UAV tracking method, device, UAV and storage medium

Abstract

The present invention discloses a method, device, drone and storage medium for tracking a drone, the method comprising: obtaining a plurality of continuous image frames containing a target to be tracked, dividing the plurality of continuous image frames into a plurality of groups, each group including an inference frame and a plurality of moving frames; based on a target tracking model, identifying the target from the inference frames in the preceding plurality of groups, and determining the border of the region of interest of the target on the inference frames; determining the pixel points within the border of the region of interest from the following group based on a corner point detection model; based on a backtracking algorithm, re-determining the pixel points within the border of the region of interest from the moving frames in the following group; determining the motion vector of the pixel point based on an optical flow motion estimation model, and tracking the target using the motion vector. The above scheme uses a backtracking algorithm model to realize tracking from the border of the region of interest to the centroid, reduces the target tracking delay, and improves the tracking accuracy.

Description

Technical Field

[0001] The present invention relates to the technical field of unmanned aerial vehicles, and in particular to a method and device for tracking unmanned aerial vehicles, an unmanned aerial vehicle and a storage medium. Background Art

[0002] Target tracking is a basic computer vision task. It tracks the selected target frame by frame by using the target information determined in the first frame. It plays a vital role in unmanned control, industrial manufacturing, traffic monitoring, etc.

[0003] In recent years, drone target tracking has received extensive attention due to its strong flexibility and its use in a wide range of industrial fields, such as aerial photography, mapping, and post-disaster rescue. It has broad application prospects. Its advantages are that the camera can track the target based on visual feedback, and the tracking performance can be optimized by changing the tracking angle to track the direction and position of the target. However, in drone tracking, there are also some difficulties that are difficult to encounter in traditional tracking problems: the environment of the tracking object is changeable (such as sudden changes in illumination), background clutter and severe occlusion, and tracking delay. Due to the limited computing power of current embedded mobile platforms, how to design a drone tracker with fast tracking speed, high accuracy, and low latency is still a very challenging task. Summary of the invention

[0004] In view of the above problems, the present invention is proposed to provide a drone tracking method, device, drone and storage medium that overcome the above problems or at least partially solve the above problems.

[0005] According to one aspect of the present invention, a method for tracking a drone is provided, the method comprising:

[0006] Acquire a plurality of continuous image frames containing the target to be tracked, and divide the plurality of continuous image frames into a plurality of groups, each group including an inference frame and a plurality of moving frames;

[0007] Based on the target tracking model, identifying the target from the reasoning frames in the previous groups, and determining the bounding box of the region of interest of the target on the reasoning frame;

[0008] Determine the pixel points within the border of the region of interest from the subsequent group based on the corner detection model;

[0009] Based on a backtracking algorithm, re-determine the pixel points in the border of the region of interest from the moving frames in the subsequent group;

[0010] The motion vector of the pixel is determined based on an optical flow motion estimation model, and the target is tracked using the motion vector.

[0011] In some embodiments, the target tracking model includes a YOLO series model, and the backbone feature extraction network of the YOLO series model is Darknet53.

[0012] In some implementations, determining pixels within the border of the region of interest from the subsequent group based on a corner detection model includes:

[0013] The Harris corner detection model is used to determine the border of the pixel.

[0014] In some implementations, determining the motion vector of the pixel based on the optical flow motion estimation model includes:

[0015] Determining the centroid of the plurality of pixels;

[0016] A motion vector of the centroid is determined based on the optical flow motion estimation model.

[0017] In some embodiments, the optical flow motion estimation model includes a Lucas-Kanade estimation model.

[0018] In some implementations, tracking the target using the motion vector includes:

[0019] Determine the position of the centroid in the current image frame using an IoU alignment algorithm;

[0020] Based on the motion vector and the position of the centroid in the current image frame, the position of the target in the current image frame is determined.

