A face recognition method, apparatus, device, storage medium and program product

Abstract

The application discloses a face recognition method, device and equipment, a storage medium and a program product, and relates to the technical field of face recognition. The method comprises the following steps: performing multi-scale feature extraction on an input face image through a plurality of groups of pre-constructed Gabor filters to obtain a target feature vector; evaluating each participating node based on a plurality of evaluation dimensions to obtain multi-dimensional quantization parameters; aggregating gradient training results of each participating node according to the multi-dimensional quantization parameters to obtain global gradient parameters; updating a local face recognition model based on the global gradient parameters; and inputting the target feature vector into the updated local face recognition model to perform face recognition. Since the features of the input face image are extracted in multiple scales, the representation ability of the image features is enhanced, the data contribution weights provided by the distributed nodes are effectively balanced through node evaluation in multiple dimensions, the convergence speed and recognition accuracy of the face recognition model are improved, and the accuracy of face recognition is greatly improved.

Description

Technical Field

[0001] This invention relates to the field of facial recognition, and more particularly to a facial recognition method, apparatus, device, storage medium, and program product. Background Technology

[0002] With the rapid development of social informatization, facial recognition systems are increasingly widely used in security, finance, and social networking. Especially in distributed computing environments, achieving efficient, secure, and fair model training has become a current research hotspot. For multi-party collaborative scenarios involving multiple institutions and regions, traditional facial recognition model training methods face severe challenges in areas such as data privacy protection, model performance optimization, and incentive mechanism design.

[0003] Currently, training methods for distributed face recognition models mainly employ centralized aggregation or federated averaging algorithms. These training modes are relatively simple, with fixed model weight allocation, making dynamic adjustments impossible based on the data quality of different participants. If some participants provide low-quality data or have an unbalanced sample distribution, it directly impacts the convergence speed and recognition accuracy of the global model. Conversely, applying the same aggregation weights to all participants fails to fully incentivize participation from those holding high-quality data. Furthermore, existing methods typically employ a single convolutional neural network structure at the feature extraction level, which has limited ability to represent facial features under different lighting conditions and pose variations, thus restricting the model's generalization performance. Summary of the Invention

[0004] The main objective of this invention is to provide a face recognition method, apparatus, device, storage medium, and program product, which aims to solve the technical problems of low face recognition accuracy caused by the inability of existing technologies to effectively balance the contribution weights of different participants in distributed face recognition processes and the single feature extraction method.

[0005] To achieve the above objectives, the present invention provides a face recognition method, which is applied to participating nodes in a federated learning blockchain, wherein the federated learning blockchain includes multiple participating nodes, and the method includes the following steps:

[0006] The input face image is subjected to multi-scale feature extraction by multiple pre-constructed Gabor filters to obtain the target feature vector. The Gabor filters are composed of multiple complex filter kernels generated based on different directions and spatial frequencies.

[0007] Each participating node is evaluated based on multiple evaluation dimensions to obtain multidimensional quantitative parameters. The evaluation dimensions include marginal contribution dimension, risk dimension and reputation dimension. The multidimensional quantitative parameters include marginal contribution degree, risk weight parameter and target reputation value.

[0008] Obtain the gradient training results uploaded by each participating node to the federated learning blockchain. The gradient training results are obtained by each participating node performing gradient training on its local face recognition model.

[0009] The gradient training results of each participating node are aggregated based on the multidimensional quantization parameters to obtain the global gradient parameters;

[0010] The local face recognition model is updated based on the global gradient parameters, and the target feature vector is input into the updated local face recognition model for face recognition.

[0011] Optionally, the step of extracting multi-scale features from the input face image using multiple pre-constructed Gabor filters to obtain the target feature vector includes:

[0012] The input face image is processed to obtain an initial grayscale image;

[0013] The initial grayscale image is normalized to obtain candidate grayscale images;

[0014] Multiple complex filter kernels are generated based on multiple sets of different directions and spatial frequencies, and multiple sets of Gabor filters are generated based on the complex filter kernels.

[0015] The candidate grayscale image is decomposed at multiple scales using the multiple sets of Gabor filters. Based on the filter output results, multiple sets of texture feature maps with different directions and spatial frequencies are extracted. The wavelet function of the Gabor filter is defined by the following formula:

[0016]

[0017]

[0018] Where z represents the spatial coordinates (x, y) of a pixel in the input face image, and k u,v This indicates the center frequency of the Gabor filter. σ represents the directivity of a two-dimensional Gabor filter. 2 The square of the standard deviation of the Gaussian envelope is used to control the spatial distribution width of the Gabor filter. The Gaussian envelope term is used to constrain the spatial distribution of the plane wave, exp(ik) u,v z) represents the periodicity and directionality of the wavelet. Used for energy spectrum attenuation compensation;

[0019] Feature analysis is performed on the multiple sets of texture feature maps with different directions and spatial frequencies to generate target feature vectors.

[0020] Optionally, the step of performing feature analysis on the multiple sets of texture feature maps with different directions and spatial frequencies to generate a target feature vector includes:

[0021] Traverse each pixel in the texture feature map and calculate the LBP value of each pixel;

[0022] Based on the LBP values ​​of each pixel in the texture feature map, generate the LBP feature map corresponding to each Gabor filter;

[0023] The LBP feature map is divided into multiple sub-blocks, and a distribution histogram of LBP values ​​is plotted for each sub-block.

[0024] The distribution histograms of each sub-block are spliced ​​together to generate the target feature vector.

[0025] Optionally, the evaluation of each participating node based on multiple evaluation dimensions to obtain multi-dimensional quantitative parameters includes:

[0026] The marginal contribution of each participating node is evaluated to obtain the marginal contribution degree of each participating node, which is calculated based on the following formula:

[0027]

[0028] in, Let s(i) represent the marginal contribution of participating node i, s(i) represent the sub-alliance consisting of all participating nodes including participating node i, v(s) represent the revenue value of the sub-alliance, v(s / {i}) represent the revenue value of the sub-alliance after removing participating node i, w(|s|) represent the contribution weight of the sub-alliance, |s| represent the size of the sub-alliance, and n represent the total number of nodes in the sub-alliance.

[0029] Risk weights are assessed for each data supply node based on the integrity of each participating node, and the risk weights for each participating node are calculated based on the following formula:

[0030]

[0031] Among them, Λ i This represents the risk weight of participating node i. λ represents the integrity of participating node i within the current block time period h. i This represents the integrity vector of participating node i;

[0032] The reputation of each participating node is quantitatively evaluated based on its data quality information and task completion rate to obtain an initial reputation value for each participating node.

