Airplane detection method, device and equipment in optical remote sensing image

Abstract

The invention provides an aircraft detection method, device and equipment in an optical remote sensing image, and relates to the technical field of optical remote sensing image processing. The method comprises the following steps: during aircraft detection, firstly obtaining a target optical remote sensing image to be detected, and inputting the target optical remote sensing image into an aircraft detection model to obtain an aircraft detection result output by the aircraft detection model; wherein the aircraft detection model is obtained by extracting different levels of image features from an optical remote sensing image in a training set based on an image denoising network ADNet of an improved lightweight attention mechanism, and training an initial aircraft detection model based on the extracted multi-level image features and corresponding labeling labels. By adopting the technical scheme provided by the invention, the defect of relatively low detection accuracy of a traditional aircraft detection algorithm in the optical remote sensing image can be solved, so that the accuracy of an aircraft detection result in the target optical remote sensing image is improved.

Description

Technical Field

[0001] The present application relates to the field of image technology, and in particular to a method, device and equipment for detecting aircraft in optical remote sensing images. Background Art

[0002] In scenarios such as traffic safety, disaster assessment, economic construction, and military reconnaissance, it is crucial to detect aircraft in optical remote sensing images. However, due to the unique morphological characteristics, complex background environment, and variable posture changes of aircraft, detecting aircraft in optical remote sensing images has become a very challenging task.

[0003] Traditional aircraft detection algorithm in optical remote sensing images: use a fixed scale to extract features of aircraft in optical remote sensing images, and classify the extracted features through machine learning, so as to detect aircraft based on the classification results.

[0004] However, the use of the above-mentioned traditional aircraft detection algorithm in optical remote sensing images will result in low accuracy of detection results. Summary of the invention

[0005] The present application provides a method, device and equipment for detecting aircraft in optical remote sensing images, which are used to solve the defect of low detection result accuracy of traditional aircraft detection algorithms in optical remote sensing images, thereby improving the accuracy of aircraft detection results in optical remote sensing images.

[0006] The present application provides a method for detecting aircraft in an optical remote sensing image, comprising: Acquire an optical remote sensing image of a target to be detected; Inputting the target optical remote sensing image into an aircraft detection model to obtain an aircraft detection result output by the aircraft detection model; Among them, the aircraft detection model is obtained by extracting image features of different levels from the optical remote sensing images in the training set based on the improved lightweight attention mechanism image denoising network ADNet, and training the initial aircraft detection model based on the extracted multi-level image features and corresponding annotation labels.

[0007] According to a method for detecting aircraft in an optical remote sensing image provided by the present application, the initial aircraft detection model includes the image denoising network ADNet based on an improved lightweight attention mechanism, a bidirectional feature pyramid network BiFPN and a detection head network; The aircraft detection model is trained based on the following method: Inputting the optical remote sensing image into the image denoising network ADNet to obtain the multi-level image features output by the image denoising network ADNet; wherein the multi-level image features include image features of the first level, image features of the second level, and image features of the third level; Inputting the image features of the first level, the image features of the second level, and the image features of the third level into the bidirectional feature pyramid network BiFPN, obtaining first scale features corresponding to the image features of the first level, second scale features corresponding to the image features of the second level, and third scale features corresponding to the image features of the third level output by the bidirectional feature pyramid network BiFPN; Inputting the first scale feature, the second scale feature and the third scale feature into the detection head network to obtain the aircraft category prediction probability and the aircraft detection frame output by the detection head network; Based on the classification loss function corresponding to the aircraft category prediction probability and the loss function corresponding to the aircraft detection box, the model parameters in the initial aircraft detection model are updated to obtain the aircraft detection model.

[0008] According to a method for detecting aircraft in an optical remote sensing image provided by the present application, the image denoising network ADNet includes a sparse module SB, a feature enhancement module FEB and an attention module based on an improved lightweight attention mechanism connected in series in sequence; The step of inputting the optical remote sensing image into the image denoising network ADNet to obtain the multi-level image features output by the image denoising network ADNet includes: Inputting the optical remote sensing image into the sparse module SB to obtain the first-level image features output by the sparse module SB; Inputting the image features of the first level and the optical remote sensing image into the feature enhancement module FEB, and outputting the image features of the second level and the convolution image features through the feature enhancement module FEB; The image features of the second level and the convolutional image features are both input into the attention module based on the improved lightweight attention mechanism, and the image features of the third level are output through the attention module.

[0009] According to a method for detecting aircraft in an optical remote sensing image provided by the present application, the feature enhancement module FEB includes a first convolution normalization unit, a second convolution normalization unit, a third convolution normalization unit, a convolution layer and an activation function layer which are sequentially connected in series; The step of inputting the image features of the first level and the optical remote sensing image into the feature enhancement module FEB, and outputting the image features of the second level and the convolution image features through the feature enhancement module includes: Inputting the image feature input of the first level into the first convolution normalization unit, the second convolution normalization unit and the third convolution normalization unit in sequence, to obtain the convolution normalization feature output by the third convolution normalization unit; Inputting the convolution normalization feature into the convolution layer to obtain the convolution image feature output by the convolution layer; The sum of the optical remote sensing image and the convolution image features is input into the activation function layer to obtain the second-level image features output by the activation function layer.

