A method for constructing an air-to-ground lightweight human target posture estimation model and application

Abstract

The application is a kind of construction method and application of air-to-ground lightweight human target posture estimation model, belonging to the field of computer vision recognition technology.It includes: constructing a human posture estimation dataset;based on the HRNet network structure, designing a deep convolutional transformation module to obtain a multi-branch feature fusion module;designing a multi-scale feature fusion module to obtain a lightweight human posture estimation model in combination with the multi-branch feature fusion module;designing a lightweight model optimization algorithm to optimize the training process of the lightweight human posture estimation model and retrain the dataset in the optimized lightweight human posture estimation model.The lightweight human posture estimation model is verified by experiments, effectively solving the problems of high computational complexity and large memory occupation in existing human posture estimation algorithms, improving the computational efficiency of the model and making it more suitable for practical application scenarios.The method is simple and effective, has good robustness and better effect in air-to-ground human posture recognition.

Description

Technical Field

[0001] The present invention relates to the field of human body posture estimation, and in particular to a method for constructing and applying an air-to-ground lightweight human target posture estimation model. Background Art

[0002] Human pose estimation, an upstream task for behavior recognition, aims to locate the coordinates of key parts of ground targets in images, such as the head, shoulders, and hips, thereby constructing a skeleton sequence that provides input data for behavior recognition. Pose estimation can be used for violation detection in various dynamic scenarios, such as detecting potential threatening behaviors in intelligent security systems targeting key areas. Through specialized training for both distant and small targets, human pose estimation can be applied to aerial devices such as drones, effectively estimating human pose from aerial perspectives.

[0003] At present, neural networks for lightweight human pose estimation, such as the "GhostPoseNet" network disclosed in the patent number "CN 115187660 A" "A method and system for multi-person human pose estimation based on knowledge distillation", use knowledge distillation technology to generate joint point biases using the student heat map and the target joint point heat map of the data label, and dynamically adjust the knowledge transfer from the teacher network to the student network to construct a lightweight network. Although this method significantly reduces the number of network parameters, the network accuracy also shows a serious decline; in the document with the document number "10.48550 / arXiv.2104.06403", the "Lite-HRNet" network is disclosed, which uses the channel shuffle module Shuffle Block to replace the 3×3 convolution with high computational cost in the network. Although it significantly reduces the complexity of the model, the loss in model accuracy is relatively serious; in the document with the document number "10.48550 / arXiv.2204.10762", the "Dite-HRNet" network is disclosed, which uses dynamic segmentation convolution and adaptive context modeling to design two lightweight modules, dynamic multi-scale context and dynamic global context. Compared with previous methods, the accuracy has been improved, but it is still not ideal.

[0004] The common problem in the above-mentioned existing technologies is that the number of high-precision model parameters is too large to be deployed on platforms with limited computing power such as drones, while the accuracy of lightweight methods is often insufficient. At the same time, they cannot effectively estimate the human posture of distant targets and small targets photographed by drones 5-10 meters above the ground. Summary of the Invention

[0005] In response to the defects of the existing technologies analyzed above, the present invention proposes a method for constructing and applying a lightweight air-to-ground human target pose estimation model. It designs a lightweight human pose estimation network and introduces knowledge distillation technology to solve the problem that the existing human pose estimation algorithm model cannot effectively detect distant targets and small targets photographed by drones 5-10m above the ground on aerial equipment platforms.

[0006] To achieve the above object, the technical solution of the present invention is as follows:

[0007] In one aspect, the present invention discloses a method for constructing an air-to-ground lightweight human target pose estimation model, comprising the following steps:

[0008] S1: Construct a dataset for human pose estimation;

[0009] S2: Based on the HRNet network structure, a deep convolution transformation module is designed to replace the residual convolution module in the high-resolution network to obtain a multi-branch feature fusion module;

[0010] S3: Design a multi-scale feature fusion module and combine it with the multi-branch feature fusion module in step S2 to obtain a lightweight human pose estimation model;

[0011] S4: Design a lightweight model optimization algorithm and use it to optimize the training process of the lightweight human pose estimation model in step S3, and input the data in the dataset of step 1 into the optimized lightweight human pose estimation model for retraining.

[0012] Furthermore, the original data set in the above step S1 is selected from the distant view and small target data taken by the drone 5-10m above the ground, so as to better adapt to the air-to-ground mission, and then the obtained original data set is preprocessed to obtain the input sample.

