RFID Finger Tracking Recognition Method and Device Based on Few-Sample Learning

Abstract

This invention discloses an RFID finger tracking identification method based on few-sample learning. First, the method acquires the original finger reflection signal with temporal characteristics sent by the tag array. Next, the original finger reflection signal is used to extract features to obtain a visualized finger tracking spatial feature map. Finally, a tracking identification model is constructed, and the finger tracking spatial feature map is input into the tracking identification model to complete the identification of the finger tracking spatial feature map. Thus, high-precision, fine-grained finger tracking identification can be achieved without the need for the user to wear any device and with very few training samples.

Description

Technical Field

[0001] This invention relates to the field of finger tracking technology, and particularly to an RFID finger tracking recognition method based on few-shot learning, a computer-readable storage medium, a computer device, and an RFID finger tracking recognition device based on few-shot learning. Background Technology

[0002] In related technologies, finger tracking is generally identified using sensors, cameras, and other devices. However, traditional finger tracking recognition methods based on sensors or cameras are highly dependent on hardware, and the quality of the hardware affects its accuracy. In addition, when the surrounding environment changes, such as insufficient light intensity, the accuracy of traditional methods cannot be guaranteed. Furthermore, user privacy is easily violated, as the camera will completely record the user's behavior, and there is also a risk of leakage of the user's facial information. Summary of the Invention

[0003] The present invention aims to at least partially solve one of the technical problems in the aforementioned technologies. Therefore, one objective of the present invention is to propose an RFID finger tracking recognition method based on few-sample learning, which can achieve high-precision, fine-grained finger tracking recognition without requiring any device worn by the user and with very few training samples.

[0004] A second objective of this invention is to provide a computer-readable storage medium.

[0005] The third objective of this invention is to provide a computer device.

[0006] The fourth objective of this invention is to propose an RFID finger tracking recognition device based on few-shot learning.

[0007] To achieve the above objectives, a first aspect of the present invention proposes an RFID finger tracking identification method based on few-shot learning, comprising: acquiring raw finger reflection signals with temporal characteristics sent by a tag array; extracting features from the raw finger reflection signals to obtain a visualized finger tracking spatial feature map; constructing a tracking identification model and inputting the finger tracking spatial feature map into the tracking identification model to complete the identification of the finger tracking spatial feature map.

[0008] According to the RFID finger tracking identification method based on few-sample learning according to embodiments of the present invention, firstly, the original finger reflection signal with temporal characteristics sent by the tag array is acquired; then, the original finger reflection signal is subjected to feature extraction to obtain a visualized finger tracking spatial feature map; finally, a tracking identification model is constructed, and the finger tracking spatial feature map is input into the tracking identification model to complete the identification of the finger tracking spatial feature map; thus, high-precision, fine-grained finger tracking identification can be achieved without the need for the human body to wear any device and with very few training samples.

[0009] In addition, the RFID finger tracking recognition method based on few-sample learning proposed in the above embodiments of the present invention may also have the following additional technical features:

[0010] Optionally, feature extraction is performed on the original finger reflection signal to obtain a visualized finger trace spatial feature map, including: correcting the phase of the original finger reflection signal using one-dimensional phase unpacking, and smoothing the original finger reflection signal using a filter to obtain a preprocessed signal; segmenting the preprocessed signal to obtain the start and end points of the finger trace; obtaining a likelihood estimation matrix composed of the likelihoods corresponding to all tags in the tag array, so as to track the finger movement trend based on the likelihood estimation matrix and the start and end points of the finger trace, and plotting it into the visualized finger trace spatial feature map.

[0011] Optionally, the preprocessed signal is segmented to obtain the start and end points of the finger trace, including: normalizing the phase and corresponding RSSI data of the preprocessed signal; calculating the variance flow of the normalized phase and RSSI corresponding to each tag in the tag array using a dynamically adjusted sliding window, and combining them with corresponding weights; extracting the maximum variance of each sliding window to form the maximum variance flow, and obtaining the first and last peak values ​​as the start and end points of the finger trace.