[0033] The initial reputation value of each participating node is normalized based on the maximum reputation value of the participating node to obtain the target reputation value of each participating node.

[0034] Optionally, the risk weight assessment of each data supply node based on the integrity of each participating node to obtain the risk weight of each participating node includes:

[0035] Obtain the historical integrity score of each participating node within the historical block time period;

[0036] The noise level of the data provided by each participating node is assessed to obtain the data quality information of each participating node.

[0037] Based on the data quality information, an exponential decay coefficient is generated for each participating node;

[0038] The historical integrity score is updated based on the exponential decay coefficient to obtain the updated integrity score. The integrity score update is performed using the following formula:

[0039]

[0040] Where h represents the current block time period, and h-1 represents the previous block time period. This represents the integrity of participating node i in the previous block time period, and α represents the exponential decay coefficient. This represents the integrity of participating node i within the current block time period, and the reward. i This represents the reward value obtained by participating node i during local model training in the current block time period, where β represents a positive number. This represents the cumulative historical integrity score of participating node i;

[0041] Based on the updated integrity rating, risk weights are assessed for each data supply node to obtain the risk weights for each participating node.

[0042] Optionally, the step of aggregating the gradient training results of each participating node according to the multidimensional quantization parameters to obtain global gradient parameters includes:

[0043] Monitor abnormal behavior of each participating node;

[0044] Mark participating nodes exhibiting abnormal behavior as abnormal nodes;

[0045] The abnormal nodes are weighted according to the multidimensional quantization parameters to obtain a target weight vector, which is calculated based on the following formula:

[0046]

[0047] Where w' represents the target weight vector;

[0048] The gradient training results of each participating node are aggregated based on the target weight vector and the multidimensional quantization parameters to obtain global gradient parameters.

[0049] Furthermore, to achieve the above objectives, the present invention also proposes a face recognition device, the face recognition device comprising:

[0050] The multi-scale feature extraction module is used to extract multi-scale features from the input face image through multiple pre-constructed Gabor filters to obtain the target feature vector. The Gabor filters are composed of multiple complex filter kernels generated based on different directions and spatial frequencies.

[0051] The multidimensional evaluation module is used to evaluate each participating node based on multiple evaluation dimensions to obtain multidimensional quantitative parameters. The evaluation dimensions include marginal contribution dimension, risk dimension and reputation dimension. The multidimensional quantitative parameters include marginal contribution degree, risk weight parameter and target reputation value.

[0052] The distributed gradient training module is used to obtain the gradient training results uploaded by each participating node to the federated learning blockchain. The gradient training results are obtained by each participating node performing gradient training on its local face recognition model.

[0053] The gradient parameter aggregation module is used to aggregate the gradient training results of each participating node according to the multidimensional quantization parameters to obtain global gradient parameters.

[0054] The face recognition module is used to update the local face recognition model based on the global gradient parameters, and input the target feature vector into the updated local face recognition model for face recognition.

[0055] In addition, to achieve the above objectives, this application also proposes a face recognition device, the device comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor, the computer program being configured to implement the steps of the face recognition method as described above.

[0056] In addition, to achieve the above objectives, this application also proposes a computer-readable storage medium storing a computer program that, when executed by a processor, implements the steps of the face recognition method described above.

[0057] In addition, to achieve the above objectives, this application also provides a computer program product, which includes a computer program that, when executed by a processor, implements the steps of the face recognition method described above.

[0058] This invention extracts multi-scale features from an input face image using multiple pre-constructed Gabor filters to obtain a target feature vector. The Gabor filters consist of multiple complex filter kernels generated based on different directions and spatial frequencies. Each participating node is evaluated based on multiple evaluation dimensions to obtain multi-dimensional quantization parameters, including marginal contribution, risk, and reputation dimensions. The multi-dimensional quantization parameters include marginal contribution degree, risk weight parameter, and target reputation value. The gradient training results uploaded by each participating node to the federated learning blockchain are obtained; these gradient training results represent the localization performance of each participating node. The face recognition model is trained using gradients; the gradient training results of each participating node are aggregated according to the multi-dimensional quantization parameters to obtain global gradient parameters; the local face recognition model is updated based on the global gradient parameters, and the target feature vector is input into the updated local face recognition model for face recognition; since this invention extracts features from the input face image at multiple scales, it enhances the representation ability of image features. By evaluating nodes in multiple dimensions, it effectively balances the data contribution weights provided by distributed nodes, improves the convergence speed and recognition accuracy of the face recognition model, and significantly improves the accuracy of face recognition. Attached Figure Description

[0059] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, for those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0060] Figure 1 This is a schematic diagram of the structure of a face recognition device in the hardware operating environment involved in the embodiments of the present invention;

[0061] Figure 2 This is a flowchart illustrating an embodiment of the face recognition method of the present invention;

[0062] Figure 3 This is a schematic diagram of the multi-scale feature extraction process in one embodiment of the face recognition method of the present invention;

[0063] Figure 4 This is a schematic diagram of the framework of a face recognition system in one embodiment of the present invention;

[0064] Figure 5 This is a structural block diagram of an embodiment of the face recognition device of the present invention.

[0065] The realization of the objective, functional features and advantages of the present invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0066] It should be understood that the specific embodiments described herein are for illustrative purposes only and are not intended to limit the scope of the invention.

[0067] Reference Figure 1 , Figure 1 This is a schematic diagram of the structure of a face recognition device in the hardware operating environment of an embodiment of the present invention.

[0068] like Figure 1 As shown, the face recognition device may include: a processor 1001, such as a central processing unit (CPU), a communication bus 1002, a user interface 1003, a network interface 1004, and a memory 1005. The communication bus 1002 is used to enable communication between these components. The user interface 1003 may include a display screen and an input unit such as a keyboard; optionally, the user interface 1003 may also include a standard wired interface or a wireless interface. The network interface 1004 may optionally include a standard wired interface or a wireless interface (such as a Wireless-Fidelity (Wi-Fi) interface). The memory 1005 may be high-speed random access memory (RAM) or stable non-volatile memory (NVM), such as a disk drive. The memory 1005 may also optionally be a storage device independent of the aforementioned processor 1001.