[0013] Furthermore, the deep convolution transformation module designed in the above step S2 includes a construction module feature extraction layer and a construction module channel mapping layer.

[0014] Based on the above solution, the specific steps of constructing the module feature extraction layer are as follows:

[0015] Layer normalization is used to eliminate the differences in input samples, and 3×3 depth-separable convolution is used to extract the contextual information of the local area of ​​the pixel point of the input sample. Cascaded channel attention is introduced to learn the channel features of different input samples.

[0016] The calculation expression of the feature extraction layer is as follows:

[0017] F'=CCA(DWConv 3×3 (LayerNorm(x)))+x

[0018] Among them: x represents the input feature, LayerNorm represents the layer normalization operation, DWConv 3×3 represents a 3×3 depthwise separable convolution, CCA represents the cascaded channel attention, and F' represents the output features after the feature extraction layer.

[0019] Based on the above scheme, the channel mapping layer of the construction module includes layer normalization and two 1×1 convolution operations. The construction steps are as follows:

[0020] The output features of the feature extraction layer are used as new input features for layer normalization. The first 1×1 convolution is used to expand the feature channels by 2 times to fully learn the information in the new input features. The second 1×1 convolution is used to compress the number of feature channels to the original number of channels to retain important feature information.

[0021] The calculation expression of the channel mapping layer is as follows:

[0022]

[0023] in: They represent the first and second 1×1 convolutions respectively, and F” represents the final output feature of the depth convolution transformation module.

[0024] Furthermore, the design of the multi-scale feature fusion module in the above step S3 includes:

[0025] (1) Constructing a feature size redefinition layer;

[0026] (2) Construct feature information fusion layer.

[0027] Based on the above solution, we further construct a feature information fusion layer consisting of three key components:

[0028] The first layer of the feature information fusion layer operates on the superimposed features through 1×1 convolution to compress the number of channels;

[0029] The second layer uses 3×3 depth-wise separable convolution to fuse the spatial information of each branch feature;

[0030] Finally, cascade channel attention is used to further improve the accuracy of fusion. The calculation formula is as follows:

[0031] F b =CCA(DWConv 3×3 (Conv 1×1 (x b )))

[0032] Among them: F b Represents the output features after fusing feature information of different scales on branch b.

[0033] In the last step of feature information fusion, it is necessary to build a cascade channel attention mechanism. The specific steps are as follows:

[0034] First, construct high-resolution feature channel attention and input feature It will go through two 1×1 convolutions with different numbers of output channels and Obtain a feature matrix containing multi-spatial information and

[0035] Secondly, construct ECA channel attention, and The sizes of are defined as HW×1×1 and HW×C / 4 respectively. The SoftMax function is used to perform nonlinear activation on the feature Q, and then Q is transposed to make its size 1×1×HW, and then matrix product operation is performed with V to compress the spatial information to the weight parameter

[0036] Finally, 1×1 convolution is applied to MC Restore the number of channels and weight it with the input feature x to get the output feature The calculation formula is as follows:

[0037]

[0038] Among them, F SG and F SM Represents Sigmoid function and SoftMax function respectively, * represents convolution operation, represents matrix product, and ⊙ represents bit-by-bit multiplication operation.

[0039] Furthermore, in step S4, when constructing a lightweight model optimization algorithm, label knowledge distillation and feature knowledge distillation are included. The network model constructed in steps S1 and S2 is used as the student network, and the trained teacher network is used to supervise the student network. Specifically, the output features of each branch of the teacher model and the student model in the third stage are selected for supervised learning. In the fourth stage, the output features are only subjected to a single-layer 1×1 convolution to obtain the final key point heat map. The attention activation heat map of the b-th branch is obtained by compressing the features in each channel dimension, and its expression is as follows:

[0040]

[0041] Among them, b is the index of the branch with different resolutions, Before mapping The element value corresponding to the position (h, w) in the i-th channel feature map, To obtain the averaged attention activation heatmap;

[0042] After calculating the attention activation heatmaps of each branch of the teacher model and the student model, the loss function of feature knowledge distillation can be defined as the formula:

[0043]

[0044] Where N represents the number of branches in the third stage, and Represents the attention activation heatmaps of the student model and the teacher model on the b-th branch respectively.

[0045] On the other hand, the present invention also discloses a method for estimating the posture of a lightweight human target in an air-to-ground environment, comprising the following steps:

[0046] Get the image containing the person's body;

[0047] The acquired image is input into the optimized lightweight human pose estimation model constructed by any of the construction methods described above to obtain an output result of the human pose estimation.