[0012] Optionally, the trace recognition model includes a classifier, a discriminator, and a generator. The generator generates relevant images based on random vectors to supplement the training dataset. The discriminator is used to distinguish between samples generated by the generator and real samples. Soft thresholding is added to the classifier to generate labels corresponding to the images.

[0013] To achieve the above objectives, a second aspect of the present invention provides a computer-readable storage medium storing an RFID finger tracking recognition program based on few-shot learning, which, when executed by a processor, implements the RFID finger tracking recognition method based on few-shot learning as described above.

[0014] According to embodiments of the present invention, a computer-readable storage medium stores an RFID finger tracking recognition program based on few-shot learning, so that when a processor executes the RFID finger tracking recognition program based on few-shot learning, it implements the RFID finger tracking recognition method based on few-shot learning as described above. Thus, high-precision, fine-grained finger tracking recognition can be achieved without the need for the human body to wear any device and with very few training samples.

[0015] To achieve the above objectives, a third aspect of the present invention provides a computer device including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the program, it implements the RFID finger tracking recognition method based on few-shot learning as described above.

[0016] According to an embodiment of the present invention, the computer device stores a few-shot learning-based RFID finger tracking recognition program in a memory, so that when the processor executes the few-shot learning-based RFID finger tracking recognition program, it implements the few-shot learning-based RFID finger tracking recognition method as described above. Thus, high-precision, fine-grained finger tracking recognition can be achieved without the need for the human body to wear any device and with very few training samples.

[0017] To achieve the above objectives, a fourth aspect of the present invention proposes an RFID finger tracking identification device based on few-shot learning, comprising: an acquisition module for acquiring raw finger reflection signals with temporal characteristics sent by a tag array; a feature extraction module for extracting features from the raw finger reflection signals to obtain a visualized finger tracking spatial feature map; and a tracking identification module for constructing a tracking identification model and inputting the finger tracking spatial feature map into the tracking identification model to complete the identification of the finger tracking spatial feature map.

[0018] According to an embodiment of the present invention, an RFID finger tracking recognition device based on few-sample learning acquires the original finger reflection signal with temporal characteristics sent by the tag array through an acquisition module; a feature extraction module extracts features from the original finger reflection signal to obtain a visualized finger tracking spatial feature map; a tracking recognition module constructs a tracking recognition model and inputs the finger tracking spatial feature map into the tracking recognition model to complete the recognition of the finger tracking spatial feature map; thus, high-precision, fine-grained finger tracking recognition can be achieved without the need for the human body to wear any device and with very few training samples.

[0019] In addition, the RFID finger tracking recognition device based on few-sample learning proposed in the above embodiments of the present invention may also have the following additional technical features:

[0020] Optionally, the feature extraction module is further configured to: correct the phase of the original finger reflection signal using one-dimensional phase unpacking; and smooth the original finger reflection signal using a filter to obtain a preprocessed signal; segment the preprocessed signal to obtain the start and end points of the finger trace; and obtain a likelihood estimation matrix composed of the likelihoods corresponding to all tags in the tag array, so as to track the finger movement trend based on the likelihood estimation matrix and the start and end points of the finger trace, and plot it as the visualized finger trace spatial feature map.

[0021] Optionally, the feature extraction module is further configured to: normalize the phase of the preprocessed signal and the corresponding RSSI data; calculate the variance flow of the normalized phase and RSSI corresponding to each tag in the tag array using a dynamically adjusted sliding window, and combine them with corresponding weights; extract the maximum variance of each sliding window to form the maximum variance flow, and obtain the first and last peak values ​​as the start and end points of the finger trace.