[0010] According to a method for detecting aircraft in an optical remote sensing image provided by the present application, the detection head network includes a plurality of detection heads of different scales connected in parallel, the plurality of detection heads including a detection head corresponding to a first scale, a detection head corresponding to a second scale, and a detection head corresponding to a third scale; The first scale feature, the second scale feature and the third scale feature are all input into the detection head network to obtain the aircraft category prediction probability and the aircraft detection frame output by the detection head network, including: Inputting the first scale feature into a detection head corresponding to the first scale to obtain a corresponding first aircraft category prediction probability and a first aircraft detection frame; Inputting the second scale feature into the detection head corresponding to the second scale to obtain the corresponding second aircraft category prediction probability and second aircraft detection frame; Inputting the third scale feature into the detection head corresponding to the third scale to obtain a corresponding third aircraft category prediction probability and a third aircraft detection frame; Determining the aircraft category prediction probability based on the first aircraft category prediction probability, the second aircraft category prediction probability, and the third aircraft category prediction probability; The aircraft detection frame is determined based on the first aircraft detection frame, the second aircraft detection frame, and the third aircraft detection frame.

[0011] According to an aircraft detection method in an optical remote sensing image provided by the present application, the classification loss function corresponding to the aircraft category prediction probability can be shown in the following formula: in, represents the classification loss function corresponding to the predicted probability of the aircraft category, represents the number of the optical remote sensing images, Indicates Optical remote sensing images, Indicates Target aircraft categories in optical remote sensing images The predicted probability of Indicates the preset probability threshold.

[0012] The present application also provides an aircraft detection device in an optical remote sensing image, comprising: An acquisition unit, used for acquiring an optical remote sensing image of a target to be detected; A detection unit, used for inputting the target optical remote sensing image into an aircraft detection model to obtain an aircraft detection result output by the aircraft detection model; Among them, the aircraft detection model is obtained by extracting image features of different levels from the optical remote sensing images in the training set based on the improved lightweight attention mechanism image denoising network ADNet, and training the initial aircraft detection model based on the extracted multi-level image features and corresponding annotation labels.

[0013] The present application also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, the method for detecting an aircraft in an optical remote sensing image as described in any one of the above is implemented.

[0014] The present application also provides a non-transitory computer-readable storage medium having a computer program stored thereon. When the computer program is executed by a processor, the method for detecting an aircraft in an optical remote sensing image as described in any one of the above is implemented.

[0015] The present application also provides a computer program product, comprising a computer program, wherein when the computer program is executed by a processor, the method for detecting an aircraft in an optical remote sensing image as described in any one of the above is implemented.

[0016] The present application provides an aircraft detection method, device and apparatus in an optical remote sensing image. When performing aircraft detection, the target optical remote sensing image to be detected is input into an aircraft detection model to obtain an aircraft detection result output by the aircraft detection model; wherein the aircraft detection model is obtained by extracting image features of different levels from the optical remote sensing images in the training set based on an improved image denoising network ADNet with a lightweight attention mechanism, and training the initial aircraft detection model based on the extracted multi-level image features and the corresponding annotation labels. In view of the fact that the aircraft detection model has a good feature enhancement expression capability, it can effectively capture and utilize image features of different levels. Therefore, when the target optical remote sensing image to be detected is input into the aircraft detection model, and the aircraft detection result is determined by the aircraft detection model, the image feature information of different levels of the target optical remote sensing image can be effectively captured and utilized. In this way, the aircraft detection result is determined by combining the image feature information of different levels, and the aircraft detection result of the target optical remote sensing image can be obtained quickly and accurately, which solves the defect of low detection accuracy of the traditional aircraft detection algorithm in optical remote sensing images, thereby improving the accuracy of the aircraft detection result in the target optical remote sensing image. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] In order to more clearly illustrate the technical solutions in the present application or the prior art, a brief introduction will be given below to the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0018] Figure 1 A schematic flow chart of a method for detecting aircraft in an optical remote sensing image provided in an embodiment of the present application.

[0019] Figure 2 A schematic diagram of the structure of an initial aircraft detection model provided in an embodiment of the present application.

[0020] Figure 3 A flowchart of a method for training an aircraft detection model provided in an embodiment of the present application.

[0021] Figure 4 A schematic diagram of the structure of an aircraft detection device in an optical remote sensing image provided in an embodiment of the present application.

[0022] Figure 5 A schematic diagram of the physical structure of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0023] In order to make the purpose, technical solutions and advantages of this application clearer, the technical solutions in this application will be clearly and completely described below in conjunction with the drawings in this application. Obviously, the described embodiments are part of the embodiments of this application, not all of the embodiments. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0024] In the embodiments of the present application, "at least one" means one or more, and "more" means two or more. "And / or" describes the association relationship of associated objects, indicating that three relationships may exist. For example, A and / or B can mean: A exists alone, A and B exist at the same time, and B exists alone, where A and B can be singular or plural. In the text description of the present application, the character " / " generally indicates that the associated objects before and after are in an "or" relationship.

[0025] The technical solution provided by the embodiment of the present application can be adapted to scenarios such as traffic safety, disaster assessment, economic construction and military reconnaissance. Traditional aircraft detection algorithm in optical remote sensing images: use a fixed scale to extract features of aircraft in optical remote sensing images, and classify the extracted features through machine learning, so as to detect aircraft based on the classification results.