[0069] Those skilled in the art will understand that Figure 1 The structure shown does not constitute a limitation on the face recognition device and may include more or fewer components than shown, or combine certain components, or have different component arrangements.

[0070] like Figure 1 As shown, the memory 1005, which is a computer-readable storage medium, may include an operating system, a network communication module, a user interface module, and a face recognition program.

[0071] exist Figure 1 In the face recognition device shown, the network interface 1004 is mainly used for data communication with the network server; the user interface 1003 is mainly used for data interaction with the user; the processor 1001 and the memory 1005 in the face recognition device of the present invention can be set in the face recognition device. The face recognition device calls the face recognition program stored in the memory 1005 through the processor 1001 and executes the face recognition method provided in the embodiment of the present invention.

[0072] This invention provides a face recognition method, referring to... Figure 2 , Figure 2 This is a flowchart illustrating an embodiment of the face recognition method of the present invention.

[0073] This embodiment is applied to participating nodes in a federated learning blockchain, which includes multiple participating nodes. In this embodiment, the face recognition method includes the following steps:

[0074] Step S10: Extract multi-scale features from the input face image using multiple pre-built Gabor filters to obtain the target feature vector.

[0075] It should be noted that there are three technical bottlenecks in the training of traditional distributed face recognition models: First, there is a contradiction between efficiency and robustness in the feature extraction stage. Although multimodal feature fusion improves the discrimination, it has high computational complexity and weak adaptability to dynamic scenarios, resulting in prominent redundancy problems. Second, the evaluation of node contribution relies on static assumptions, making it difficult to objectively quantify the marginal contribution of nodes during dynamic training, which leads to evaluation bias. Third, the model aggregation strategy is sensitive to data heterogeneity, and client features and parameters lack adaptive alignment capabilities.

[0076] Currently, feature extraction is a core component of distributed training for face recognition models, and its efficiency and robustness directly affect model performance. Existing research mainly focuses on feature operator optimization and multimodal feature fusion. Nevertheless, feature extraction research still faces challenges such as high computational complexity, weak adaptability to dynamic scenes, and feature redundancy. In particular, the lack of effective dimensionality compression mechanisms in the fusion of heterogeneous features and deep networks limits its large-scale application in distributed training.

[0077] Distributed training node contribution evaluation and model aggregation have gained widespread attention for helping to screen high-quality training datasets and improve the efficiency of distributed training of face recognition models. However, existing aggregation methods lack adaptive alignment capabilities for heterogeneous client data, which limits their reliability in complex scenarios.

[0078] To address the aforementioned technical shortcomings, one embodiment of this invention proposes a distributed training method for face recognition models based on data contribution aggregation. At the feature extraction level, a composite feature fusion mechanism combining Gabor transform and Local Binary Pattern (LBP) is employed. The Gabor filter bank extracts multi-directional, multi-scale frequency domain features, while the LBP operator captures local texture details. Through feature space reconstruction and 3D feature vector dimensionality reduction, a low-dimensional feature representation with strong discriminative power is formed. This technical solution effectively enhances the model's robustness to changes in illumination and pose, laying a high-quality feature foundation for distributed collaborative training.

[0079] Regarding the model collaborative training mechanism, the Shapley Value theory is introduced to construct a data contribution evaluation model. By calculating the marginal contribution of each participant to improving model accuracy in real time, the global model aggregation weights are dynamically adjusted to establish an incentive mechanism that matches "contribution" with "reward." This technique eliminates the negative impact of low-quality data participants on the global model and stimulates the participation of high-quality data holders.

[0080] It should be understood that the executing entity of this embodiment can be a computing service device with data processing, network communication, and program execution functions, such as a personal computer, a server, etc., or a terminal electronic device capable of realizing the above functions. The following uses a face recognition device as an example to describe this embodiment and the following embodiments.

[0081] It should be noted that the Gabor filter consists of multiple complex filter kernels generated based on different directions and spatial frequencies.

[0082] Furthermore, to effectively enhance the robustness of the face model to changes in lighting and pose, lay a high-quality feature foundation for distributed collaborative training, and thus improve the accuracy of face recognition, reference was made to... Figure 3 , Figure 3 This is a schematic diagram of a multi-scale feature extraction process in one embodiment. Step S10 above may include:

[0083] Step S101: Perform grayscale processing on the input face image to obtain an initial grayscale image.

[0084] In some embodiments, the face recognition device performs image grayscale processing on the input face image, assigns weights of 0.299, 0.587, and 0.114 to the red, green, and blue channels of the image respectively (in accordance with human eye sensitivity), and uses a weighted average method to merge the three RGB channels into a single channel grayscale.

[0085] Step S102: Normalize the initial grayscale image to obtain candidate grayscale images.

[0086] In some embodiments, the face recognition device divides each pixel value of the grayscale image by 255.0, while converting the data type to float32 format, and normalizing the pixel values ​​from the integer range of [0,255] to the floating-point range of [0,1].

[0087] Step S103: Generate corresponding complex filter kernels based on multiple sets of different directions and spatial frequencies, and generate multiple sets of Gabor filters based on the complex filter kernels.

[0088] In some embodiments, when a face recognition device can use multiple Gabor filters for filtering, it first presets different directions (such as 0°, 45°, 90°, 135°, etc.) and spatial frequencies (parameters related to wavelength) to generate corresponding complex filter kernels.

[0089] Step S104: Perform multi-scale decomposition on the candidate grayscale image using the multiple sets of Gabor filters, and extract multiple sets of texture feature maps with different directions and spatial frequencies based on the filter output results.

[0090] It should be noted that the wavelet function of the Gabor filter is based on the following formula:

[0091]

[0092] Where z represents the spatial coordinates (x, y) of a pixel in the input face image, and k u,v This indicates the center frequency of the Gabor filter. σ represents the directivity of a two-dimensional Gabor filter. 2 The square of the standard deviation of the Gaussian envelope is used to control the spatial distribution width of the Gabor filter. The Gaussian envelope term is used to constrain the spatial distribution of the plane wave, exp(ik) u,v z) represents the periodicity and directionality of the wavelet. Used for energy spectrum attenuation compensation.

[0093] Step S105: Perform feature analysis on the multiple sets of texture feature maps with different directions and spatial frequencies to generate target feature vectors.

[0094] In some embodiments, the face recognition device performs a convolution operation between each Gabor filter and the grayscale image to extract the texture response at a specific direction and frequency, and takes the amplitude of the filtering result as a feature.