[0022] Optionally, the trace recognition model includes a classifier, a discriminator, and a generator. The generator generates relevant images based on random vectors to supplement the training dataset. The discriminator is used to distinguish between samples generated by the generator and real samples. Soft thresholding is added to the classifier to generate labels corresponding to the images. Attached Figure Description

[0023] Figure 1 This is a flowchart illustrating the RFID finger tracking identification method based on few-shot learning according to an embodiment of the present invention.

[0024] Figure 2 This is a schematic diagram of the structure of an RFID tag array according to an embodiment of the present invention;

[0025] Figure 3 This is a partially visualized spatial feature map of finger traces according to an embodiment of the present invention;

[0026] Figure 4 This is a schematic diagram of the overall framework of the RF-DSGAN according to an embodiment of the present invention;

[0027] Figure 5 This is a schematic diagram of the specific architecture of DS-GAN according to an embodiment of the present invention;

[0028] Figure 6 This is a block diagram of an RFID finger tracking recognition device based on few-sample learning according to an embodiment of the present invention. Detailed Implementation

[0029] Embodiments of the present invention are described in detail below, examples of which are illustrated in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and intended to explain the present invention, and should not be construed as limiting the present invention.

[0030] To better understand the above technical solutions, exemplary embodiments of the present invention will be described in more detail below with reference to the accompanying drawings. Although exemplary embodiments of the present invention are shown in the drawings, it should be understood that the present invention can be implemented in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided to enable a more thorough understanding of the present invention and to fully convey the scope of the invention to those skilled in the art.

[0031] To better understand the above technical solutions, the following will provide a detailed explanation of the technical solutions in conjunction with the accompanying drawings and specific implementation methods.

[0032] Figure 1 This is a flowchart illustrating the RFID finger tracking recognition method based on few-shot learning according to an embodiment of the present invention, as shown below. Figure 1 As shown, this RFID finger tracking recognition method based on few-shot learning includes the following steps:

[0033] S101, acquire the raw finger reflection signal with temporal characteristics sent by the tag array.

[0034] As an example, such as Figure 2 As shown, the tag array is a 6*6 matrix array composed of 36 RFID tags. The RF-DSGAN captures the time-series reflection signal changes caused by waving a finger in front of the tag array, thereby obtaining the original finger reflection signal with temporal characteristics.

[0035] In other words, the original finger reflection signal is a radio frequency signal, which includes spatial and temporal information.

[0036] S102, extract features from the original finger reflection signal to obtain a visualized spatial feature map of the finger trace.

[0037] As one embodiment, feature extraction is performed on the original finger reflection signal to obtain a visualized finger trace spatial feature map. This includes: correcting the phase of the original finger reflection signal using one-dimensional phase unpacking, and smoothing the original finger reflection signal using a filter to obtain a preprocessed signal; segmenting the preprocessed signal to obtain the start and end points of the finger trace; and obtaining a likelihood estimation matrix composed of the likelihoods corresponding to all tags in the tag array, so as to track the finger movement trend based on the likelihood estimation matrix and the start and end points of the finger trace, and plotting it into a visualized finger trace spatial feature map.

[0038] In other words, the raw radio frequency signals received by the RFID reader are first preprocessed, then trace segmented, and finally visualized to obtain a visualized spatial feature map of finger traces.

[0039] As an example, the preprocessed signal is segmented to obtain the start and end points of the finger trace, including: normalizing the phase of the preprocessed signal and the corresponding RSSI data; using a dynamically adjusted sliding window to calculate the variance flow of the normalized phase and RSSI corresponding to each tag in the tag array, and combining them with corresponding weights; extracting the maximum variance of each sliding window to form the maximum variance flow, and obtaining the first and last peaks as the start and end points of the finger trace.

[0040] As a specific example, since the raw radio frequency signal received by the RFID reader may have problems such as measurement defects, misreading, noise and phase jump, data preprocessing is required to improve the reliability of the signal. The phase value before unpacking fluctuates greatly and is irregular, so it cannot be used as experimental data. Therefore, a one-dimensional phase unpacking method is needed to correct the phase. At the same time, a filter is used to smooth the noise.