[0021] In some embodiments, the method further comprises:

[0022] After the pixel association between groups fails, a preset number of image frames are saved, which can be extracted and identified as targets;

[0023] When the border of the region of interest cannot be identified from the new image frame in the group, the saved image frame is called and the border of the region of interest therein is identified, and the border of the region of interest is used for the next step of identification and tracking, thereby improving the confidence of target identification.

[0024] According to another aspect of the present invention, a drone tracking device is provided, the device comprising:

[0025] An acquisition module, adapted to acquire a plurality of continuous image frames containing a target to be tracked, and divide the plurality of continuous image frames into a plurality of groups, each group including an inference frame and a plurality of moving frames;

[0026] A recognition module, adapted to recognize a target from the reasoning frames in the previous groups based on a target tracking model, and determine a bounding box of an area of ​​interest of the target on the reasoning frame;

[0027] A determination module, adapted to determine the pixel points within the border of the region of interest from the subsequent group based on a corner detection model;

[0028] A backtracking module, adapted to re-determine the pixel points in the border of the region of interest from the running frames in the following group based on a backtracking algorithm;

[0029] The tracking module is adapted to determine the motion vector of the pixel point based on the optical flow motion estimation model, and track the target using the motion vector.

[0030] According to another aspect of the present invention, a drone is provided, comprising: a processor and a memory arranged to store computer executable instructions, wherein when the executable instructions are executed, the processor executes the drone tracking method according to any one of the above embodiments.

[0031] According to another aspect of the present invention, a computer-readable storage medium is provided, wherein the computer-readable storage medium stores one or more programs, and when the one or more programs are executed by a processor, they implement any of the above-mentioned drone tracking methods.

[0032] From the above, it can be seen that according to the drone tracking method disclosed in the present invention, the acquired image frames are first grouped, the key inference frames and the moving frames after the inference frames are identified, and the backtracking algorithm model is used to realize tracking from the border of the area of ​​interest to the center of mass, thereby reducing the target tracking delay and improving the tracking accuracy.

[0033] The above description is only an overview of the technical solution of the present invention. In order to more clearly understand the technical means of the present invention, it can be implemented according to the contents of the specification. In order to make the above and other purposes, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are listed below. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] Various other advantages and benefits will become apparent to those of ordinary skill in the art by reading the detailed description of the preferred embodiments below. The accompanying drawings are only for the purpose of illustrating the preferred embodiments and are not to be considered as limiting the present invention. Moreover, the same reference symbols are used throughout the accompanying drawings to represent the same components. In the accompanying drawings:

[0035] Figure 1 A schematic diagram of a flow chart of a method for tracking a drone according to an embodiment of the present invention is shown;

[0036] Figure 2 A schematic diagram of image frame grouping and target recognition in drone target tracking according to an embodiment of the present invention is shown;

[0037] Figure 3 A schematic diagram of the structure of a drone tracking device according to an embodiment of the present invention is shown;

[0038] Figure 4 A schematic structural diagram of a drone according to an embodiment of the present invention is shown. DETAILED DESCRIPTION

[0039] The exemplary embodiments of the present invention will be described in more detail below with reference to the accompanying drawings. Although the exemplary embodiments of the present invention are shown in the accompanying drawings, it should be understood that the present invention can be implemented in various forms and should not be limited by the embodiments set forth herein. On the contrary, these embodiments are provided in order to enable a more thorough understanding of the present invention and to enable the scope of the present invention to be fully communicated to those skilled in the art.

