network model.

7. A detection device, characterized in that, The device includes: The acquisition module is used to acquire different frame data corresponding to the current scene; The input module is used to input the different frame data into the semantic segmentation network model according to the frame-separated strategy, and adaptively obtain the CFAR weights corresponding to each point in the range-Doppler spectrum; wherein, one frame of data corresponds to one range-Doppler spectrum; The determination module is used to determine the CFAR weights of different frame data based on the CFAR weights of each point in the distance Doppler spectrum; The calculation module is used to calculate the CFAR weights of the different frame data according to the CFAR detection algorithm to obtain the detection result; The determining module is specifically used to calculate the average value of the CFAR weight of the current frame data by adding the CFAR weight calculation results of the previous k CFAR weight calculations to the corresponding point in the Doppler spectrum of the current frame data; wherein, the CFAR weight calculation results of the previous k CFAR weight calculations include the CFAR weight calculation results of k frames of data, and the interval between two adjacent frames of data in the k frames of data is a preset number of frames of data; the CFAR weight of the current frame data is used as the CFAR weight of the preset number of frames of data.

8. An electronic device, characterized in that, The electronic device includes: At least one processor; and A memory communicatively connected to the at least one processor; wherein, The memory stores a computer program that can be executed by the at least one processor, the computer program being executed by the at least one processor to enable the at least one processor to perform the detection method according to any one of claims 1-6.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer instructions that cause a processor to execute the detection method according to any one of claims 1-6.

10. A vehicle, characterized in that, The vehicle includes the electronic equipment and multiple sensors as described in claim 8.

Images