[0041] The phase unpacking formula is as follows:

[0042]

[0043]

[0044] in, Let i be the initial phase value at a certain moment.

[0045] As a specific implementation, since gestures are diverse and different gestures have different effects on the tag array, and gestures are unstable—even if the same person makes the same gesture at different speeds, the signal sequence will look significantly different—an improved maximum variance stream is used to determine the start and end points of the finger traces in order to more accurately extract the features of finger movements without adding extra noise or losing important information. Specifically: 1) Normalization: To ensure that the signal changes of all tags in the tag array are at the same magnitude, the preprocessed phase and RSSI data are normalized. 2) Weight Distribution: For each tag, the variance stream of the corresponding phase and RSSI is calculated using a dynamically adjusted sliding window, and an ablation experiment is designed to select the optimal value combining phase and RSSI. 3) Variance Maximization: The maximum variance of each sliding window is extracted to form the maximum variance stream, and the first and last peaks are taken as the start and end points of the gesture.

[0046] As a specific implementation, the Pearson correlation coefficient between the actual reflected energy distribution matrix and the theoretical reflected energy distribution matrix of the entire array is calculated to track the finger's movement trajectory at time t. A likelihood estimation matrix is ​​constructed by combining the likelihoods L(t,i) of the finger's position on all tags at time t. This likelihood estimation matrix reflects the finger's position on the tag array at time t. After recording the point with the highest probability at each time point, the trace is smoothed to track the finger's movement trend, and then plotted as shown in the figure. Figure 3 The image shown is in the form of a picture.

[0047] Furthermore, considering the label array as a 6x6 grid, the likelihood L(t,i) of a finger being above the i-th label at time t can be defined using the Pearson correlation coefficient:

[0048]

[0049] in:

[0050] P actual =|S reflect | 2 (4)

[0051]

[0052]

[0053] e ix =cosx + isinx (7)

[0054] Where N is the number of tags in the tag array, and C is a constant, which is set to 1 here. μ is the distance between the finger and the label, where P is the distance between the finger and the label. theoretical (i) and P actual The corresponding mean, P theoretical (i) refers to the theoretical reflection energy distribution matrix of the entire array if the finger lands on the i-th label, which is calculated by formula (5); P actual (t) refers to the actual reflected energy distribution matrix calculated by formula (6,7) using the phase and RSSI received by the reader at time t.

[0055] It should be noted that each tag i has a corresponding phase and RSSI at each time step. Concatenating the actual reflected energy calculated for each tag yields an actual reflected energy distribution matrix; for example... Figure 3 The algorithm was used to reconstruct letters written with a finger; to improve the resolution of the feature map, a 6*6 image was enlarged to 18*18 using two-dimensional linear interpolation.

[0056] S103, construct a trace recognition model and input the spatial feature map of the finger trace into the trace recognition model in order to complete the recognition of the spatial feature map of the finger trace.

[0057] As an example, the trace recognition model includes a classifier, a discriminator, and a generator. The generator generates relevant images based on random vectors to supplement the training dataset. The discriminator is used to distinguish between samples generated by the generator and real samples. Soft thresholding is added to the classifier to generate labels corresponding to the images.

[0058] In other words, such as Figure 5 As shown, a 100-dimensional random vector is input into the DS-GAN generator to generate relevant images and supplement the training dataset. Simultaneously, a discriminator is continuously trained to distinguish between generated and real samples. Finally, a classifier with an independent architecture predicts labels. Addressing the shortcomings of soft thresholding in image denoising, soft thresholding is incorporated into the classifier to enhance its performance on noisy data. Since the amount of real RFID feature map data is too small, resulting in insufficient generalization ability of conventional deep learning models on the test set, images generated by the generator are needed to supplement the training data. For this scenario, soft thresholding is embedded in the classifier to remove noise from the generated noisy virtual images. Then, a pseudo-labeling technique is used to assign the most probable label to the generated image based on the classifier's current capability. Only when the model predicts the sample's category with high confidence, or when the probability exceeds a certain threshold, are the generated image and label retained. Each mini-batch generates new images, which are immediately provided to the classifier.