[0095] In some embodiments, the face recognition device employs Gabor transform to decompose the image at multiple scales, effectively extracting local information of the signal at different scales and directions. By selecting different frequencies and directions, multiple sets of two-dimensional Gabor wavelets can be generated, thereby enabling image analysis from multiple angles during feature extraction.

[0096] Furthermore, in order to accurately extract local and global features of the image, thereby enhancing the image's feature representation capability, step S105 above may include:

[0097] Step S1051: Traverse each pixel in the texture feature map and calculate the LBP value of each pixel;

[0098] Step S1052: Generate LBP feature maps corresponding to each Gabor filter based on the LBP values ​​of each pixel in the texture feature map;

[0099] Step S1053: Divide the LBP feature map into multiple sub-blocks and generate a distribution histogram of LBP values ​​for each sub-block;

[0100] Step S1054: Concatenate the distribution histograms of each sub-block to generate the target feature vector.

[0101] In some embodiments, the face recognition device performs Local Binary Pattern (LBP) calculation on each filtered image, comparing the difference between the value of each pixel and its neighboring pixels to generate a binary code, which is then converted into a decimal value to form an LBP feature map.

[0102] It should be noted that the Local Binary Pattern (LBP) is calculated using the following formula:

[0103]

[0104] Among them, LBP P,R This represents the LBP value, where P represents the number of neighboring points around the center pixel, R represents the distance between the neighboring points and the center pixel, and g... p Let g represent the gray value of the p-th neighboring point. c s(c) represents the grayscale value of the center pixel, and is used to compare the grayscale values ​​of neighboring points with those of the center pixel. p This is used to weight the comparison results (0 or 1) of each neighboring point according to the bit weight, and finally form a binary number (which is the LBP value after being converted to decimal).

[0105] In some embodiments, the face recognition device divides the LBP feature maps of all Gabor filter channels into blocks to calculate histograms. Each Gabor-filtered image is divided into several small blocks, and within each small block, the LBP histogram is calculated (i.e., assuming the LBP mode has b bins, the histogram of each block is a b-dimensional vector. These are concatenated into a high-dimensional feature vector, which comprehensively describes the local texture and global structure information of the image).

[0106] In some embodiments, the face recognition device may employ a lightweight convolutional network (CNN) structure for algorithm optimization. In the CNN network, the input layer is set to adapt to the standardized image size (64×64 pixels), and two convolutional layers (kernel size 3×3, stride 1) and max pooling (kernel size 2×2) are defined in the feature extraction part. The output dimension of the fully connected layer is made consistent with the number of categories in the classification task, resulting in the final target feature vector (i.e., the comprehensive feature vector). (Refer to...) Figure 3 , Figure 3This is a schematic diagram of the feature extraction process in one embodiment. The face recognition device performs grayscale and normalization processing on the input face image, filters the processed image through multiple Gabor filters of different scales to obtain texture feature maps of multiple scales, calculates and statistically analyzes the texture distribution through LBP feature calculation, and determines whether each Gabor filter has completed the feature map processing. If not, feature processing continues; if so, the feature histograms output by each Gabor filter of different scales are concatenated to output a comprehensive feature vector (i.e., the target feature vector).

[0107] It should be noted that, Figure 3 This diagram illustrates the feature extraction framework designed in this scheme. The entire processing flow begins with the original image input, and grayscale conversion and data normalization are performed sequentially in the preprocessing stage. To further enhance texture representation capabilities, the system employs a multi-channel Gabor filter bank for feature enhancement. By configuring multiple sets of filters with different frequency and directional parameters, local texture details in the image are effectively extracted. Based on this, an improved LBP operator is introduced for secondary processing of the feature map. This algorithm establishes the grayscale contrast relationship between the center pixel and its circular neighboring pixels, ultimately forming a binary feature code with strong discriminative power. Histograms are calculated for the LBP features output by each Gabor filter, and the distribution of different texture patterns is statistically analyzed to form a texture description under a single filter. The final output comprehensive feature vector integrates multi-scale and multi-directional texture information, making it suitable for tasks such as image classification and recognition. This method combines the frequency domain analysis capabilities of Gabor with the local texture description advantages of LBP, enhancing the feature representation capabilities.

[0108] Step S20: Evaluate each participating node based on multiple evaluation dimensions to obtain multidimensional quantitative parameters.

[0109] It should be noted that the evaluation dimensions include marginal contribution dimension, risk dimension and reputation dimension, and the multidimensional quantitative parameters include marginal contribution degree, risk weight parameter and target reputation value.

[0110] It should be noted that, in terms of contribution evaluation mechanism, in order to screen high-quality data from data stored in different nodes and reduce the impact of low-quality data on the performance of the final trained model, a quality index and data freshness are designed for each round of model update. Based on Shapley value theory, a contribution evaluation mechanism is designed to quantify the marginal contribution of nodes and introduce risk weight correction to achieve dynamic matching between data quality and aggregation weight.

[0111] Furthermore, in order to accurately assess the contribution, risk, and reputation of each participating node, in one embodiment, step S20 above may include:

[0112] Step S201: Evaluate the marginal contribution of each participating node to obtain the marginal contribution degree of each participating node.

[0113] It should be noted that this embodiment can utilize Shapley value sampling to calculate the marginal contribution value of each data provider in different federated learning sub-coalitions, quantifying the marginal contribution of each data provider. The marginal contribution is calculated based on the following formula:

[0114]

[0115] in, Let s(i) represent the marginal contribution of participating node i, s(i) represent the sub-alliance consisting of all participating nodes including node i, v(s) represent the profit value of the sub-alliance, v(s / {i}) represent the profit value of the sub-alliance after removing participating node i, w(|s|) represent the contribution weight of the sub-alliance, |s| represent the size of the sub-alliance, and n represent the total number of nodes in the sub-alliance. In federated learning, a collaborative group composed of a subset of nodes (members) is called a sub-alliance. Members i in sub-alliance s(i) participate in the local training of federated learning by sharing data or model parameters, jointly optimizing the global model. s(i) is the set of all sub-allies containing member i, where each member i provides data or computation.

[0116] Step S202: Based on the integrity of each participating node, assess the risk weight of each data supply node to obtain the risk weight of each participating node.