[0026] However, the use of the above-mentioned traditional aircraft detection algorithm in optical remote sensing images has the problem of low detection accuracy.

[0027] In order to solve the defect of low detection result accuracy of traditional aircraft detection algorithms in optical remote sensing images, thereby improving the accuracy of aircraft detection results in optical remote sensing images, an embodiment of the present application provides a method for detecting aircraft in optical remote sensing images, which extracts image features of different levels from optical remote sensing images in a training set based on an improved lightweight attention mechanism image denoising network (Attention-guided Denoising Convolutional Neural Network, ADNet), and trains an initial aircraft detection model based on the extracted multi-level image features and corresponding annotation labels to obtain an aircraft detection model. Given that the aircraft detection model has good feature enhancement expression capabilities and can effectively capture and utilize image features at different levels, when the target optical remote sensing image to be detected is input into the aircraft detection model and the aircraft detection result is determined by the aircraft detection model, the image feature information at different levels of the target optical remote sensing image can be effectively captured and utilized. In this way, by combining image feature information at different levels to determine the aircraft detection result, the aircraft detection result of the target optical remote sensing image can be obtained quickly and accurately, which solves the defect of low detection accuracy of traditional aircraft detection algorithms in optical remote sensing images, thereby improving the accuracy of aircraft detection results in target optical remote sensing images.

[0028] It can be understood that the executor of the present method can be an electronic device such as an aircraft detection device, a computer or a server, or an aircraft detection device in an optical remote sensing image set in the electronic device. The aircraft detection device in the optical remote sensing image can be implemented by software, hardware or a combination of the two, and can be specifically set according to actual needs.

[0029] The following specific embodiments will be used to describe the aircraft detection method in optical remote sensing images provided by the present application in detail. It is understandable that the following specific embodiments can be combined with each other, and the same or similar concepts or processes may not be described in detail in some embodiments.

[0030] Figure 1 A flowchart of a method for detecting aircraft in an optical remote sensing image provided by an embodiment of the present application is provided. For example, see Figure 1 As shown, the method for detecting aircraft in an optical remote sensing image may include: S101, obtaining an optical remote sensing image of a target to be detected.

[0031] For example, the optical remote sensing image of the target to be detected can be obtained by remote sensing satellite photography, can be obtained from an open source remote sensing image database, or can be obtained by other means, such as obtaining the optical remote sensing image of the target to be detected from a remote sensing data sharing library, etc. The specific settings can be made according to actual needs.

[0032] Typically, before inputting the target optical remote sensing image into the aircraft detection model, the target optical remote sensing image may be preprocessed, such as denoising, geometric correction, atmospheric correction, resolution adjustment, etc., and the preprocessed target optical remote sensing image is used as the input of the aircraft detection model to output the aircraft detection result through the aircraft detection model, i.e., execute the following S102.

[0033] S102: Input the target optical remote sensing image into the aircraft detection model to obtain the aircraft detection result output by the aircraft detection model.

[0034] Among them, the aircraft detection model is obtained by extracting image features of different levels from the optical remote sensing images in the training set based on the improved lightweight attention mechanism image denoising network ADNet, and training the initial aircraft detection model based on the extracted multi-level image features and corresponding annotation labels.

[0035] For example, the aircraft detection result may include the position of the aircraft in the optical remote sensing image, the probability of the aircraft category to which the aircraft belongs, etc., which may be specifically set according to actual needs.

[0036] It can be seen that in the embodiment of the present application, when performing aircraft detection, the target optical remote sensing image to be detected is input into the aircraft detection model to obtain the aircraft detection result output by the aircraft detection model; wherein, the aircraft detection model is obtained by extracting image features of different levels from the optical remote sensing images in the training set based on the image denoising network ADNet with an improved lightweight attention mechanism, and training the initial aircraft detection model based on the extracted multi-level image features and the corresponding annotation labels. In view of the fact that the aircraft detection model has a good feature enhancement expression ability, it can effectively capture and utilize image features of different levels. Therefore, when the target optical remote sensing image to be detected is input into the aircraft detection model, and the aircraft detection result is determined by the aircraft detection model, the image feature information of different levels of the target optical remote sensing image can be effectively captured and utilized. In this way, the aircraft detection result is determined by combining the image feature information of different levels, and the aircraft detection result of the target optical remote sensing image can be obtained quickly and accurately, which solves the defect of low detection accuracy of the aircraft detection algorithm in the traditional optical remote sensing image, thereby improving the accuracy of the aircraft detection result in the target optical remote sensing image.

[0037] based on Figure 1In the embodiment shown, for example, in the embodiment of the present application, the above-mentioned initial aircraft detection model can be an improved YOLOv10 network model, and the backbone network Backbone in the YOLOv10 network model is changed to an image denoising network ADNet based on an improved lightweight attention mechanism, and the lightweight attention mechanism is used to replace the AB module in the YOLOv10 network model; and the Neck structure in the YOLOv10 network model is modified, and the original path aggregation network (Path Aggregation Network, PAN) structure is replaced with a bidirectional feature pyramid network (Bidirectional Feature Pyramid Network, BiFPN) structure, and the improved YOLOv10 network model is used as the initial aircraft detection model.