[0059] In other words, such as Figure 4 As shown, RF-DSGAN comprises two key components: feature extraction and trace recognition. RF-DSGAN first takes the raw time-series signal received from the label array as input; a data preprocessing module creates a smooth, uniform time-slot signal from the raw data; then, a finger trajectory visualization module receives the gesture time-series data as input and generates a visualized trace image; the visualized image is passed to the trace recognition module, which uses DS-GAN (a GAN that performs fully supervised classification using a semi-supervised algorithm) to recognize the finger trace image, such as identifying the handwritten trace letter "a" as the corresponding category number.

[0060] In summary, RF-DSGAN utilizes few-shot learning techniques to improve the system's recognition performance with limited data. Furthermore, it employs soft thresholding to compensate for the challenge of effectively training GANs on limited datasets, enabling the system to learn key features from low-quality virtual images. Extensive experiments demonstrate that RF-DSGAN achieves an average accuracy of 94.69% on 160 training data points and 640 test data points across eight finger traces. When encountering different traces, users, finger speeds, and environments, it can reach up to 97.50% with fewer samples, showcasing excellent robustness and cross-environmental capabilities. The results of RF-DSGAN compared with other methods are shown in the table below.

[0061]

[0062] It can be seen that RF-DSGAN outperforms other baseline methods in various finger traces.

[0063] To implement the above embodiments, this invention also proposes a computer-readable storage medium storing an RFID finger trace recognition program based on few-shot learning. When executed by a processor, the RFID finger trace recognition program based on few-shot learning implements the RFID finger trace recognition method based on few-shot learning as described above.

[0064] According to embodiments of the present invention, a computer-readable storage medium stores an RFID finger tracking recognition program based on few-shot learning, so that when a processor executes the RFID finger tracking recognition program based on few-shot learning, it implements the RFID finger tracking recognition method based on few-shot learning as described above. Thus, high-precision, fine-grained finger tracking recognition can be achieved without the need for the human body to wear any device and with very few training samples.

[0065] To implement the above embodiments, this invention proposes a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the program, it implements the RFID finger tracking recognition method based on few-shot learning as described above.

[0066] According to an embodiment of the present invention, the computer device stores a few-shot learning-based RFID finger tracking recognition program in a memory, so that when the processor executes the few-shot learning-based RFID finger tracking recognition program, it implements the few-shot learning-based RFID finger tracking recognition method as described above. Thus, high-precision, fine-grained finger tracking recognition can be achieved without the need for the human body to wear any device and with very few training samples.

[0067] To achieve the above embodiments, this invention proposes an RFID finger tracking recognition device based on few-shot learning, such as... Figure 6 As shown, the RFID finger tracking recognition device based on few-shot learning includes: an acquisition module 10, a feature extraction module 20, and a tracking recognition module 30.

[0068] The acquisition module 10 is used to acquire the original finger reflection signal with temporal characteristics sent by the tag array; the feature extraction module 20 is used to extract features from the original finger reflection signal to obtain a visualized finger trace spatial feature map; the trace recognition module 30 is used to construct a trace recognition model and input the finger trace spatial feature map into the trace recognition model in order to complete the recognition of the finger trace spatial feature map.

[0069] Optionally, the feature extraction module 20 is further configured to: correct the phase of the original finger reflection signal by using one-dimensional phase unpacking; and smooth the original finger reflection signal using a filter to obtain a preprocessed signal; segment the preprocessed signal to obtain the start and end points of the finger trace; and obtain a likelihood estimation matrix composed of the likelihoods corresponding to all labels in the label array, so as to track the finger movement trend based on the likelihood estimation matrix and the start and end points of the finger trace, and draw it into a visualized finger trace spatial feature map.