[0117] It should be noted that this embodiment may introduce an additional set of data risk weight vectors Λ=(Λ1,Λ2,…,Λ). n ), where: Λ i This represents the data privacy risk weight of the i-th distributed training participant. Taking into account the past integrity of the participating node, the Shapley value calculation formula has been modified according to the characteristics of the distributed training data source itself, making the weight correlated with the integrity of the participating node.

[0118] set up Let λ represent the integrity score of miner node i within block time period h. Since all source data is known before the start of the distributed training task, the integrity vector λ is consistent across a single distributed training task, and its weights sum to 1. The risk weight is calculated based on the following formula:

[0119]

[0120] Among them, Λ i This represents the risk weight of participating node i. λ represents the integrity score of participating node i within the current block time period h, and is dynamically updated over time. iThis represents the integrity vector of participating node i.

[0121] Step S203: Quantitatively evaluate the reputation of each participating node based on its data quality information and task completion rate to obtain the initial reputation value of each participating node.

[0122] It should be noted that the data quality information can be the data quality level provided by each participating node in the current period and historical periods. For example, data quality information can be obtained by evaluating the data noise level of the participating nodes. The task completion rate mentioned above can be the task completion rate of the participating nodes in distributed training in historical periods.

[0123] Step S204: Normalize the initial reputation value of each participating node based on the maximum reputation value of the participating node to obtain the target reputation value of each participating node.

[0124] In some embodiments, the face recognition device may set a random hash value Convert(h) i This maps the participant's unique identifier hash value to a random number in the range (0, 1). The mapping priority is determined by the parameter value. Where r i This represents the node's reputation value, where max(r) represents the highest reputation value among all nodes. The final priority is determined by the reputation value and the random hash value. (Convert(h)) i (To be decided jointly)

[0125] Furthermore, in order to accurately assess the integrity of each participating node and effectively reduce the risk of facial recognition data leakage, in some embodiments, step S202 above may include:

[0126] Step S2021: Obtain the historical integrity score of each participating node within the historical block time period;

[0127] Step S2022: Evaluate the noise level of the data provided by each participating node and obtain the data quality information of each participating node;

[0128] Step S2023: Generate the exponential decay coefficient for each participating node based on the data quality information;

[0129] Step S2024: Update the historical integrity score according to the exponential decay coefficient to obtain the updated integrity score;

[0130] Step S2025: Based on the updated integrity level, perform risk weight assessment on each data supply node to obtain the risk weight of each participating node.

[0131] It should be noted that facial recognition devices can use dynamic weight integration, which dynamically adjusts the weight based on the data quality (noise level) of the participating nodes. Higher quality data receives higher weight. At the same time, the influence of historical contributions is adjusted through an exponential decay coefficient, with earlier contributions decaying more significantly.

[0132] In some embodiments, the integrity score of participating nodes can be recorded in the smart contract of the main chain and bound to the account of each data contributor. This integrity score is continuously updated with each training round and evaluation score, and the update is based on the following formula:

[0133]

[0134] Where h represents the current block time period, and h-1 represents the previous block time period. This represents the integrity of participating node i in the previous block time period, and α represents the exponential decay coefficient. The larger α is, the higher the weight of the reward in the current period, and the faster the historical contribution decays. This represents the integrity of participating node i within the current block time period, and the reward. i This represents the reward value obtained by participating node i during local model training in the current block time period, where β represents a positive number. This represents the cumulative historical integrity score of participating node i.

[0135] Step S30: Obtain the gradient training results uploaded by each participating node to the federated learning blockchain.

[0136] It should be noted that the gradient training results are obtained by each participating node performing gradient training on the local face recognition model.

[0137] In some embodiments, during the model aggregation process, each participating node follows a formula. Train the local model gradient by finding the minimum parameters. Minimize loss function Build a decentralized training architecture. Indicates from dataset x ik y ik ∈D i The main parameters obtained from training. Each node propagates its gradient results throughout the network.

[0138] In some embodiments, each participating node may introduce a verifiable random function (VRF) to encrypt the data and randomly select an aggregation node to ensure the fairness of node selection.

[0139] In some embodiments, key generation can be based on formula (pk) i ,sk i )←VRF.KeyGen(1 λGenerate a public-private key pair for each participating node, (pk i ,sk i ) represent the public and private key pairs, respectively, and λ is a security parameter; the cryptographic sorting algorithm calculates the priority of each node as a leader or committee member based on its private key and a global seed random number, and outputs a sorting credential. The calculation formula is expressed as: (π i ,h i )←VRF.Eval(sk i ,seed|r i ). Where h i The priority hash value used for sorting, π i It is a verifiable credential.

[0140] In some embodiments, the face recognition device can verify whether the node priority is valid, and the verification formula is VRF.Verify(pk i ,seed|r i ,π i ,h i The formula outputs 1 to {0,1}, indicating that the verification passed and 0 failed.

[0141] Step S40: Aggregate the gradient training results of each participating node according to the multidimensional quantization parameters to obtain global gradient parameters.

[0142] In practical implementation, the face recognition device can aggregate multi-dimensional quantization parameters to obtain the aggregate weights of each participating node, and then aggregate the gradient training results of each participating node based on the aggregate weights to obtain global gradient parameters.

[0143] In some embodiments, after each training round, the smart contract can assign the aggregate weight w' and risk weight Λ of the current round to the facial recognition device. i Marginal contribution and dynamic integrity weight Key parameters are permanently written to the blockchain. This data not only provides historical evidence for subsequent rounds of weight calculation but also ensures the transparency and auditability of the entire aggregation process through the blockchain's immutability. A reputation score is generated based on the data quality and task completion rate of each data contributor (node), serving as cumulative proof of long-term contributions. This score is then normalized by dividing by the maximum reputation score (max(r)) to eliminate absolute numerical differences and establish a relative evaluation system. This normalization result can adjust its proportion according to changes in the highest reputation score, allowing for rapid priority adjustments even when new nodes are added.

[0144] Furthermore, in order to accurately assign data weights to each participating node, step S40 above may include:

[0145] Step S401: The face recognition device monitors abnormal behavior of each participating node;

[0146] Step S402: Mark participating nodes exhibiting abnormal behavior as abnormal nodes;

[0147] Step S403: Adjust the weights of the abnormal nodes according to the multidimensional quantization parameters to obtain the target weight vector;

[0148] Step S404: Aggregate the gradient training results of each participating node based on the target weight vector and the multidimensional quantization parameters to obtain global gradient parameters.