[0048] Compared with the prior art, the present invention has the following beneficial effects:

[0049] (1) A lightweight human pose estimation module based on cascaded channel attention and a deep convolutional transformation module are constructed, which can improve the accuracy of pose estimation while keeping the model lightweight.

[0050] (2) A multi-scale feature fusion module is designed to enhance the semantic fusion effect between features of different scales, thereby enhancing the feature richness of the model, which helps the model better understand and process inputs with features of different scales and improves the model's adaptability to diverse scenarios.

[0051] (3) Design a knowledge distillation technology to retrain and optimize the network model, that is, use a more complex but more accurate teacher model to guide the lightweight student model to learn, so that the student model can improve its own detection accuracy without increasing the computational cost.

[0052] (4) Compared with similar methods, the designed network is more accurate in estimating the human pose of distant targets and small targets. It can effectively distinguish the estimation errors caused by the close proximity of key points of small targets and is more suitable for air-to-ground human target pose estimation. BRIEF DESCRIPTION OF THE DRAWINGS

[0053] Figure 1 Design a flow chart for lightweight human pose estimation based on knowledge distillation of the present invention;

[0054] Figure 2 It is a deep convolution transformation module diagram;

[0055] Figure 3 It is a multi-scale feature fusion module diagram;

[0056] Figure 4 is the cascade channel attention map;

[0057] Figure 5 is the feature knowledge distillation graph;

[0058] Figure 6 This is the human posture visualization result diagram. DETAILED DESCRIPTION

[0059] In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments.

[0060] The present invention provides a method and application for constructing an air-to-ground lightweight human target posture estimation model. Figure 1 As shown, the present invention first constructs a dataset specifically for training air-to-ground tasks, then designs a deep convolution transformation module and a multi-scale feature fusion module to produce a lightweight pose estimation network. Finally, the lightweight human pose estimation optimization based on knowledge distillation is used to complete the design of an efficient and compact pose estimation network model. The specific implementation steps are as follows:

[0061] Step 1: Obtain and preprocess the data set for human posture estimation;

[0062] 1.1 Obtain the original dataset for human pose estimation.

[0063] We use the COCO2017 public dataset and combine it with images of distant and small human objects taken by drones 5-10 meters above the ground to create an air-to-ground human pose estimation task dataset.

[0064] 1.2 Preprocess the original dataset.

[0065] The height and width of the air-to-ground human pose estimation task dataset images are expanded to a fixed aspect ratio of 4:3, and the adjusted boxes are cropped from the images and resized to a fixed size of 256×192mm. The method uses a random scale factor ([0.75, 1.25]), random rotation ([-30°, 30°]), horizontal flipping, and additional half-body augmentation to randomly enhance each person's sample, which is then used as the input sample.

[0066] Step 2: Based on the HRNet network structure, a deep convolution transformation module is designed to replace the residual convolution module in the high-resolution network to obtain a multi-branch feature fusion module.

[0067] like Figure 2 As shown in the figure, the module uses a combination of 3×3 depthwise separable convolution and cascaded channel attention to construct the feature extraction layer, thereby ensuring the lightweight model while capturing local detail information more specifically. The specific steps of the construction are as follows:

[0068] 2.1 Building module feature extraction layer;

[0069] Layer normalization is used to eliminate the differences between input samples. Layer normalization helps ensure that the distribution of input data between different samples is relatively consistent, providing more stable input for subsequent processing. At the same time, a depth-wise separable convolution with a stride of 3×3 is used to extract contextual information of the local area of ​​the pixel point. Cascaded channel attention is then introduced to learn the importance of each channel feature of different inputs, so that the model can focus more specifically on different channels of the input, thereby improving the perception of multi-channel features. Finally, a residual structure is added to alleviate the gradient vanishing and explosion problems during training. The calculation expression of the feature extraction layer is as follows:

[0070] F'=CCA(DWConv 3×3 (LayerNorm(x)))+x

[0071] Among them, x represents the input feature, LayerNorm represents the layer normalization operation, and DWConv 3×3 represents a 3×3 depthwise separable convolution, CCA represents the cascaded channel attention, and F' represents the output features after the feature extraction layer.