[0040] Figure 1 The following is a flow chart of a method for tracking a drone according to an embodiment of the present invention. The method can be implemented by a program in a flight control system of the drone. The method includes the following steps:

[0041] Step S110, obtaining a plurality of continuous image frames containing the target to be tracked, dividing the plurality of continuous image frames into a plurality of groups, and Figure 2 As shown, each group includes an inference frame I and several moving frames M;

[0042] Step S120, based on the target tracking model, identifying the target from the reasoning frames in the previous groups, and determining the bounding box of the region of interest of the target on the reasoning frame;

[0043] Step S130, determining the pixel points within the border of the region of interest from the subsequent group based on the corner detection model;

[0044] Step S140, based on a backtracking algorithm, re-determine the pixel points in the border of the region of interest from the moving frames in the subsequent group;

[0045] The retrospective tracking in step S140 is actually implemented based on motion interpolation of target detection in the video sequence, wherein the retrospective tracking aims to reduce the tracking delay by adjusting the past detection information using the recorded moving frames, wherein the motion recording and interpolation are necessary parts of the retrospective tracking. Figure 2 As shown, a visual description of the backtracking can be seen based on the frame sequence, with the drone image frames in the second row showing the bounding boxes and pixels derived from the backtracking algorithm.

[0046] Step S150, determining the motion vector of the pixel point based on the optical flow motion estimation model, and tracking the target using the motion vector.

[0047] According to the drone tracking method of this embodiment, the acquired image frames are first grouped, the key inference frames and the moving frames after the inference frames are identified, and the backtracking algorithm model is used to realize tracking from the border of the area of ​​interest to the center of mass, thereby reducing the target tracking delay and improving the tracking accuracy.

[0048] In some embodiments, the target tracking model includes a YOLO series model, and the backbone feature extraction network of the YOLO series model is Darknet53.

[0049] YOLO is the abbreviation of You Look once. It is a commonly used algorithm model in target detection algorithms. This algorithm model achieves a compromise between detection speed and detection accuracy.

[0050] The main feature extraction network used by the existing YOLO is Darknet53, which has two important features: One important feature of Darknet53 is the use of a residual network. The residual convolution in Darknet53 first performs a convolution with a kernel size of 3X3 and a step size of 2. This convolution will compress the width and height of the input feature layer, and a feature layer can be obtained at this time; then the feature layer is subjected to a 1X1 convolution and a 3X3 convolution, and this result is added to the feature layer to form a residual structure. Through continuous 1X1 convolution and 3X3 convolution and the superposition of residual edges, the network is greatly deepened.

[0051] The residual network is easy to optimize and can improve accuracy by increasing the depth. The internal residual block uses skip connections to alleviate the gradient vanishing problem caused by increasing the depth of the deep neural network. In addition, each convolution part of Darknet53 uses a unique DarknetConv2D structure, and l2 regularization is performed during each convolution, and BatchNormalization and LeakyReLU are performed after the convolution is completed.

[0052] In some embodiments, step S130 determines the pixel points within the border of the region of interest from the subsequent group based on the corner detection model, including:

[0053] The Harris corner detection model is used to determine the border of the pixel.

[0054] In some embodiments, determining the motion vector of the pixel point based on the optical flow motion estimation model in step S150 includes:

[0055] Determining the centroid of the plurality of pixels;

[0056] A motion vector of the centroid is determined based on the optical flow motion estimation model.

[0057] In some embodiments, the optical flow motion estimation model comprises a Lucas-Kanade estimation model.

[0058] Specifically, Lucas-Kanade optical flow motion estimation is the preferred technology of the embodiment of the present invention because of its speed and ease of use and can be implemented together with OpenCV. In addition, full frame, region of interest (ROI) and ROI to centroid tracking solutions can also be implemented through Lucas-Kanade optical flow motion estimation.

[0059] Optionally, the image may be subdivided into discrete macroblocks, and the size of the formed optical flow block is 8 × 8. In addition, for the recognition of pixel points, the Harris corner detection module is used to generate the ROI border of the pixel points.

[0060] In some embodiments, tracking the target using the motion vector in step S150 includes:

[0061] Determine the position of the centroid in the current image frame using an IoU alignment algorithm;

[0062] Based on the motion vector and the position of the centroid in the current image frame, the position of the target in the current image frame is determined.