[0038] For example, see Figure 2 As shown, Figure 2 A schematic diagram of the structure of an initial aircraft detection model provided in an embodiment of the present application, wherein the initial aircraft detection model may include an image denoising network ADNet based on an improved lightweight attention mechanism, a bidirectional feature pyramid network BiFPN and a detection head network, combined with Figure 2 The initial aircraft detection model shown below will be Figure 3 The illustrated embodiment describes a method for training an aircraft detection model.

[0039] Normally, when training the initial aircraft detection model, it is necessary to first obtain optical remote sensing images, perform image preprocessing on the optical remote sensing images and annotate them into data sets, and divide the rotating box detection targets of different aircraft categories according to a preset ratio, and randomly divide them into training sets, validation sets, and test sets. Among them, the training set and validation set are used to train the initial aircraft detection model to obtain the aircraft detection model; the test data is used to test and verify the trained aircraft detection model to obtain the final aircraft detection model. For example, the ratio of the training set, validation set, and test set can be 8:1:1, which can be set according to actual needs.

[0040] Combination Figure 2 The initial aircraft detection model shown in FIG. 1 can be used as the basis of the aircraft detection model to be trained to train the aircraft detection model. For example, see Figure 3 As shown, Figure 3 A flowchart of a method for training an aircraft detection model provided in an embodiment of the present application is provided. The method for training an aircraft detection model may include: S301, inputting the optical remote sensing image into the image denoising network ADNet, and obtaining the multi-level image features output by the image denoising network ADNet; wherein the multi-level image features include the image features of the first level, the image features of the second level and the image features of the third level.

[0041] The image denoising network can be an ADNet network. In an embodiment of the present application, the backbone network Backbone in the YOLOv10 network model can be changed to an image denoising network based on an improved lightweight attention mechanism (Coordinate Attention), and the lightweight attention mechanism can be used to replace the AB module in ADNet.

[0042] For example, in the embodiments of the present application, it can be combined with Figure 2 As shown in the figure, the image denoising network includes a sparse block (SB), a feature enhancement block (FEB), and an attention (Coordinate Attention) module based on an improved lightweight attention mechanism, which are connected in series.

[0043] When the optical remote sensing image is input into the image denoising network ADNet to obtain the multi-level image features output by the image denoising network ADNet, the optical remote sensing image can be first input into the sparse module SB to obtain the first-level image features output by the sparse module; and the first-level image features and the optical remote sensing image are both input into the feature enhancement module FEB, and the second-level image features and convolution image features are output through the feature enhancement module FEB; and the second-level image features and convolution image features are both input into the attention module based on the improved lightweight attention mechanism, and the third-level image features are output through the attention module.

[0044] Among them, the optical remote sensing image is input into the sparse module, and the sparse module SB can use hollow convolution and ordinary convolution to implement the sparse mechanism to obtain sparse convolution features. The sparse convolution features are the first-level image features output by the sparse module SB to achieve a balance between efficiency and performance.

[0045] The first-level image features and optical remote sensing images are input into the feature enhancement module FEB. The feature enhancement module FEB integrates shallow and deep feature information through long-range paths to obtain enhanced convolution features. The enhanced convolution features are the second-level image features output by the feature enhancement module FEB to enhance the feature expression capability of the aircraft detection model.

[0046] The second-level image features and convolutional image features are input into the attention module based on the improved lightweight attention mechanism to capture the long-range dependencies between spaces and channels, obtain the local and global relationships of pixels in space, and obtain the third-level image features output by the attention module.

[0047] It can be understood that in the embodiment of the present application, a deep residual structure and a deep supervised 1×1 convolutional layer are introduced for the above-mentioned sparse module SB, feature enhancement module FEB and attention module based on the improved lightweight attention mechanism.

[0048] Among them, the deep residual structure adds residual branch output to the sparse module SB, the feature enhancement module FEB and the attention module based on the improved lightweight attention mechanism for multi-scale feature extraction; the deep supervision 1×1 convolution layer is the connection layer after the deep residual structure, which is used to adjust the number of input feature channels.

[0049] For example, the second-level image features and the convolutional image features are input into an attention module based on the improved lightweight attention mechanism. When the third-level image features are output through the attention module, each channel is encoded from the horizontal coordinate direction and the vertical coordinate direction. The encoding expressions of these two spatial directions can be seen in the following formulas 1 and 2.

[0050] Formula 1 Formula 2 in, represents the width of the input feature, H represents the height of the input feature, Represents pixel The coordinates of Indicates that the second-level image features and the convolution image features are fused in the horizontal direction for the channel Features, Indicates that the second-level image features and the convolution image features are fused in the vertical direction for the channel Features, Indicates the channel The encoded horizontal feature map, Indicates that the channel Encoded vertical feature map.

[0051] The encoded horizontal feature map and vertical feature map are transformed using a shared 1×1 convolution to obtain an aggregated feature map, as shown in the following formula 3.

[0052] Formula 3 in, represents the aggregate feature map, represents the horizontal feature map, Represents a vertical feature map without distinguishing channels.

[0053] The above aggregate feature map Split into two separate tensors along the spatial dimension and , using two 1×1 convolutions to transform the feature map and feature map The feature map whose number of channels is consistent with the input feature of the attention module is transformed, as shown in the following formula 4.