[0072] 2.2 Building Module Channel Mapping Layer;

[0073] The channel mapping layer structure includes normalization and two 1×1 convolution operations, aiming to improve the model's nonlinear expression capabilities. The first 1×1 convolution is used to double the number of feature channels. Through this operation, the model can more fully learn the information in the input features and increase its understanding of rich feature representations. Next, the second 1×1 convolution compresses the number of feature channels to the original number of channels. The purpose is to retain important feature information through nonlinear transformation while reducing the model's computational burden. When outputting features, the channel mapping layer also uses a residual structure to reduce the risk of gradient explosion during model training. The calculation formula of the channel mapping layer is as follows:

[0074]

[0075] in, They represent the first and second 1×1 convolutions respectively, and F” represents the final output feature of the depth convolution transformation module.

[0076] In step 3, a multi-scale feature fusion module is designed and combined with the multi-branch feature fusion module in step 2 to obtain a lightweight human posture estimation model.

[0077] Construct a multi-scale feature fusion module, which mainly includes a feature size redefinition layer and a feature information fusion layer. The former redefines the size of multi-scale features from different branches while keeping the number of channels unchanged, and superimposes the unified features on the channel dimension as the input of the feature information fusion layer; the latter is responsible for fusing the feature information output from the previous step, thereby improving the fusion accuracy and ensuring that features from different scales can be better integrated. The module structure is as follows: Figure 3 As shown, the specific steps of the construction are as follows:

[0078] 3.1 Constructing the feature size redefinition layer;

[0079] The feature resizing layer consists of three components: AdaptiveAvgPool, Identity, and Linear Upsampling. When fusing the resolution features of other branches, this layer applies AdaptiveAvgPool, Identity, and Linear Upsampling to the input high-resolution (HR), medium-resolution (MR), and low-resolution (LR) features, respectively, to unify their feature sizes. Subsequently, by stacking these equally sized features in the channel dimension, features of the same dimension are provided to each branch. The calculation formula is as follows:

[0080]

[0081] x b =Concat(x bi )(i∈1,...,s)

[0082] Among them, x i represents the characteristics of the i-th branch, x bi Indicates that AdaptiveAvgPool, Identity or Upsamling is used to pool x i The size of is defined as the feature after the feature size of the bth branch is the same, s represents the total number of branches in the current stage, x b Indicates that the feature x bi Features after superposition in the channel dimension.

[0083] 3.2 Construct feature information fusion layer;

[0084] The feature information fusion layer consists of three key components: 1×1 convolution, 3×3 depthwise separable convolution, and cascaded channel attention. This layer is designed to fuse feature information from different branches through these three operations to improve fusion accuracy.

[0085] The first layer of the feature information fusion layer operates on the superimposed features through 1×1 convolution, compressing its number of channels to make it equal to the number of feature channels of the current branch. This operation helps to fuse the information between feature channels and ensure that each channel can work better together. The second layer uses 3×3 depthwise separable convolution to fuse the spatial information of each branch feature. While maintaining computational efficiency, the depthwise separable convolution effectively captures the relationship between features in the spatial dimension and enhances the expressiveness of features. Finally, cascaded channel attention is used to further improve the accuracy of fusion. The cascaded channel attention mechanism can learn the importance of each channel feature of different inputs, making the fusion more targeted. Its calculation formula is as follows:

[0086] F b =CCA(DWConv 3×3 (Conv 1×1 (x b )))

[0087] Among them, F b Represents the output features after fusing feature information of different scales on branch b.

[0088] The third layer of the feature information fusion layer is the cascade channel attention mechanism, whose structure is as follows Figure 4 As shown in Figure 2, it consists of two parts: high-resolution feature channel attention and ECA (Efficient Channel Attention) channel attention.

[0089] In the design of high-resolution feature channel attention, the input feature First, it will go through two 1×1 convolutions with different numbers of output channels and To obtain a feature matrix containing more spatial information and

[0090] Secondly, and The sizes of are defined as HW×1×1 and HW×C / 4 respectively, and the SoftMax function is used to perform nonlinear activation on the feature Q, and then Q is transposed to make its size 1×1×HW, and then matrix product operation is performed with V, so that the spatial information is compressed to the weight parameter

[0091] Finally, 1×1 convolution is used for MC Restore the number of channels and weight it with the input feature x to get the output feature The calculation formula is as follows:

[0092]

[0093] Among them, F SG and F SM Represents Sigmoid function and SoftMax function respectively, * represents convolution operation, represents matrix product, and ⊙ represents bit-by-bit multiplication operation.