[0063] IoU refers to the alignment overlap of the front and back bounding boxes, and is calculated by aligning the top left corner of the box, as shown in the following formula.

[0064]

[0065] Where IoU is the intersection of the joint calculations; (W a , H a ) and ( W b , H b ) are the scales of the two anchor boxes. In theory, the more similar the anchor box is to the bounding box, the smaller the metric value should be.

[0066] It should be pointed out that if Figure 2As shown, starting from the 6th frame, the object enters the tracking mode, where the centroid of the bounding box represented by the star moves forward through the single-point optical flow. The embodiment of the present invention realizes the transition and transformation from multi-point ROI-based tracking to single-point centroid tracking, that is, from ROI to centroid tracking, which has the advantage of further improving the running speed through the backtracking algorithm. In the tracking mode, once the next frame is read in, the bounding box is located on the actual target, and the object can be tracked using the centroid generated by its box position.

[0067] By this centroid tracking, the delay in recognizing past detection frames can be reduced. Specifically, in frame 7, the motion vector from frame 6 to frame 7 determined by the optical flow on the centroid is determined, and the border is displayed. At frame 8 [3N−1], the motion vector is determined from frame 6 to frame 8. In this way, when reaching frame 9 [3N], the motion vector linking or traveling from frame 6 to frame 9 has been determined.

[0068] In some embodiments, the method further comprises:

[0069] After the pixel association between groups fails, a preset number of image frames are saved, which can be extracted and identified as targets;

[0070] When the border of the region of interest cannot be identified from the new image frame in the group, the saved image frame is called and the border of the region of interest therein is identified, and the border of the region of interest is used for the next step of identification and tracking, thereby improving the confidence of target identification.

[0071] According to the embodiment, during target recognition, when the association from one pixel to the next pixel fails, a certain number of image frames are saved in the memory in an active state. The purpose of the method is to reduce the false detection of the target object not being detected in the I frame, but actually existing in the video sequence. When a target object may not be displayed in the image frame sequence due to occlusion, small object size, poor network training, etc., and exceeds the number of frames that remain active. However, since the tracker disclosed in the embodiment of the present invention generates a tracking chain for N image frames, each tracking frame can be linked to the next tracking chain through an IoU-based matching alignment algorithm. For indexing, a unique ID can be assigned to each matched tracking target, and unmatched objects will receive a new ID. At this time, due to the use of the above-mentioned saving function, the last detection result can be extracted from the previous tracking pixel, and the border recognized by the saved image frame can be reused when the target recognition model fails to detect the existence of the border, thereby improving the confidence of target recognition.

[0072] In summary, the above embodiments of the present invention can achieve the following technical effects:

[0073] First, the YOLO series and Darknet53 backbone extraction network and optical flow motion estimation are used to achieve target detection and motion interpolation;

[0074] Secondly, the backtracking algorithm model is used to realize the tracking from the ROI region of interest to the centroid, which reduces the computational overhead;

[0075] Finally, based on the tracking chain, IoU matching and save activity functions, the core capabilities of the tracker are enhanced, the tracking delay is reduced and the tracking accuracy is improved.

[0076] Combination Figure 3 As shown, the present invention also protects a drone tracking device 300, and the drone tracking device 300 includes:

[0077] An acquisition module 310 is adapted to acquire a plurality of continuous image frames including a target to be tracked, and divide the plurality of continuous image frames into a plurality of groups, each group including an inference frame and a plurality of moving frames;

[0078] The recognition module 320 is adapted to recognize the target from the reasoning frames in the previous groups based on the target tracking model, and determine the bounding box of the region of interest of the target on the reasoning frame;

[0079] A determination module 330, adapted to determine the pixel points within the border of the region of interest from the subsequent group based on a corner detection model;

[0080] A backtracking module 340, adapted to re-determine the pixel points in the border of the region of interest from the moving frames in the subsequent group based on a backtracking algorithm;

[0081] The tracking module 350 is adapted to determine the motion vector of the pixel point based on the optical flow motion estimation model, and track the target using the motion vector.