[0149] It should be noted that in the distributed federated learning architecture, the collaboration paradigm of participating nodes adopts a consensus-like mechanism design, and its operating logic is analogous to the node verification system of a blockchain network. The reputation evaluation system adopts a three-state determination mechanism: when a node generates positive feedback, a reputation weight gain is executed; in a steady state, the current evaluation coefficient is maintained; if a negative event is detected, reputation decay is triggered.

[0150] In practical implementation, to defend against malicious node attacks, the facial recognition device can preset a dynamic penalty protocol: when abnormal behavior is verified, account permissions are first suspended and incentives are rolled back, followed by zeroing out the reputation weight; if the violation is serious, a ban is initiated to participate in federated learning as a penalty. Based on the Shapley value calculation formula, a comprehensive weight evaluation is performed to obtain the aggregate weight (i.e., the target weight vector), which is calculated based on the following formula:

[0151]

[0152] Here, w' represents the target weight vector, which adjusts the contribution distribution ratio of each node (participant) in federated learning, combining the dynamic weights of integrity and Shapley value. The numerator represents the basic contribution weight of node i in the sub-coalition s, while also considering its integrity weight Λ. i Marginal contribution is used in the denominator to ensure The adjusted weights w' sum to 1, achieving a fair distribution.

[0153] Step S50: Update the local face recognition model based on the global gradient parameters, and input the target feature vector into the updated local face recognition model for face recognition.

[0154] In some embodiments, the face recognition device employs a lightweight convolutional network (CNN) structure for algorithm optimization. The input layer of the CNN is adapted to the standardized image size (64×64 pixels). Two convolutional layers (kernel size 3×3, stride 1) and a max-pooling layer (kernel size 2×2) are defined in the feature extraction part to ensure that the output dimension of the fully connected layer matches the number of categories in the classification task, thus obtaining the final features.

[0155] In some embodiments, the face recognition device may set a random hash value to map the participant's unique identifier hash value to a random number in the range (0,1). The mapping priority is determined by the parameter value, where represents the node reputation value, and represents the highest reputation value among all nodes. The final priority is determined jointly by the reputation value and the random hash value.

[0156] In some embodiments, when a node i is selected as a verification node, signed gradient data is generated using worker nodes to verify the gradient information of other nodes. First, the system checks the marginal contribution of the participating nodes. Whether a dynamic threshold is reached (e.g., the median contribution of all nodes in the current round) determines whether only nodes with sufficient contributions are allowed to participate in gradient verification. A report is submitted based on the verification results. Worker nodes are responsible for broadcasting gradient collection requests, sending signed gradients and metadata, and encapsulating and storing gradient blocks in the blockchain / database. Ultimately, all components collaborate to complete the gradient verification and data storage tasks.

[0157] Once the newly generated block is verified by a majority of nodes, each node will, according to... Update its local model to use the new global gradient in the validated block.

[0158] This embodiment extracts multi-scale features from the input face image using multiple pre-constructed Gabor filters to obtain the target feature vector. The Gabor filters consist of multiple complex filter kernels generated based on different directions and spatial frequencies. Each participating node is evaluated based on multiple evaluation dimensions to obtain multi-dimensional quantization parameters. These evaluation dimensions include marginal contribution, risk, and reputation dimensions. The multi-dimensional quantization parameters include marginal contribution degree, risk weight parameter, and target reputation value. The gradient training results uploaded by each participating node to the federated learning blockchain are obtained. These gradient training results represent the localization performance of each participating node. The face recognition model is trained using gradients; the gradient training results of each participating node are aggregated according to the multi-dimensional quantization parameters to obtain global gradient parameters; the local face recognition model is updated based on the global gradient parameters, and the target feature vector is input into the updated local face recognition model for face recognition; since this embodiment extracts features from the input face image at multiple scales, it enhances the representation ability of image features. By evaluating nodes in multiple dimensions, it effectively balances the data contribution weights provided by distributed nodes, improves the convergence speed and recognition accuracy of the face recognition model, and significantly improves the accuracy of face recognition.

[0159] Furthermore, embodiments of the present invention also propose a face recognition system, referring to... Figure 4 , Figure 4 This is a schematic diagram of the framework of a face recognition system in one embodiment. Figure 4 This paper presents a comprehensive system framework for a face recognition system, aiming to improve the efficiency and fairness of distributed collaboration through model aggregation and incentive mechanisms. Its content can be divided into four parts: First, based on literature analysis and current status analysis, a local feature extraction scheme centered on a convolutional neural network (CNN) is constructed, encompassing modules such as convolutional layers, GCS-LBP feature extraction (a hybrid method integrating traditional texture features and deep learning), and pooling layers, for efficient data feature mining. Second, in the model aggregation stage, a blockchain-based federated learning (BCFL) framework is introduced, combined with a verifiable random function (VRF) to ensure the security and randomness of communication between nodes, preventing malicious tampering. Third, to address the differences in contributions among participants in federated learning, a fair incentive mechanism is designed. Node contributions are evaluated through risk weight adjustment and Shapley Value correction, balancing the impact of computing resources and data quality on the results, thereby optimizing collaborative fairness. Finally, through experimental verification, the robustness, convergence efficiency, and incentive rationality of the framework are systematically tested by setting up the environment, preprocessing data, and configuring multiple nodes, ensuring the adaptability of the theoretical design to real-world scenarios. Overall, the framework of the face recognition system focuses on secure aggregation and fair incentives as its core innovations, balancing the accuracy of feature extraction with the reliability of distributed collaboration, and providing methodological support for the application of federated learning in complex scenarios.

[0160] Furthermore, embodiments of the present invention also propose a computer-readable storage medium storing a face recognition program, which, when executed by a processor, implements the steps of the face recognition method described above.

[0161] The computer-readable storage medium provided in this application may be, for example, a USB flash drive, but is not limited to, electrical, magnetic, optical, electromagnetic, infrared, or semiconductor systems, devices, or any combination thereof. More specific examples of computer-readable storage media may include, but are not limited to: electrical connections having 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 fiber, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof. In this embodiment, the computer-readable storage medium may be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, system, or device. The program code contained on the computer-readable storage medium may be transmitted using any suitable medium, including but not limited to: wires, optical cables, RF (Radio Frequency), etc., or any suitable combination thereof.

[0162] The aforementioned computer-readable storage medium may be included in the face recognition device; or it may exist independently and not assembled into the face recognition device.

