Image-based unmanned aerial vehicle terrain category detection method
By combining a single zoom camera with YOLOX-s and GhostNetV2 classification models, the accuracy problem of UAV terrain detection in harsh scenarios and complex road conditions is solved, achieving efficient and low-cost recognition of diverse terrain features.
Patent Information
- Application Number
- CN202511479815.7
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-10-16
- Publication Date
- 2026-02-24
AI Technical Summary
Existing UAV terrain detection technology has limited accuracy in harsh environments and lacks robustness in road detection under complex road conditions, making it difficult to meet diverse terrain detection needs.
A single zoom camera is used to acquire images. The terrain detection model based on the YOLOX-s algorithm and the classification model based on GhostNetV2 as the backbone network are combined. The feature fusion is enhanced by the DFC attention self-attention module to achieve terrain classification.
It improves the applicability and accuracy of terrain detection, reduces costs, and achieves efficient and accurate recognition of diverse terrain features.
Smart Images
Figure CN121564577A_ABST
Abstract
Description
Technical Field
[0001] This disclosure relates to the field of unmanned aerial vehicle (UAV) terrain category detection, and more specifically, to an image-based UAV terrain category detection method, apparatus, electronic device, and computer-readable storage medium. Background Technology
[0002] Drones, through terrain detection technology, can acquire high-resolution terrain data containing detailed information such as surface topography, vegetation cover, rivers, and lakes. Simultaneously, drones can optimize their navigation and positioning capabilities by accurately identifying terrain features. Especially in complex terrain conditions such as mountains, forests, and deserts, terrain detection ensures the safe and stable flight of drones, accurately reaching predetermined targets. Drone terrain detection avoids human intrusion into target areas, reducing potential safety risks during the measurement process. Particularly in dangerous or inaccessible areas, drone terrain detection ensures personnel safety while simultaneously collecting terrain data. Compared to traditional manual or ground-based surveying, drones can quickly cover large areas, acquire massive amounts of data in real time, and reduce the investment of manpower, resources, and capital. Therefore, drone terrain detection is more efficient and cost-effective.
[0003] In existing technologies, a UAV terrain perception and warning method includes: acquiring DEM data and preprocessing it to generate multi-level pyramid data; extracting buildings, towers, roads, and bridges from the vector data in OSM data to generate a 3D model; acquiring UAV onboard data using a flight control computer module; acquiring sensor data using a sensor module; generating an initial terrain layered warning map using a terrain warning module; combining sensor data and elevation data to generate collision information and a surface elevation map; optimizing using Bayesian nonparametric methods; generating a UAV collision response based on the collision information; and issuing a collision warning signal; and displaying the data using a visual display module. This invention generates an initial terrain layered warning map according to aviation standards using elevation data, OSM data, and onboard data, and optimizes it using sensor data and Bayesian nonparametric methods, eliminating elevation differences between surface buildings and achieving higher warning accuracy. However, DEM data and OSM data are difficult to obtain in special scenarios, such as the Gobi Desert and plateaus, and their accuracy is limited. Our solution relies solely on camera image data, making it applicable to a wider range of terrain detection scenarios and achieving higher accuracy.
[0004] Another existing technology involves a ground-based visual road detection method for unmanned vehicles (UAVs) based on UAV images. This invention discloses a method for ground-based visual road detection based on UAV images. First, two road detection models are trained for road region identification in aerial and vehicle-mounted images, respectively. Then, a color particle tracking algorithm, a direction matching method based on line segments, and inverse projection transformation are used to obtain common interest region (COP) images from the aerial and vehicle-mounted images, respectively. Road detection is then performed on the two COP images separately, and the recognition results are weighted and averaged to obtain a fused top-view road detection result. This invention uses the UAV to expand the field of view, increasing the perception range of the unmanned vehicle, and combines top-view information for road detection, effectively improving the accuracy of distant road detection and enhancing the robustness of road detection. The above scheme combines the UAV's field of view and perception to perform accurate terrain detection, but requires the integration of vehicle-mounted images for identification. However, it is limited in complex road conditions or scenarios where vehicles cannot drive.
[0005] Therefore, one or more methods are needed to solve the above problems.
[0006] It should be noted that the information disclosed in the background section above is only used to enhance the understanding of the background of this disclosure, and therefore may include information that does not constitute prior art known to those skilled in the art. Summary of the Invention
[0007] The purpose of this disclosure is to provide an image-based method, apparatus, electronic device, and computer-readable storage medium for unmanned aerial vehicle (UAV) terrain category detection, thereby overcoming, at least to some extent, one or more problems caused by limitations and defects in related technologies.