[0082] According to this embodiment, the acquired image frames are grouped, the key inference frames and the moving frames after the inference frames are identified, and the backtracking algorithm model is used to achieve tracking from the border of the region of interest to the center of mass, thereby reducing the target tracking delay and improving the tracking accuracy.

[0083] In some embodiments, the target tracking model includes a YOLO series model, and the backbone feature extraction network of the YOLO series model is Darknet53.

[0084] In some embodiments, the determination module 330 is specifically adapted to:

[0085] The Harris corner detection model is used to determine the border of the pixel.

[0086] In some embodiments, determining the motion vector of the pixel point based on the optical flow motion estimation model in the tracking module 350 includes:

[0087] Determining the centroid of the plurality of pixels;

[0088] A motion vector of the centroid is determined based on the optical flow motion estimation model.

[0089] In some embodiments, the optical flow motion estimation model comprises a Lucas-Kanade estimation model.

[0090] In some embodiments, tracking the target using the motion vector in the tracking module 350 includes:

[0091] Determine the position of the centroid in the current image frame using an IoU alignment algorithm;

[0092] Based on the motion vector and the position of the centroid in the current image frame, the position of the target in the current image frame is determined.

[0093] In some embodiments, the device is further adapted to:

[0094] After the pixel association between groups fails, a preset number of image frames are saved, which can be extracted and identified as targets;

[0095] When the border of the region of interest cannot be identified from the new image frame in the group, the saved image frame is called and the border of the region of interest therein is identified, and the border of the region of interest is used for the next step of identification and tracking, thereby improving the confidence of target identification.

[0096] It should be noted that the specific implementation of the above-mentioned device embodiments can be carried out with reference to the specific implementation of the above-mentioned corresponding method embodiments, which will not be repeated here.

[0097] It should be noted that:

[0098] The algorithm and display provided herein are not inherently related to any particular computer, virtual device or other equipment. Various general-purpose devices can also be used together with the teachings based on this. According to the above description, it is obvious to construct the structure required for this type of device. In addition, the present invention is not directed to any specific programming language either. It should be understood that various programming languages ​​can be utilized to realize the content of the present invention described herein, and the description of the above specific language is for the purpose of disclosing the best mode of the present invention.

[0099] In the description provided herein, a large number of specific details are described. However, it is understood that embodiments of the present invention can be practiced without these specific details. In some instances, well-known methods, structures and techniques are not shown in detail so as not to obscure the understanding of this description.

[0100] Similarly, it should be understood that in order to streamline the present invention and aid in understanding one or more of the various inventive aspects, in the above description of exemplary embodiments of the present invention, various features of the present invention are sometimes grouped together into a single embodiment, figure, or description thereof. However, this method of disclosure should not be interpreted as reflecting an intention that the claimed invention requires more features than are expressly recited in each claim.

[0101] Those skilled in the art will appreciate that the modules in the devices in the embodiments may be adaptively changed and arranged in one or more devices different from the embodiments. The modules or units or components in the embodiments may be combined into one module or unit or component, and further may be divided into a plurality of submodules or subunits or subcomponents. All features disclosed in this specification (including the accompanying claims, abstracts and drawings) and all processes or units of any method or device so disclosed may be combined in any combination, except that at least some of such features and / or processes or units are mutually exclusive. Unless otherwise expressly stated, each feature disclosed in this specification (including the accompanying claims, abstracts and drawings) may be replaced by an alternative feature that provides the same, equivalent or similar purpose.

[0102] Furthermore, those skilled in the art will appreciate that although some embodiments described herein include certain features included in other embodiments but not other features, the combination of features from different embodiments is meant to be within the scope of the present invention and to form different embodiments.