[0094] In the design of ECA channel attention, after performing global average pooling (GAP) on the input features, one-dimensional convolution Conv1d is used to realize cross-channel information interaction. The calculation formula is as follows:

[0095]

[0096] Step 4: Design a lightweight model optimization algorithm, use the lightweight model optimization algorithm to optimize the training process of the lightweight human pose estimation model obtained in step 3, and input the data in the data set of step 1 into the optimized lightweight human pose estimation model for retraining.

[0097] The lightweight human pose estimation model is retrained and optimized using offline distillation. At the same time, a feature distillation method based on attention activation is constructed to make the intermediate feature dimensions of the teacher model and the student model the same, so that the loss function can be directly calculated. Feature knowledge distillation is as follows: Figure 5 shown.

[0098] To achieve feature knowledge distillation, the present invention selects the output features of each branch of the third stage of the teacher model and the student model for supervised learning. In the third stage, both the teacher model and the student model have three feature branches of different scales, so that the model can perform multi-scale feature representation. In the student model, the number of feature channels on each branch is only half of that of the teacher model. Because feature distillation in the early stage of the model will force the student model to fit the output of the teacher model, resulting in excessive learning difficulty and poor training effect, only the features in the third stage are selected for knowledge transfer. In the final fourth stage of the model, the output features are only subjected to a single-layer 1×1 convolution to obtain the final key point heat map. The difference between its feature distribution and the predicted heat map is small, so label knowledge distillation can be directly used.

[0099] The feature distillation process based on attention activation is as follows: the feature map size on the b-th branch can be expressed as C b ×H b×W b , then the attention activation heat map of the b-th branch can be obtained by compressing the features in each channel dimension, as shown in the following formula:

[0100]

[0101] Among them, b is the index of the branch with different resolutions, Before mapping The element value corresponding to the position (h, w) in the i-th channel feature map, To obtain the averaged attention activation heat map; after calculating the attention activation heat map of each branch of the teacher model and the student model, the loss function of feature knowledge distillation can be defined as the formula:

[0102]

[0103] N represents the number of branches in the third stage, and The following table represents the attention activation heatmaps of the student model and the teacher model on branch b, respectively. Through the above loss function, the student model can learn the degree of attention and importance distribution of key points from the teacher model, making up for the limited transfer of high-level semantic feature knowledge by label knowledge distillation.

[0104] The present invention uses Object Keypoint Similarity (OKS) as the evaluation criterion, which is calculated as follows:

[0105]

[0106] The indicators of model accuracy include average precision AP, AP 50 、AP 75 、AP M 、AP L Series values ​​and average recall rate AR. As shown in Table 1, the present invention is compared with several traditional lightweight models in experiments.

[0107] Table 1 Comparative experimental results of the present invention and the lightweight model

[0108]

[0109] As can be seen from Table 1, the present invention has achieved significant improvements in various indicators after teacher model knowledge supervision training. When compared with traditional models such as Dite-HRNet-30, Lite-HRnet-30 and GhostPoseNet that use high-resolution backbone feature extraction networks, the performance indicators AP and AR of the lightweight human pose estimation model proposed in this invention have increased by 1.8 to 14.9 percentage points and 1.6 to 13.7 percentage points respectively, indicating that it can still maintain better performance under a relatively lightweight setting, while also achieving higher precision indicators AP 50 The results show that the lightweight human pose estimation model of the present invention is more effective in identifying small targets.

[0110] like Figure 6 As shown, the present invention performs human posture estimation on human body pictures taken from air to ground. It can be clearly observed that the lightweight human posture estimation model of the present invention has better robustness for scenes such as incompleteness and occlusion, and has better recognition effect for joint points in image details. These observation results emphasize that the present invention is more effective in air-to-ground scenarios, and has more accurate positioning performance in long-range scenes, small targets, and when joint points are too close when the drone is 5-10m above the ground, proving that the lightweight human posture estimation model of the present invention has better effect for air-to-ground human posture recognition.

[0111] The above description is an explanation of the specific implementation of the present invention, not a limitation of the present invention. Those skilled in the relevant technical field may also make various equivalent technical solutions without departing from the scope of the present invention, and therefore all equivalent technical solutions should be included in the patent protection scope of the present invention.