[0008] According to one aspect of this disclosure, an image-based method for detecting unmanned aerial vehicle (UAV) terrain categories is provided, comprising:
[0009] The original planar image of the area to be identified is acquired using a preset image acquisition device;
[0010] The original planar image is detected based on a preset detection model, and detection results are generated.
[0011] If the detection result is that the terrain is not identified, the detection result is sent to the preset classification model;
[0012] The detection results of unidentified terrain are classified based on a preset classification model to generate terrain classification results.
[0013] In one exemplary embodiment of this disclosure, the method further includes:
[0014] The preset image acquisition device is a single variable-focus camera.
[0015] In one exemplary embodiment of this disclosure, the method further includes:
[0016] The preset detection model is a terrain detection model based on the YOLOX-s algorithm.
[0017] In one exemplary embodiment of this disclosure, the method further includes:
[0018] The original planar image is detected based on a preset detection model to generate a detection result. If the detection result is identifiable terrain, the identifiable terrain is output as a terrain classification result.
[0019] If the detection result indicates that the terrain needs to be further refined, then the detection result is sent to the preset classification model.
[0020] In one exemplary embodiment of this disclosure, the method further includes:
[0021] The preset classification model is a classification model with GhostNetV2 as the backbone network. The bottleneck part of the preset classification model is supplemented with a decoupled fully connected DFC attention self-attention module.
[0022] In one exemplary embodiment of this disclosure, the method further includes:
[0023] The input features are fed into parallel Ghost Module branches and DFC module branches, respectively. The Ghost Module branch is used to output features, and the DFC module branch is used to generate attention maps.
[0024] In one exemplary embodiment of this disclosure, the preset classification model in the method further includes:
[0025] Each attention value in the preset classification model is associated with a large range of image patches, so that the output features can cover the global information of the image patches;
[0026] Feature fusion is performed through dot product operations to enhance extended features;
[0027] The enhanced features are fed into another Ghost module to generate output features that include long-range spatial dependencies.
[0028] In one aspect of this disclosure, an image-based unmanned aerial vehicle (UAV) terrain category detection device is provided, comprising:
[0029] The image acquisition module is used to acquire the original planar image of the area to be identified based on a preset image acquisition device;
[0030] The image detection module is used to detect the original planar image based on a preset detection model and generate detection results;
[0031] The result recognition module is used to send the detection result to a preset classification model if the detection result is unrecognized terrain.
[0032] The result classification module is used to classify the detection results of unidentified terrain based on a preset classification model and generate terrain classification results.
[0033] In one aspect of this disclosure, an electronic device is provided, comprising:
[0034] Processor; and
[0035] A memory storing computer-readable instructions that, when executed by the processor, implement the method according to any one of the preceding claims.
[0036] In one aspect of this disclosure, a computer-readable storage medium is provided having a computer program stored thereon that, when executed by a processor, implements the method according to any one of the preceding claims.
[0037] An exemplary embodiment of this disclosure provides an image-based drone terrain category detection method. The method includes: acquiring an original planar image of a region to be identified using a preset image acquisition device; detecting the original planar image based on a preset detection model to generate a detection result; if the detection result indicates unidentified terrain, sending the detection result to a preset classification model; and classifying the detection result of unidentified terrain based on the preset classification model to generate a terrain classification result. This disclosure requires only a single variable-focus camera, resulting in a simple operating environment, wider applicability, easier installation and deployment, lower cost, and higher accuracy. Furthermore, the terrain classification method employed in this disclosure combines the advantages of diversity and high accuracy, enabling efficient and accurate differentiation of terrain features when integrated into the entire recognition system.
[0038] It should be understood that the above general description and the following detailed description are exemplary and explanatory only, and are not intended to limit this disclosure. Attached Figure Description
[0039] The above and other features and advantages of this disclosure will become more apparent from the detailed description of exemplary embodiments thereof with reference to the accompanying drawings.
[0040] Figure 1 A flowchart of an image-based unmanned aerial vehicle (UAV) terrain category detection method according to an exemplary embodiment of the present disclosure is shown;
[0041] Figure 2A logic diagram of an image-based unmanned aerial vehicle (UAV) terrain category detection method according to an exemplary embodiment of the present disclosure is shown;
[0042] Figure 3 A schematic diagram of the bottleneck structure of an image-based UAV terrain category detection method according to an exemplary embodiment of the present disclosure is shown.