[0054] Formula 4 in, The feature map The feature map obtained after transformation is The feature map The feature map obtained after transformation.

[0055] The transformed feature map is then calculated as shown in the following formula 5 to obtain the third-level image features output by the attention module.

[0056] Formula 5 in, Indicates that in the channel The third level image features, Represents the features after the fusion of the second-level image features and the convolutional image features. Represents the feature map of channel c The feature map obtained after transformation is Represents the feature map of channel c The feature map obtained after transformation.

[0057] It can be seen that in the embodiment of the present application, when training the aircraft detection model, by improving the YOLOv10 model, the backbone network Backbone in the YOLOv10 network model is changed to an image denoising network ADNet based on an improved lightweight attention mechanism, which can not only effectively cope with noise interference and enhance the stability and reliability of the aircraft detection system; but also the introduced ADNet, the attention module based on the improved lightweight attention mechanism, and the bidirectional feature pyramid network BiFPN have lower computational complexity, which significantly improves the aircraft detection efficiency, meets the application scenarios with high real-time requirements, greatly improves the detection accuracy, reduces the probability of false detection and missed detection, and provides an innovative and efficient solution for aircraft target detection in optical remote sensing images.

[0058] For example, in the embodiments of the present application, it can be combined with Figure 2 As shown, the feature enhancement module FEB may include a first convolution normalization unit, a second convolution normalization unit, a third convolution normalization unit, a convolution layer, and an activation function layer which are sequentially connected in series.

[0059] The image features of the first level and the optical remote sensing image are input into the feature enhancement module FEB. When the image features and convolution image features of the second level are output through the feature enhancement module FEB, the image features of the first level can be input into the first convolution normalization unit, the second convolution normalization unit and the third convolution normalization unit in sequence to obtain the convolution normalization features output by the third convolution normalization unit; and the convolution normalization features are input into the convolution layer to obtain the convolution image features output by the convolution layer; and the sum of the optical remote sensing image and the convolution image features is input into the activation function layer to obtain the image features of the second level output by the activation function layer, and the image features of the second level are the enhanced convolution features output by the feature enhancement module.

[0060] After the first-level image features are obtained through the sparse module SB, the second-level image features are obtained through the feature enhancement module FEB, and the third-level image features are obtained through the attention module, the following S302 can be executed: S302, inputting the image features of the first level, the image features of the second level, and the image features of the third level into the bidirectional feature pyramid network BiFPN, and obtaining the first scale features corresponding to the image features of the first level, the second scale features corresponding to the image features of the second level, and the third scale features corresponding to the image features of the third level output by the bidirectional feature pyramid network BiFPN.

[0061] In the embodiment of the present application, the Neck structure in the YOLOv10 network model can be modified, and the original path aggregation network PAN structure can be replaced with a bidirectional feature pyramid network BiFPN structure. Among them, the bidirectional feature pyramid network BiFPN is an improved feature pyramid network (Feature Pyramid Networks, FPN), which realizes bidirectional feature fusion by introducing three top-down and bottom-up paths, obtains the first scale feature corresponding to the image feature of the first level, the second scale feature corresponding to the image feature of the second level, and the third scale feature corresponding to the image feature of the third level, adds learnable weights to better utilize the feature information of different levels, simplifies the feature fusion path and nodes, and reduces the computational complexity.

[0062] S303: Input the first scale feature, the second scale feature, and the third scale feature into the detection head network to obtain the aircraft category prediction probability and the aircraft detection box output by the detection head network.

[0063] For example, in the embodiments of the present application, it can be combined with Figure 2 As shown, the detection head network includes multiple detection heads of different scales connected in parallel, and the multiple detection heads include a detection head corresponding to a first scale, a detection head corresponding to a second scale, and a detection head corresponding to a third scale.

[0064] When the first scale feature, the second scale feature and the third scale feature are all input into the detection head network to obtain the aircraft category prediction probability and the aircraft detection frame output by the detection head network, the first scale feature can be input into the detection head corresponding to the first scale to obtain the corresponding first aircraft category prediction probability and the first aircraft detection frame; the second scale feature is input into the detection head corresponding to the second scale to obtain the corresponding second aircraft category prediction probability and the second aircraft detection frame; the third scale feature is input into the detection head corresponding to the third scale to obtain the corresponding third aircraft category prediction probability and the third aircraft detection frame; then, based on the first aircraft category prediction probability, the second aircraft category prediction probability and the third aircraft category prediction probability, the aircraft category prediction probability is determined; based on the first aircraft detection frame, the second aircraft detection frame and the third aircraft detection frame, the aircraft detection frame is determined.

[0065] Taking determining the aircraft category prediction probability based on the first aircraft category prediction probability, the second aircraft category prediction probability and the third aircraft category prediction probability as an example, for example, the three aircraft category prediction probabilities can be weighted, and the weighted result can be used as the final aircraft category prediction probability, which can be set according to actual needs.

[0066] After determining the loss function corresponding to the aircraft category prediction probability and the aircraft detection box, the model parameters in the initial aircraft detection model can be updated based on the classification loss function corresponding to the aircraft category prediction probability and the loss function corresponding to the aircraft detection box to train the aircraft detection model, that is, execute the following S304.

[0067] S304: Based on the classification loss function corresponding to the aircraft category prediction probability and the loss function corresponding to the aircraft detection box, the model parameters in the initial aircraft detection model are updated to obtain the aircraft detection model.