[0070] Optionally, the feature extraction module 20 is further configured to: normalize the phase of the preprocessed signal and the corresponding RSSI data; calculate the variance flow of the normalized phase and RSSI corresponding to each tag in the tag array using a dynamically adjusted sliding window, and combine them with corresponding weights; extract the maximum variance of each sliding window to form the maximum variance flow, and obtain the first and last peak values ​​as the start and end points of the finger trace.

[0071] Optionally, the trace recognition model includes a classifier, a discriminator, and a generator. The generator generates relevant images based on random vectors to supplement the training dataset. The discriminator is used to distinguish between samples generated by the generator and real samples. Soft thresholding is added to the classifier to generate labels corresponding to the images.

[0072] It should be noted that the above regarding Figure 1 The description of the RFID finger tracking recognition method based on few-shot learning also applies to this RFID finger tracking recognition device based on few-shot learning, and will not be repeated here.

[0073] In summary, the RFID finger tracking recognition device based on few-shot learning according to embodiments of the present invention acquires the original finger reflection signal with temporal characteristics sent by the tag array through the acquisition module; the feature extraction module extracts features from the original finger reflection signal to obtain a visualized finger tracking spatial feature map; the tracking recognition module constructs a tracking recognition model and inputs the finger tracking spatial feature map into the tracking recognition model to complete the recognition of the finger tracking spatial feature map; thus, high-precision, fine-grained finger tracking recognition can be achieved without the need for the human body to wear any device and with very few training samples.

[0074] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0075] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0076] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0077] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0078] It should be noted that any reference signs placed between parentheses in the claims should not be construed as limiting the claims. The word "comprising" does not exclude the presence of components or steps not listed in the claims. The word "a" or "an" preceding a component does not exclude the presence of a plurality of such components. The invention can be implemented by means of hardware comprising several different components and by means of a suitably programmed computer. In a unit claim enumerating several means, several of these means may be embodied by the same item of hardware. The use of the words first, second, and third, etc., does not indicate any order. These words can be interpreted as names.

[0079] Although preferred embodiments of the invention have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments as well as all changes and modifications falling within the scope of the invention.

[0080] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims of this invention and their equivalents, this invention also intends to include these modifications and variations.

[0081] In the description of this invention, it should be understood that the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of indicated technical features. Therefore, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature. In the description of this invention, "a plurality of" means two or more, unless otherwise explicitly specified.

[0082] In this invention, unless otherwise explicitly specified and limited, the terms "installation," "connection," "linking," and "fixing," etc., should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral part; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; they can refer to the internal communication of two components or the interaction between two components. Those skilled in the art can understand the specific meaning of the above terms in this invention according to the specific circumstances.

[0083] In this invention, unless otherwise explicitly specified and limited, "above" or "below" the second feature can mean that the first feature is in direct contact with the second feature, or that the first feature is in indirect contact with the second feature through an intermediate medium. Furthermore, "above," "over," and "on top" of the second feature can mean that the first feature is directly above or diagonally above the second feature, or simply that the first feature is at a higher horizontal level than the second feature. "Below," "below," and "under" the second feature can mean that the first feature is directly below or diagonally below the second feature, or simply that the first feature is at a lower horizontal level than the second feature.

[0084] In the description of this specification, the references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of the present invention. In this specification, the illustrative expressions of the above terms should not be construed as necessarily referring to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples. Moreover, without contradiction, those skilled in the art can combine and integrate the different embodiments or examples described in this specification, as well as the features of different embodiments or examples.

[0085] Although embodiments of the present invention have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting the present invention. Those skilled in the art can make changes, modifications, substitutions and variations to the above embodiments within the scope of the present invention.