[0043] Figure 4 A schematic block diagram of an image-based unmanned aerial vehicle (UAV) terrain category detection apparatus according to an exemplary embodiment of the present disclosure is shown.
[0044] Figure 5 A block diagram of an electronic device according to an exemplary embodiment of the present disclosure is shown schematically;
[0045] Figure 6 The illustration shows a schematic diagram of a computer-readable storage medium according to an exemplary embodiment of the present disclosure. Detailed Implementation
[0046] Exemplary embodiments will now be described more fully with reference to the accompanying drawings. However, these exemplary embodiments can be implemented in many forms and should not be construed as limited to the embodiments set forth herein; rather, they are provided so that this disclosure will be thorough and complete, and will fully convey the concept of the exemplary embodiments to those skilled in the art. The same reference numerals in the drawings denote the same or similar parts, and therefore repeated descriptions of them will be omitted.
[0047] Furthermore, the described features, structures, or characteristics can be combined in any suitable manner in one or more embodiments. Numerous specific details are provided in the following description to give a thorough understanding of embodiments of this disclosure. However, those skilled in the art will recognize that the technical solutions of this disclosure can be practiced without one or more of the specific details described, or other methods, components, materials, apparatuses, steps, etc., can be employed. In other instances, well-known structures, methods, apparatuses, implementations, materials, or operations are not shown or described in detail to avoid obscuring various aspects of this disclosure.
[0048] The block diagrams shown in the accompanying drawings are merely functional entities and do not necessarily correspond to physically independent entities. That is, these functional entities can be implemented in software, or in one or more software-hardened modules, or in different network and / or processor devices and / or microcontroller devices.
[0049] In this example embodiment, an image-based method for UAV terrain category detection is first provided; refer to Figure 1 As shown, this image-based drone terrain category detection method may include the following steps:
[0050] Step S110: Acquire the original planar image of the area to be identified using a preset image acquisition device;
[0051] Step S120: Detect the original planar image based on a preset detection model and generate detection results;
[0052] Step S130: If the detection result is unidentified terrain, then the detection result is sent to the preset classification model;
[0053] Step S140: Classify the detection results of unidentified terrain based on a preset classification model to generate terrain classification results.
[0054] An exemplary embodiment of this disclosure provides an image-based drone terrain category detection method. The method includes: acquiring an original planar image of a region to be identified using a preset image acquisition device; detecting the original planar image based on a preset detection model to generate a detection result; if the detection result indicates unidentified terrain, sending the detection result to a preset classification model; and classifying the detection result of unidentified terrain based on the preset classification model to generate a terrain classification result. This disclosure requires only a single variable-focus camera, resulting in a simple operating environment, wider applicability, easier installation and deployment, lower cost, and higher accuracy. Furthermore, the terrain classification method employed in this disclosure combines the advantages of diversity and high accuracy, enabling efficient and accurate differentiation of terrain features when integrated into the entire recognition system.
[0055] The following will further explain an image-based UAV terrain category detection method in this example embodiment.
[0056] Example 1:
[0057] In step S110, the original planar image of the area to be identified can be acquired based on a preset image acquisition device.
[0058] In this example embodiment, the method further includes:
[0059] The preset image acquisition device is a single variable-focus camera.
[0060] In this example embodiment, the method further includes:
[0061] The preset detection model is a terrain detection model based on the YOLOX-s algorithm.
[0062] In step S120, the original planar image can be detected based on a preset detection model to generate detection results.
[0063] In this example embodiment, the method further includes:
[0064] The original planar image is detected based on a preset detection model to generate a detection result. If the detection result is identifiable terrain, the identifiable terrain is output as a terrain classification result.
[0065] If the detection result indicates that the terrain needs to be further refined, then the detection result is sent to the preset classification model.
[0066] In step S130, if the detection result is unidentified terrain, the detection result can be sent to a preset classification model.
[0067] In this example embodiment, the method further includes:
[0068] The preset classification model is a classification model with GhostNetV2 as the backbone network. The bottleneck part of the preset classification model is supplemented with a decoupled fully connected DFC attention self-attention module.
[0069] In this example embodiment, the method further includes:
[0070] The input features are fed into parallel Ghost Module branches and DFC module branches, respectively. The Ghost Module branch is used to output features, and the DFC module branch is used to generate attention maps.