[0068] For example, in an embodiment of the present application, the OHEM loss function may be used to update the model parameters in the initial aircraft detection model, and the classification loss function corresponding to the aircraft category prediction probability may be shown in the following formula 6.

[0069] Formula 6 in, represents the classification loss function corresponding to the predicted probability of the aircraft category, represents the number of optical remote sensing images, Indicates Optical remote sensing images, Indicates Target aircraft categories in optical remote sensing images The predicted probability of represents a preset probability threshold. For example, It can be set to 0.7, and can be set according to actual needs.

[0070] In this way, based on the classification loss function corresponding to the predicted probability of the aircraft category and the loss function corresponding to the aircraft detection frame, the model parameters in the initial aircraft detection model are updated until the aircraft detection model is trained. Given that the aircraft detection model has good feature enhancement expression capabilities and can effectively capture and utilize image features at different levels, when the target optical remote sensing image to be detected is input into the aircraft detection model and the aircraft detection result is determined by the aircraft detection model, the image feature information at different levels of the target optical remote sensing image can be effectively captured and utilized. In this way, the aircraft detection result is determined by combining the image feature information at different levels, and the aircraft detection result of the target optical remote sensing image can be obtained quickly and accurately, which solves the defect of low detection accuracy of the traditional aircraft detection algorithm in optical remote sensing images, thereby improving the accuracy of the aircraft detection result in the target optical remote sensing image.

[0071] The following is a description of the aircraft detection device in the optical remote sensing image provided by the present application. The aircraft detection device in the optical remote sensing image described below and the aircraft detection method in the optical remote sensing image described above can be referenced to each other.

[0072] Figure 4 This is a schematic diagram of the structure of an aircraft detection device in an optical remote sensing image provided in an embodiment of the present application. For example, see Figure 4 As shown, the aircraft detection device 40 in the optical remote sensing image may include: An acquisition unit 401 is used to acquire an optical remote sensing image of a target to be detected; A detection unit 402, configured to input the target optical remote sensing image into an aircraft detection model to obtain an aircraft detection result output by the aircraft detection model; Among them, the aircraft detection model is obtained by extracting image features of different levels from the optical remote sensing images in the training set based on the improved lightweight attention mechanism image denoising network ADNet, and training the initial aircraft detection model based on the extracted multi-level image features and corresponding annotation labels.

[0073] For example, in an embodiment of the present application, the initial aircraft detection model includes the image denoising network ADNet based on an improved lightweight attention mechanism, a bidirectional feature pyramid network BiFPN, and a detection head network; The aircraft detection model is trained based on the following method: Inputting the optical remote sensing image into the image denoising network ADNet to obtain the multi-level image features output by the image denoising network ADNet; wherein the multi-level image features include image features of the first level, image features of the second level, and image features of the third level; Inputting the image features of the first level, the image features of the second level, and the image features of the third level into the bidirectional feature pyramid network BiFPN, obtaining first scale features corresponding to the image features of the first level, second scale features corresponding to the image features of the second level, and third scale features corresponding to the image features of the third level output by the bidirectional feature pyramid network BiFPN; Inputting the first scale feature, the second scale feature and the third scale feature into the detection head network to obtain the aircraft category prediction probability and the aircraft detection frame output by the detection head network; Based on the classification loss function corresponding to the aircraft category prediction probability and the loss function corresponding to the aircraft detection box, the model parameters in the initial aircraft detection model are updated to obtain the aircraft detection model.

[0074] For example, in an embodiment of the present application, the image denoising network ADNet includes a sparse module SB, a feature enhancement module FEB, and an attention module based on an improved lightweight attention mechanism, which are sequentially connected in series; The step of inputting the optical remote sensing image into the image denoising network ADNet to obtain the multi-level image features output by the image denoising network ADNet includes: Inputting the optical remote sensing image into the sparse module SB to obtain the first-level image features output by the sparse module SB; Inputting the image features of the first level and the optical remote sensing image into the feature enhancement module FEB, and outputting the image features of the second level and the convolution image features through the feature enhancement module FEB; The image features of the second level and the convolutional image features are both input into the attention module based on the improved lightweight attention mechanism, and the image features of the third level are output through the attention module.

[0075] For example, in an embodiment of the present application, the feature enhancement module FEB includes a first convolution normalization unit, a second convolution normalization unit, a third convolution normalization unit, a convolution layer, and an activation function layer which are sequentially connected in series; The step of inputting the image features of the first level and the optical remote sensing image into the feature enhancement module FEB, and outputting the image features of the second level and the convolution image features through the feature enhancement module includes: Inputting the image feature input of the first level into the first convolution normalization unit, the second convolution normalization unit and the third convolution normalization unit in sequence, to obtain the convolution normalization feature output by the third convolution normalization unit; Inputting the convolution normalization feature into the convolution layer to obtain the convolution image feature output by the convolution layer; The sum of the optical remote sensing image and the convolution image features is input into the activation function layer to obtain the second-level image features output by the activation function layer.