[0163] Furthermore, this invention also proposes a computer program product, including a face recognition program, which, when executed by a processor, implements the steps of the face recognition method described above.

[0164] The specific implementation of the computer program product of the present invention is basically the same as the various embodiments of the face recognition method described above, and will not be repeated here.

[0165] Reference Figure 5 , Figure 5 This is a structural block diagram of an embodiment of the face recognition device of the present invention.

[0166] like Figure 5 As shown in the embodiment of the present invention, the face recognition device is applied to participating nodes in a federated learning blockchain. The federated learning blockchain includes multiple participating nodes, and the face recognition device includes:

[0167] The multi-scale feature extraction module 10 is used to extract multi-scale features from the input face image through multiple pre-constructed Gabor filters to obtain the target feature vector. The Gabor filters are composed of multiple complex filter kernels generated based on different directions and spatial frequencies.

[0168] The multidimensional evaluation module 20 is used to evaluate each participating node based on multiple evaluation dimensions to obtain multidimensional quantitative parameters. The evaluation dimensions include marginal contribution dimension, risk dimension and reputation dimension. The multidimensional quantitative parameters include marginal contribution degree, risk weight parameter and target reputation value.

[0169] The distributed gradient training module 30 is used to obtain the gradient training results uploaded by each participating node to the federated learning blockchain. The gradient training results are obtained by each participating node performing gradient training on the local face recognition model.

[0170] The gradient parameter aggregation module 40 is used to aggregate the gradient training results of each participating node according to the multidimensional quantization parameters to obtain global gradient parameters.

[0171] The face recognition module 50 is used to update the local face recognition model based on the global gradient parameters, and input the target feature vector into the updated local face recognition model for face recognition.

[0172] This embodiment extracts multi-scale features from the input face image using multiple pre-constructed Gabor filters to obtain the target feature vector. The Gabor filters consist of multiple complex filter kernels generated based on different directions and spatial frequencies. Each participating node is evaluated based on multiple evaluation dimensions to obtain multi-dimensional quantization parameters. These evaluation dimensions include marginal contribution, risk, and reputation dimensions. The multi-dimensional quantization parameters include marginal contribution degree, risk weight parameter, and target reputation value. The gradient training results uploaded by each participating node to the federated learning blockchain are obtained. These gradient training results represent the localization performance of each participating node. The face recognition model is trained using gradients; the gradient training results of each participating node are aggregated according to the multi-dimensional quantization parameters to obtain global gradient parameters; the local face recognition model is updated based on the global gradient parameters, and the target feature vector is input into the updated local face recognition model for face recognition; since this embodiment extracts features from the input face image at multiple scales, it enhances the representation ability of image features. By evaluating nodes in multiple dimensions, it effectively balances the data contribution weights provided by distributed nodes, improves the convergence speed and recognition accuracy of the face recognition model, and significantly improves the accuracy of face recognition.

[0173] The face recognition device provided in this application, employing the face recognition method described in the above embodiments, can solve the technical problems of face recognition. Compared with the prior art, the beneficial effects of the face recognition device provided in this application are the same as those of the face recognition method described in the above embodiments, and other technical features in the face recognition device are the same as those disclosed in the methods of the above embodiments, and will not be repeated here.

[0174] It should be understood that the above are merely illustrative examples and do not constitute any limitation on the technical solutions of the present invention. In specific applications, those skilled in the art can make settings as needed, and the present invention does not impose any restrictions on this.

[0175] It should be noted that the workflow described above is merely illustrative and does not limit the scope of protection of this invention. In practical applications, those skilled in the art can select some or all of the workflow to achieve the purpose of this embodiment according to actual needs, and no restrictions are imposed here.

[0176] In addition, for technical details not described in detail in this embodiment, please refer to the face recognition method provided in any embodiment of the present invention, which will not be repeated here.

[0177] It should be noted that, in this document, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or system that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or system. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or system that includes that element.

[0178] The sequence numbers of the above embodiments of the present invention are for descriptive purposes only and do not represent the superiority or inferiority of the embodiments.

[0179] Through the above description of the embodiments, those skilled in the art can clearly understand that the methods of the above embodiments can be implemented by means of software plus necessary general-purpose hardware platforms. Of course, they can also be implemented by hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product is stored in a storage medium (such as read-only memory / random access memory, magnetic disk, optical disk) and includes several instructions to cause a terminal device (which may be a mobile phone, computer, server, air conditioner, or network device, etc.) to execute the methods described in the various embodiments of the present invention.

[0180] The above are merely preferred embodiments of the present invention and do not limit the scope of the patent. Any equivalent structural or procedural transformations made based on the description and drawings of the present invention, or direct or indirect applications in other related technical fields, are similarly included within the scope of patent protection of the present invention.

Claims

1. A face recognition method, characterized in that, The face recognition method is applied to participating nodes in a federated learning blockchain, which includes multiple participating nodes. The face recognition method includes: The input face image is subjected to multi-scale feature extraction by multiple pre-constructed Gabor filters to obtain the target feature vector. The Gabor filters are composed of multiple complex filter kernels generated based on different directions and spatial frequencies. Each participating node is evaluated based on multiple evaluation dimensions to obtain multidimensional quantitative parameters. The evaluation dimensions include marginal contribution dimension, risk dimension and reputation dimension. The multidimensional quantitative parameters include marginal contribution degree, risk weight parameter and target reputation value. Obtain the gradient training results uploaded by each participating node to the federated learning blockchain. The gradient training results are obtained by each participating node performing gradient training on its local face recognition model. The gradient training results of each participating node are aggregated based on the multidimensional quantization parameters to obtain the global gradient parameters; The local face recognition model is updated based on the global gradient parameters, and the target feature vector is input into the updated local face recognition model for face recognition. The evaluation of each participating node based on multiple evaluation dimensions yields multidimensional quantitative parameters, including: The marginal contribution of each participating node is evaluated to obtain the marginal contribution degree of each participating node, which is calculated based on the following formula: in, Indicates participating nodes The marginal contribution Indicates that it includes participating nodes. A sub-alliance composed of all participating nodes This represents the revenue value of the sub-alliance. This indicates that the sub-alliance has removed participating nodes. The subsequent profit value, This indicates the contribution weight of the sub-alliance. Indicates the size of the sub-alliance, This represents the total number of nodes in the sub-alliance; Risk weights are assessed for each data supply node based on the integrity of each participating node, and the risk weights for each participating node are calculated based on the following formula: in, Indicates participating nodes Risk weights, Indicates participating nodes In the current block time period Integrity within the organization Indicates participating nodes The integrity vector; The reputation of each participating node is quantitatively evaluated based on its data quality information and task completion rate to obtain an initial reputation value for each participating node. The initial reputation value of each participating node is normalized based on the maximum reputation value of the participating node to obtain the target reputation value of each participating node.