[0071] In step S140, the detection results of unidentified terrain can be classified based on a preset classification model to generate terrain classification results.
[0072] In this example embodiment, the preset classification model in the method further includes:
[0073] Each attention value in the preset classification model is associated with a large range of image patches, so that the output features can cover the global information of the image patches;
[0074] Feature fusion is performed through dot product operations to enhance extended features;
[0075] The enhanced features are fed into another Ghost module to generate output features that include long-range spatial dependencies.
[0076] In this example embodiment, the present disclosure only requires a single zoom camera, which is simple to operate in, applicable to a wider range of scenarios, easier to install and deploy, low in cost, and high in accuracy. At the same time, the present disclosure adopts a terrain classification method, which takes into account both the advantages of diversity and high accuracy. When combined with the entire recognition system, it can efficiently and accurately distinguish terrain features.
[0077] Example 2:
[0078] In the embodiments of this example, as Figure 2 The diagram illustrates the logic of the detection method. Both the detection and classification models are 2D object detection and classification algorithms. Detection model 1 uses the YOLOX-s algorithm, directly inputting the raw 2D image captured by the camera. The model then infers and detects the terrain type within the image. Common terrain types are categorized into three main types (convex, concave, and cliff-like). Prominent terrain features such as steep slopes and protrusions are classified as convex terrain; concave terrain features such as trenches and pits are classified as concave terrain; and terrain features with high lateral verticality, such as cliffs and precipices, are classified as cliff-like terrain. Once the terrain type is determined, it is input into the classification model for further evaluation.
[0079] The images detected in the previous step are then classified. Due to the high real-time requirements of drone detection, this disclosure employs a classification model with GhostNetV2 as the backbone network. Compared to GhostNetV1, its bottleneck part adds a DFC attention module. The specific bottleneck structure is as follows... Figure 3 As shown.
[0080] In the embodiments of this example, as Figure 3 As shown, DFC attention is used to enhance the output feature Y of the Ghost Module, thereby capturing long-range dependencies between pixels in different spatial locations. The input feature X is fed into two branches: the Ghost Module branch for output feature Y, and the DFC module branch for generating the attention map. Using the same input features, the Ghost Module and DFC attention are two parallel branches extracting information from different perspectives. The output features are their element-wise information, containing features from the Ghost Module and information from the DFC attention. Each attention value involves a wide range of patches so that the output features can contain information from these patches. Dot product merging is used to enhance the expanded features. The enhanced features are then fed into a second Ghost module to produce the output features. It captures long-range dependencies between pixels in different spatial locations, enhancing the model's expressive power. The final result is obtained.
[0081] In this example embodiment, the detection algorithm throughout the system can be deployed using the OpenVino tool, resulting in fast inference speed. This significantly improves the system's versatility and allows it to be deployed and run on processors.
[0082] It should be noted that although the steps of the method in this disclosure are described in a specific order in the accompanying drawings, this does not require or imply that the steps must be performed in that specific order, or that all the steps shown must be performed to achieve the desired result. Additional or alternative steps may be omitted, multiple steps may be combined into one step, and / or a step may be broken down into multiple steps.
[0083] Furthermore, in this example embodiment, an image-based drone terrain category detection device is also provided. (See reference...) Figure 4 As shown, the image-based UAV terrain category detection device 200 may include: an image acquisition module 210, an image detection module 220, a result recognition module 230, and a result classification module 240. Wherein:
[0084] Image acquisition module 210 is used to acquire the original planar image of the area to be identified based on a preset image acquisition device;
[0085] Image detection module 220 is used to detect the original planar image based on a preset detection model and generate detection results;
[0086] The result recognition module 230 is used to send the detection result to a preset classification model if the detection result is unrecognized terrain.
[0087] The result classification module 240 is used to classify the detection results of unidentified terrain based on a preset classification model and generate terrain classification results.
[0088] The specific details of each of the above-mentioned image-based UAV terrain category detection device modules have been described in detail in the corresponding image-based UAV terrain category detection method, so they will not be repeated here.
[0089] It should be noted that although several modules or units of an image-based UAV terrain category detection device 200 have been mentioned in the detailed description above, this division is not mandatory. In fact, according to embodiments of this disclosure, the features and functions of two or more modules or units described above can be embodied in one module or unit. Conversely, the features and functions of one module or unit described above can be further divided and embodied by multiple modules or units.