Claims

1. A method for constructing a lightweight air-to-ground human target pose estimation model, characterized in that: The following steps are involved: S1: Construct a dataset for human pose estimation; We use the COCO2017 public dataset and combine it with images of distant and small objects taken by drones 5-10 meters above the ground to create an air-to-ground human pose estimation dataset. S2: Based on the HRNet network structure, a deep convolution transformation module is designed to replace the residual convolution module in the high-resolution network to obtain a multi-branch feature fusion module; S3: Design a multi-scale feature fusion module and combine it with the multi-branch feature fusion module in step S2 to obtain a lightweight human pose estimation model; S4: Design a lightweight model optimization algorithm and use it to optimize the training process of the lightweight human pose estimation model in step S3, and input the data set in step S1 into the optimized lightweight human pose estimation model for retraining; In step S4, when constructing the lightweight model optimization algorithm, the output features of each branch of the teacher model and the student model in the third stage are selected for supervised learning. In the fourth stage, the output features are only subjected to a single layer of 1×1 convolution to obtain the final key point heat map. The attention activation heat map of the strip branch is obtained by compressing the features in each channel dimension, and its expression is as follows: in, is the index of branches with different resolutions, Before mapping The Channel feature map The element value corresponding to the position, To obtain the averaged attention activation heatmap; After calculating the attention activation heatmaps of each branch of the teacher model and the student model, the loss function of feature knowledge distillation is defined as the formula: in, represents the number of branches in the third stage, and Represents the student model and teacher model in the Heatmap of attention activations on the strip branches.

2. The method for constructing an air-to-ground lightweight human target pose estimation model according to claim 1, wherein: In step S1, the dataset construction includes: (1) Obtain the original dataset for human pose estimation; (2) Preprocess the original data set to obtain input samples.

3. The method for constructing an air-to-ground lightweight human target pose estimation model according to claim 2, wherein: In step S2, the design of the deep convolution transformation module includes: (1) Constructing module feature extraction layer; (2) Construct the module channel mapping layer.

4. The method for constructing an air-to-ground lightweight human target pose estimation model according to claim 3, wherein: The specific steps to construct the feature extraction layer are: Layer normalization is used to eliminate the differences in input samples, and 3×3 depth-separable convolution is used to extract the contextual information of the local area of ​​the pixel point of the input sample. Cascaded channel attention is introduced to learn the channel features of different input samples. The calculation expression of the feature extraction layer is as follows: in: represents the input features, Representation layer normalization operation, represents a 3×3 depthwise separable convolution, represents the cascaded channel attention, Represents the output features after the feature extraction layer.

5. The method for constructing an air-to-ground lightweight human target pose estimation model according to claim 4, wherein: Constructing the channel mapping layer includes layer normalization and two 1×1 convolution operations. The construction steps are: The output features of the feature extraction layer are used as new input features for layer normalization. The first 1×1 convolution is used to expand the feature channels by 2 times to fully learn the information in the new input features. The second 1×1 convolution is used to compress the number of feature channels to the original number of channels to retain important feature information. The calculation expression of the channel mapping layer is as follows: in: , Represent the first and second 1×1 convolutions respectively, Represents the final output features of the deep convolution transformation module.

6. The method for constructing an air-to-ground lightweight human target pose estimation model according to claim 1, wherein: In step S3, the design of the multi-scale feature fusion module includes: (1) Constructing a feature size redefinition layer; (2) Construct the feature information fusion layer.

7. The method for constructing an air-to-ground lightweight human target pose estimation model according to claim 6, characterized in that: Constructing the feature information fusion layer consists of three key components: 1×1 convolution, 3×3 depth-wise separable convolution, and cascaded channel attention.

8. The method for constructing an air-to-ground lightweight human target pose estimation model according to claim 7, wherein: The specific steps to build the cascade channel attention module are: First, construct high-resolution feature channel attention and input feature It will go through two 1×1 convolutions with different numbers of output channels and , obtain the feature matrix containing multi-spatial information and ; Secondly, construct ECA channel attention, and The sizes are defined as and , using the SoftMax function to Perform nonlinear activation and then Transpose it so that its size is , and then with Perform matrix product operations to compress spatial information into weight parameters ; Finally, yes Using 1×1 convolution Recover the number of channels and compare with the input features Weighted output features , The calculation formula is as follows: in, and Represent the Sigmoid function and SoftMax function respectively, represents the convolution operation, represents matrix product, Represents a bitwise multiplication operation.

9. A lightweight air-to-ground human target pose estimation method, characterized in that: The following steps are involved: Get the image containing the person's body; The acquired image is input into the optimized lightweight human posture estimation model constructed by the construction method according to any one of claims 1 to 8 to obtain an output result of the human posture estimation.

Images