2. The face recognition method as described in claim 1, characterized in that, The step of extracting multi-scale features from the input face image using multiple pre-constructed Gabor filters to obtain the target feature vector includes: The input face image is processed to obtain an initial grayscale image; The initial grayscale image is normalized to obtain candidate grayscale images; Multiple complex filter kernels are generated based on multiple sets of different directions and spatial frequencies, and multiple sets of Gabor filters are generated based on the complex filter kernels. The candidate grayscale image is decomposed at multiple scales using the multiple sets of Gabor filters. Based on the filter outputs, multiple texture feature maps with different directions and spatial frequencies are extracted. The wavelet function of the Gabor filter is defined by the following formula: in, Represents the spatial coordinates of pixels in the input face image. , This indicates the center frequency of the Gabor filter. Represents the directivity of a two-dimensional Gabor filter. The square of the standard deviation of the Gaussian envelope is used to control the spatial distribution width of the Gabor filter. This represents the Gaussian envelope term, used to constrain the spatial distribution of plane waves. This indicates the periodicity and directionality of the wavelet. Used for energy spectrum attenuation compensation; Feature analysis is performed on the multiple sets of texture feature maps with different directions and spatial frequencies to generate target feature vectors.

3. The face recognition method as described in claim 2, characterized in that, The step of performing feature analysis on the multiple sets of texture feature maps with different directions and spatial frequencies to generate target feature vectors includes: Traverse each pixel in the texture feature map and calculate the LBP value of each pixel; Based on the LBP values ​​of each pixel in the texture feature map, generate the LBP feature map corresponding to each Gabor filter; The LBP feature map is divided into multiple sub-blocks, and a distribution histogram of LBP values ​​is plotted for each sub-block. The distribution histograms of each sub-block are spliced ​​together to generate the target feature vector.

4. The face recognition method as described in claim 1, characterized in that, The risk weight assessment of each data supply node based on the integrity of each participating node, to obtain the risk weight of each participating node, includes: Obtain the historical integrity score of each participating node within the historical block time period; The noise level of the data provided by each participating node is assessed to obtain the data quality information of each participating node. Based on the data quality information, an exponential decay coefficient is generated for each participating node; The historical integrity score is updated based on the exponential decay coefficient to obtain the updated integrity score. The integrity score update is performed using the following formula: in, Indicates the current block time period. Indicates the time period of the previous block. Indicates participating nodes Integrity level in the previous block time period, Represents the exponential decay coefficient. Indicates participating nodes Integrity level within the current block time period, Indicates participating nodes The reward value obtained during local model training in the current block time period. Represents positive numbers. Indicates participating nodes Historical integrity score; Based on the updated integrity rating, risk weights are assessed for each data supply node to obtain the risk weights for each participating node.

5. The face recognition method as described in claim 4, characterized in that, The step of aggregating the gradient training results of each participating node according to the multidimensional quantization parameters to obtain global gradient parameters includes: Monitor abnormal behavior of each participating node; Mark participating nodes exhibiting abnormal behavior as abnormal nodes; The abnormal nodes are weighted according to the multidimensional quantization parameters to obtain a target weight vector, which is calculated based on the following formula: in, Represents the target weight vector; The gradient training results of each participating node are aggregated based on the target weight vector and the multidimensional quantization parameters to obtain global gradient parameters.

6. A face recognition device, characterized in that, The facial recognition device is applied to participating nodes in a federated learning blockchain, which includes multiple participating nodes. The facial recognition device includes: The multi-scale feature extraction module is used to extract multi-scale features from the input face image through multiple pre-constructed Gabor filters to obtain the target feature vector. The Gabor filters are composed of multiple complex filter kernels generated based on different directions and spatial frequencies. The multidimensional evaluation module is used to evaluate each participating node based on multiple evaluation dimensions to obtain multidimensional quantitative parameters. The evaluation dimensions include marginal contribution dimension, risk dimension and reputation dimension. The multidimensional quantitative parameters include marginal contribution degree, risk weight parameter and target reputation value. The distributed gradient training module is used to obtain the gradient training results uploaded by each participating node to the federated learning blockchain. The gradient training results are obtained by each participating node performing gradient training on its local face recognition model. The gradient parameter aggregation module is used to aggregate the gradient training results of each participating node according to the multidimensional quantization parameters to obtain global gradient parameters. The face recognition module is used to update the local face recognition model based on the global gradient parameters, and input the target feature vector into the updated local face recognition model for face recognition; The multidimensional evaluation module is also used to evaluate the marginal contribution of each participating node to obtain the marginal contribution degree of each participating node, which is calculated based on the following formula: in, Indicates participating nodes The marginal contribution Indicates that it includes participating nodes. A sub-alliance composed of all participating nodes This represents the revenue value of the sub-alliance. This indicates that the sub-alliance has removed participating nodes. The subsequent profit value, This indicates the contribution weight of the sub-alliance. Indicates the size of the sub-alliance, This represents the total number of nodes in the sub-alliance; Risk weights are assessed for each data supply node based on the integrity of each participating node, and the risk weights for each participating node are calculated based on the following formula: in, Indicates participating nodes Risk weights, Indicates participating nodes In the current block time period Integrity within the organization Indicates participating nodes The integrity vector; The reputation of each participating node is quantitatively evaluated based on its data quality information and task completion rate to obtain its initial reputation value. The initial reputation value of each participating node is then normalized based on its maximum reputation value to obtain its target reputation value.

7. A facial recognition device, characterized in that, The face recognition device includes: a memory, a processor, and a face recognition program stored in the memory and executable on the processor, the face recognition program being configured to implement the face recognition method as described in any one of claims 1 to 5.

8. A computer-readable storage medium, characterized in that, A face recognition program is stored on the computer-readable storage medium, which, when executed by a processor, implements the face recognition method as described in any one of claims 1 to 5.

9. A computer program product, characterized in that, The computer program product includes a face recognition program, which, when executed by a processor, implements the steps of the face recognition method as described in any one of claims 1 to 5.

Images