[0090] Furthermore, in an exemplary embodiment of this disclosure, an electronic device capable of implementing the above-described method is also provided.
[0091] Those skilled in the art will understand that various aspects of the present invention can be implemented as systems, methods, or program products. Therefore, various aspects of the present invention can be specifically implemented as entirely hardware embodiments, entirely software embodiments (including firmware, microcode, etc.), or embodiments combining hardware and software aspects, collectively referred to herein as “circuit,” “module,” or “system.”
[0092] The following reference Figure 5 To describe an electronic device 500 according to such an embodiment of the present invention. Figure 5 The electronic device 500 shown is merely an example and should not impose any limitations on the functionality and scope of use of the embodiments of the present invention.
[0093] like Figure 5 As shown, the electronic device 500 is manifested in the form of a general-purpose computing device. The components of the electronic device 500 may include, but are not limited to: at least one processing unit 510, at least one storage unit 520, a bus 530 connecting different system components (including storage unit 520 and processing unit 510), and a display unit 540.
[0094] The storage unit stores program code that can be executed by the processing unit 510, causing the processing unit 510 to perform the steps described in the "Exemplary Methods" section of this specification according to various exemplary embodiments of the present invention. For example, the processing unit 510 can perform actions such as... Figure 1 Steps S110 to S140 are shown in the diagram.
[0095] Storage unit 520 may include a readable medium in the form of a volatile storage unit, such as random access memory (RAM) 5201 and / or cache memory 5202, and may further include a read-only memory (ROM) 5203.
[0096] Storage unit 520 may also include a program / utility 5204 having a set (at least one) program module 5205, such program module 5205 including but not limited to: operating system, one or more application programs, other program modules and program data, each or some combination of these examples may include an implementation of a network environment.
[0097] Bus 530 can represent one or more of several types of bus structures, including a memory cell bus or memory cell controller, a peripheral bus, a graphics acceleration port, a processing unit, or a local bus using any of the various bus structures.
[0098] Electronic device 500 can also communicate with one or more external devices 570 (e.g., keyboard, pointing device, Bluetooth device, etc.), and with one or more devices that enable a user to interact with electronic device 500, and / or with any device that enables electronic device 500 to communicate with one or more other computing devices (e.g., router, modem, etc.). This communication can be performed via input / output (I / O) interface 550. Furthermore, electronic device 500 can also communicate with one or more networks (e.g., local area network (LAN), wide area network (WAN), and / or public networks, such as the Internet) via network adapter 560. As shown, network adapter 560 communicates with other modules of electronic device 500 via bus 530. It should be understood that, although not shown in the figures, other hardware and / or software modules can be used in conjunction with electronic device 500, including but not limited to: microcode, device drivers, redundant processing units, external disk drive arrays, RAID systems, tape drives, and data backup storage systems.
[0099] From the above description of the embodiments, those skilled in the art will readily understand that the exemplary embodiments described herein can be implemented by software or by combining software with necessary hardware. Therefore, the technical solutions according to the embodiments of this disclosure can be embodied in the form of a software product, which can be stored in a non-volatile storage medium (such as a CD-ROM, USB flash drive, external hard drive, etc.) or on a network, including several instructions to cause a computing device (such as a personal computer, server, terminal device, or network device, etc.) to execute the methods according to the embodiments of this disclosure.
[0100] In exemplary embodiments of this disclosure, a computer-readable storage medium is also provided, on which a program product capable of implementing the methods described above is stored. In some possible embodiments, various aspects of the invention may also be implemented as a program product comprising program code that, when the program product is run on a terminal device, causes the terminal device to perform the steps of the various exemplary embodiments of the invention described in the "Exemplary Methods" section above.
[0101] refer to Figure 6 As shown, a program product 600 for implementing the above-described method according to an embodiment of the present invention is described. It may employ a portable compact disc read-only memory (CD-ROM) and include program code, and may run on a terminal device, such as a personal computer. However, the program product of the present invention is not limited thereto. In this document, the readable storage medium may be any tangible medium containing or storing a program that may be used by or in conjunction with an instruction execution system, apparatus, or device.
[0102] The program product may employ any combination of one or more readable media. A readable medium may be a readable signal medium or a readable storage medium. A readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples of readable storage media (a non-exhaustive list) include: an electrical connection having one or more wires, a portable disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof.