[0076] For example, in an embodiment of the present application, the detection head network includes a plurality of detection heads of different scales connected in parallel, and the plurality of detection heads include a detection head corresponding to a first scale, a detection head corresponding to a second scale, and a detection head corresponding to a third scale; The first scale feature, the second scale feature and the third scale feature are all input into the detection head network to obtain the aircraft category prediction probability and the aircraft detection frame output by the detection head network, including: Inputting the first scale feature into a detection head corresponding to the first scale to obtain a corresponding first aircraft category prediction probability and a first aircraft detection frame; Inputting the second scale feature into the detection head corresponding to the second scale to obtain the corresponding second aircraft category prediction probability and second aircraft detection frame; Inputting the third scale feature into the detection head corresponding to the third scale to obtain a corresponding third aircraft category prediction probability and a third aircraft detection frame; Determining the aircraft category prediction probability based on the first aircraft category prediction probability, the second aircraft category prediction probability, and the third aircraft category prediction probability; The aircraft detection frame is determined based on the first aircraft detection frame, the second aircraft detection frame, and the third aircraft detection frame.

[0077] For example, in the embodiment of the present application, the classification loss function corresponding to the predicted probability of the aircraft category can be shown in the following formula: in, represents the classification loss function corresponding to the predicted probability of the aircraft category, represents the number of the optical remote sensing images, Indicates Optical remote sensing images, Indicates Target aircraft categories in optical remote sensing images The predicted probability of Indicates the preset probability threshold.

[0078] The aircraft detection device 40 in the optical remote sensing image provided in the embodiment of the present application can execute the technical solution of the aircraft detection method in the optical remote sensing image in any of the above-mentioned embodiments. Its implementation principle and beneficial effects are similar to the implementation principle and beneficial effects of the aircraft detection method in the optical remote sensing image. Please refer to the implementation principle and beneficial effects of the aircraft detection method in the optical remote sensing image, and no further details will be given here.

[0079] Figure 5 A schematic diagram of the physical structure of an electronic device provided in an embodiment of the present application, such as Figure 5 As shown, the electronic device may include: a processor 510, a communication interface 520, a memory 530 and a communication bus 540, wherein the processor 510, the communication interface 520 and the memory 530 communicate with each other through the communication bus 540. The processor 510 may call the logic instructions in the memory 530 to execute the method for detecting an aircraft in an optical remote sensing image, the method comprising: obtaining an optical remote sensing image of a target to be detected; inputting the optical remote sensing image of the target into an aircraft detection model to obtain an aircraft detection result output by the aircraft detection model; wherein the aircraft detection model is obtained by extracting image features of different levels from the optical remote sensing images in the training set based on an image denoising network ADNet with an improved lightweight attention mechanism, and training the initial aircraft detection model based on the extracted multi-level image features and the corresponding annotation labels.

[0080] In addition, the logic instructions in the above-mentioned memory 530 can be implemented in the form of a software functional unit and can be stored in a computer-readable storage medium when it is sold or used as an independent product. Based on this understanding, the technical solution of the present application can be essentially or partly embodied in the form of a software product that contributes to the prior art. The computer software product is stored in a storage medium, including several instructions to enable a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present application. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk, etc. Various media that can store program codes.

[0081] On the other hand, the present application also provides a computer program product, which includes a computer program, and the computer program can be stored on a non-transitory computer-readable storage medium. When the computer program is executed by a processor, the computer can execute the aircraft detection method in the optical remote sensing image provided by the above-mentioned methods, and the method includes: obtaining an optical remote sensing image of the target to be detected; inputting the target optical remote sensing image into an aircraft detection model to obtain an aircraft detection result output by the aircraft detection model; wherein the aircraft detection model is based on an improved lightweight attention mechanism image denoising network ADNet that extracts image features of different levels from the optical remote sensing images in the training set, and trains the initial aircraft detection model based on the extracted multi-level image features and corresponding annotation labels.

[0082] On the other hand, the present application also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, is implemented to execute the aircraft detection method in the optical remote sensing image provided by the above-mentioned methods, the method comprising: obtaining an optical remote sensing image of a target to be detected; inputting the target optical remote sensing image into an aircraft detection model to obtain an aircraft detection result output by the aircraft detection model; wherein the aircraft detection model is an image denoising network ADNet based on an improved lightweight attention mechanism, which extracts image features of different levels from the optical remote sensing images in a training set, and trains an initial aircraft detection model based on the extracted multi-level image features and corresponding annotation labels.

[0083] The device embodiments described above are merely illustrative, wherein the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment. Ordinary technicians in this field can understand and implement it without paying creative labor.

[0084] Through the description of the above implementation methods, those skilled in the art can clearly understand that each implementation method can be implemented by means of software plus a necessary general hardware platform, and of course, can also be implemented by hardware. Based on this understanding, the above technical solution is essentially or the part that contributes to the prior art can be embodied in the form of a software product, and the computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, a disk, an optical disk, etc., including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each embodiment or some parts of the embodiments.

[0085] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit it. Although the present application has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A method for detecting aircraft in optical remote sensing images, characterized in that: Acquire an optical remote sensing image of a target to be detected; Inputting the target optical remote sensing image into an aircraft detection model to obtain an aircraft detection result output by the aircraft detection model; Among them, the aircraft detection model is obtained by extracting image features of different levels from the optical remote sensing images in the training set based on the improved lightweight attention mechanism image denoising network ADNet, and training the initial aircraft detection model based on the extracted multi-level image features and corresponding annotation labels.