Claims

1. A method for RFID finger tracking identification based on few-shot learning, characterized in that, Includes the following steps: Acquire the raw finger reflection signal with temporal characteristics sent by the tag array; Feature extraction is performed on the original finger reflection signal to obtain a visualized spatial feature map of finger traces; A trace recognition model is constructed, and the spatial feature map of the finger trace is input into the trace recognition model in order to complete the recognition of the spatial feature map of the finger trace; The process includes feature extraction from the original finger reflection signal to obtain a visualized spatial feature map of the finger trace, including: The phase of the original finger reflection signal is corrected by one-dimensional phase unpacking, and the original finger reflection signal is smoothed by a filter to obtain a preprocessed signal; The preprocessed signal is segmented to obtain the start and end points of the finger trace; Obtain a likelihood estimation matrix composed of the likelihoods corresponding to all labels in the label array, so as to track the finger movement trend based on the likelihood estimation matrix and the start and end points of the finger trace, and draw it into the visualized finger trace spatial feature map. The preprocessed signal is segmented to obtain the start and end points of the finger trace, including: The phase of the preprocessed signal and the corresponding RSSI data are normalized. The normalized phase and RSSI variance currents for each tag in the tag array are calculated using a dynamically adjusted sliding window, and then combined with appropriate weights. The maximum variance of each sliding window is extracted to form a maximum variance stream, and the first and last peaks are taken as the start and end points of the finger trace. The phase unpacking formula is as follows: in, Let i be the initial phase value at a certain moment.

2. The RFID finger tracking recognition method based on few-shot learning as described in claim 1, characterized in that, The trace recognition model includes a classifier, a discriminator, and a generator. The generator generates relevant images based on random vectors to supplement the training dataset. The discriminator is used to distinguish between samples generated by the generator and real samples. Soft thresholding is added to the classifier to generate labels corresponding to the images.

3. A computer-readable storage medium, characterized in that, It stores an RFID finger tracking recognition program based on few-shot learning, which, when executed by a processor, implements the RFID finger tracking recognition method based on few-shot learning as described in any one of claims 1-2.

4. A computer device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the program, it implements the RFID finger tracking recognition method based on few-shot learning as described in any one of claims 1-2.

5. An RFID finger tracking recognition device based on few-shot learning, characterized in that, include The acquisition module is used to acquire the raw finger reflection signal with temporal characteristics sent by the tag array; The feature extraction module is used to extract features from the original finger reflection signal to obtain a visualized spatial feature map of finger traces; The trace recognition module is used to construct a trace recognition model and input the spatial feature map of the finger trace into the trace recognition model in order to complete the recognition of the spatial feature map of the finger trace; The feature extraction module is further used for, The phase of the original finger reflection signal is corrected by one-dimensional phase unpacking, and the original finger reflection signal is smoothed by a filter to obtain a preprocessed signal; The preprocessed signal is segmented to obtain the start and end points of the finger trace; Obtain a likelihood estimation matrix composed of the likelihoods corresponding to all labels in the label array, so as to track the finger movement trend based on the likelihood estimation matrix and the start and end points of the finger trace, and draw it into the visualized finger trace spatial feature map. The feature extraction module is further used for, The phase of the preprocessed signal and the corresponding RSSI data are normalized. The normalized phase and RSSI variance currents for each tag in the tag array are calculated using a dynamically adjusted sliding window, and then combined with appropriate weights. The maximum variance of each sliding window is extracted to form a maximum variance stream, and the first and last peaks are taken as the start and end points of the finger trace. The phase unpacking formula is as follows: in, Let i be the initial phase value at a certain moment.

6. The RFID finger tracking recognition device based on few-shot learning as described in claim 5, characterized in that, The trace recognition model includes a classifier, a discriminator, and a generator. The generator generates relevant images based on random vectors to supplement the training dataset. The discriminator is used to distinguish between samples generated by the generator and real samples. Soft thresholding is added to the classifier to generate labels corresponding to the images.

Images