[0103] Computer-readable signal media may include data signals propagated in baseband or as part of a carrier wave, carrying readable program code. Such propagated data signals may take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. A readable signal medium may also be any readable medium other than a readable storage medium, capable of sending, propagating, or transmitting programs for use by or in conjunction with an instruction execution system, apparatus, or device.
[0104] The program code contained on the readable medium may be transmitted using any suitable medium, including but not limited to wireless, wired, optical fiber, RF, etc., or any suitable combination thereof.
[0105] Program code for performing the operations of this invention can be written in any combination of one or more programming languages, including object-oriented programming languages such as Java and C++, and conventional procedural programming languages such as C or similar languages. The program code can execute entirely on the user's computing device, partially on the user's device, as a standalone software package, partially on the user's computing device and partially on a remote computing device, or entirely on a remote computing device or server. In cases involving remote computing devices, the remote computing device can be connected to the user's computing device via any type of network, including a local area network (LAN) or a wide area network (WAN), or it can be connected to an external computing device (e.g., via the Internet using an Internet service provider).
[0106] Furthermore, the above figures are merely illustrative of the processes included in the method according to exemplary embodiments of the present invention, and are not intended to be limiting. It is readily understood that the processes shown in the above figures do not indicate or limit the temporal order of these processes. Additionally, it is readily understood that these processes may be executed synchronously or asynchronously, for example, in multiple modules.
[0107] Other embodiments of this disclosure will readily occur to those skilled in the art upon consideration of the specification and practice of the invention disclosed herein. This application is intended to cover any variations, uses, or adaptations of this disclosure that follow the general principles of this disclosure and include common knowledge or customary techniques in the art not disclosed herein. The specification and embodiments are to be considered exemplary only, and the true scope and spirit of this disclosure are indicated by the claims.
[0108] It should be understood that this disclosure is not limited to the precise structures described above and shown in the accompanying drawings, and various modifications and changes can be made without departing from its scope. The scope of this disclosure is limited only by the appended claims.
Claims
1. An image-based method for UAV terrain category detection, characterized in that, The method includes: The original planar image of the area to be identified is acquired using a preset image acquisition device; The original planar image is detected based on a preset detection model, and detection results are generated. If the detection result is that the terrain is not identified, the detection result is sent to the preset classification model; The detection results of unidentified terrain are classified based on a preset classification model to generate terrain classification results.
2. The method as described in claim 1, characterized in that, The method further includes: The preset image acquisition device is a single variable-focus camera.
3. The method as described in claim 1, characterized in that, The method further includes: The preset detection model is a terrain detection model based on the YOLOX-s algorithm.
4. The method as described in claim 3, characterized in that, The method further includes: The original planar image is detected based on a preset detection model to generate a detection result. If the detection result is identifiable terrain, the identifiable terrain is output as a terrain classification result. If the detection result indicates that the terrain needs to be further refined, then the detection result is sent to the preset classification model.
5. The method as described in claim 1, characterized in that, The method further includes: The preset classification model is a classification model with GhostNetV2 as the backbone network. The bottleneck part of the preset classification model is supplemented with a decoupled fully connected DFC attention self-attention module.
6. The method as described in claim 5, characterized in that, The method further includes: The input features are fed into parallel Ghost Module branches and DFC module branches, respectively. The Ghost Module branches are used to output features, and the DFC module branches are used to generate attention maps.
7. The method as described in claim 6, characterized in that, The preset classification model in the method also includes: Each attention value in the preset classification model is associated with a large range of image patches, so that the output features can cover the global information of the image patches; Feature fusion is performed through dot product operations to enhance extended features; The enhanced features are fed into another Ghost module to generate output features that include long-range spatial dependencies.
8. An image-based UAV terrain category detection device, characterized in that, The device includes: The image acquisition module is used to acquire the original planar image of the area to be identified based on a preset image acquisition device; The image detection module is used to detect the original planar image based on a preset detection model and generate detection results; The result recognition module is used to send the detection result to a preset classification model if the detection result is unrecognized terrain. The result classification module is used to classify the detection results of unidentified terrain based on a preset classification model and generate terrain classification results.
9. An electronic device, characterized in that, include Processor; and A memory storing computer-readable instructions that, when executed by the processor, implement the method according to any one of claims 1 to 7.
10. A computer-readable storage medium, characterized in that, It stores a computer program that, when executed by a processor, implements the method according to any one of claims 1 to 7.