2. The method for detecting aircraft in optical remote sensing images according to claim 1, characterized in that: The initial aircraft detection model includes the image denoising network ADNet based on the improved lightweight attention mechanism, the bidirectional feature pyramid network BiFPN and the detection head network; The aircraft detection model is trained based on the following method: Inputting the optical remote sensing image into the image denoising network ADNet to obtain the multi-level image features output by the image denoising network ADNet; wherein the multi-level image features include image features of the first level, image features of the second level, and image features of the third level; Inputting the image features of the first level, the image features of the second level, and the image features of the third level into the bidirectional feature pyramid network BiFPN, obtaining first scale features corresponding to the image features of the first level, second scale features corresponding to the image features of the second level, and third scale features corresponding to the image features of the third level output by the bidirectional feature pyramid network BiFPN; Inputting the first scale feature, the second scale feature and the third scale feature into the detection head network to obtain the aircraft category prediction probability and the aircraft detection frame output by the detection head network; Based on the classification loss function corresponding to the aircraft category prediction probability and the loss function corresponding to the aircraft detection box, the model parameters in the initial aircraft detection model are updated to obtain the aircraft detection model.

3. The method for detecting aircraft in optical remote sensing images according to claim 2, characterized in that: The image denoising network ADNet includes a sparse module SB, a feature enhancement module FEB and an attention module based on an improved lightweight attention mechanism which are connected in series in sequence; The step of inputting the optical remote sensing image into the image denoising network ADNet to obtain the multi-level image features output by the image denoising network ADNet includes: Inputting the optical remote sensing image into the sparse module SB to obtain the first-level image features output by the sparse module SB; Inputting the image features of the first level and the optical remote sensing image into the feature enhancement module FEB, and outputting the image features of the second level and the convolution image features through the feature enhancement module FEB; The image features of the second level and the convolutional image features are both input into the attention module based on the improved lightweight attention mechanism, and the image features of the third level are output through the attention module.

4. The method for detecting aircraft in optical remote sensing images according to claim 3, characterized in that: The feature enhancement module FEB includes a first convolution normalization unit, a second convolution normalization unit, a third convolution normalization unit, a convolution layer and an activation function layer which are sequentially connected in series; The step of inputting the image features of the first level and the optical remote sensing image into the feature enhancement module FEB, and outputting the image features of the second level and the convolution image features through the feature enhancement module includes: Inputting the image feature input of the first level into the first convolution normalization unit, the second convolution normalization unit and the third convolution normalization unit in sequence, to obtain the convolution normalization feature output by the third convolution normalization unit; Inputting the convolution normalization feature into the convolution layer to obtain the convolution image feature output by the convolution layer; The sum of the optical remote sensing image and the convolution image features is input into the activation function layer to obtain the second-level image features output by the activation function layer.

5. The method for detecting aircraft in optical remote sensing images according to claim 2, characterized in that: The detection head network includes a plurality of detection heads of different scales connected in parallel, wherein the plurality of detection heads include a detection head corresponding to a first scale, a detection head corresponding to a second scale, and a detection head corresponding to a third scale; The first scale feature, the second scale feature and the third scale feature are all input into the detection head network to obtain the aircraft category prediction probability and the aircraft detection frame output by the detection head network, including: Inputting the first scale feature into a detection head corresponding to the first scale to obtain a corresponding first aircraft category prediction probability and a first aircraft detection frame; Inputting the second scale feature into the detection head corresponding to the second scale to obtain the corresponding second aircraft category prediction probability and second aircraft detection frame; Inputting the third scale feature into the detection head corresponding to the third scale to obtain a corresponding third aircraft category prediction probability and a third aircraft detection frame; Determining the aircraft category prediction probability based on the first aircraft category prediction probability, the second aircraft category prediction probability, and the third aircraft category prediction probability; The aircraft detection frame is determined based on the first aircraft detection frame, the second aircraft detection frame, and the third aircraft detection frame.

6. The method for detecting aircraft in optical remote sensing images according to any one of claims 2 to 5, characterized in that: The classification loss function corresponding to the predicted probability of the aircraft category can be shown in the following formula: in, represents the classification loss function corresponding to the predicted probability of the aircraft category, represents the number of the optical remote sensing images, Indicates Optical remote sensing images, Indicates Target aircraft categories in optical remote sensing images The predicted probability of Indicates the preset probability threshold.

7. An aircraft detection device in an optical remote sensing image, characterized in that: include: An acquisition unit, used for acquiring an optical remote sensing image of a target to be detected; A detection unit, used for inputting the target optical remote sensing image into an aircraft detection model to obtain an aircraft detection result output by the aircraft detection model; Among them, the aircraft detection model is obtained by extracting image features of different levels from the optical remote sensing images in the training set based on the improved lightweight attention mechanism image denoising network ADNet, and training the initial aircraft detection model based on the extracted multi-level image features and corresponding annotation labels.

8. An electronic device comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that: When the processor executes the computer program, the method for detecting aircraft in an optical remote sensing image according to any one of claims 1 to 6 is implemented.

9. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the method for detecting aircraft in an optical remote sensing image according to any one of claims 1 to 6 is implemented.

10. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the method for detecting aircraft in an optical remote sensing image according to any one of claims 1 to 6 is implemented.

Images