[0103] The various component embodiments of the present invention may be implemented in hardware, or in software modules running on one or more processors, or in a combination thereof. Those skilled in the art will appreciate that a microprocessor or digital signal processor (DSP) may be used in practice to implement some or all of the functions of some or all of the components of the drone tracking device according to an embodiment of the present invention. The present invention may also be implemented as a device or apparatus program (e.g., a computer program and a computer program product) for executing part or all of the methods described herein. Such a program for implementing the present invention may be stored on a computer-readable medium, or may be in the form of one or more signals. Such a signal may be downloaded from an Internet website, or provided on a carrier signal, or provided in any other form.

[0104] An embodiment of the present invention provides a non-volatile computer storage medium, wherein the computer storage medium stores at least one executable instruction, and the computer executable instruction can execute the above-mentioned drone tracking method in any of the above-mentioned method embodiments.

[0105] Figure 4 A schematic diagram of the structure of an embodiment of a drone of the present invention is shown, and the specific embodiment of the present invention does not limit the specific structure of the drone.

[0106] like Figure 4 As shown, the UAV (including the flight control system) may include: a processor (processor) 402 , a communications interface (Communications Interface) 404 , a memory (memory) 406 , and a communication bus 408 .

[0107] The processor 402, the communication interface 404, and the memory 406 communicate with each other via the communication bus 408. The communication interface 404 is used to communicate with other devices such as a client or other server network elements. The processor 402 is used to execute the program 410, which can specifically execute the relevant steps in the above-mentioned drone tracking method embodiment for the drone.

[0108] Specifically, the program 410 may include program codes, which include computer operation instructions.

[0109] The processor 402 may be a central processing unit (CPU), or an application-specific integrated circuit (ASIC), or one or more integrated circuits configured to implement the embodiments of the present invention. The one or more processors included in the drone may be processors of the same type, such as one or more CPUs; or processors of different types, such as one or more CPUs and one or more ASICs.

[0110] The memory 406 is used to store the program 410. The memory 406 may include a high-speed RAM memory, and may also include a non-volatile memory (non-volatile memory), such as at least one disk memory.

[0111] Program 410 can be specifically used to enable processor 402 to perform operations corresponding to the above-mentioned drone tracking method embodiment.

[0112] It should be noted that the above embodiments illustrate the present invention rather than limit it, and that those skilled in the art may devise alternative embodiments without departing from the scope of the appended claims. In the claims, any reference symbol between brackets shall not be construed as a limitation on the claims. The word "comprising" does not exclude the presence of elements or steps not listed in the claims. The word "one" or "an" preceding an element does not exclude the presence of a plurality of such elements. The present invention may be implemented by means of hardware comprising a number of different elements and by means of a suitably programmed computer. In a unit claim enumerating a number of devices, several of these devices may be embodied by the same hardware item. The use of the words first, second, and third, etc., does not indicate any order. These words may be interpreted as names.

Claims

1. A drone tracking method, characterized in that: The method comprises: Acquire a plurality of continuous image frames containing the target to be tracked, and divide the plurality of continuous image frames into a plurality of groups, each group including an inference frame and a plurality of moving frames; Based on the target tracking model, identifying the target from the reasoning frames in the previous groups, and determining the bounding box of the region of interest of the target on the reasoning frame; Determine the pixel points within the border of the region of interest from the subsequent group based on the corner detection model; Based on a backtracking algorithm of object detection motion interpolation in a video sequence, re-determine the pixel points in the border of the region of interest from the moving frames in the subsequent group; Determine the motion vector of the pixel based on an optical flow motion estimation model, and track the target using the motion vector; The method further comprises: After the pixel association between groups fails, a preset number of image frames are saved, which can be extracted and identified as targets; When the border of the region of interest cannot be identified from the new image frame in the group, the saved image frame is called and the border of the region of interest therein is identified, and the border of the region of interest is used for the next step of identification and tracking, thereby improving the confidence of target identification.

2. The method according to claim 1, characterized in that The target tracking model includes a YOLO series model, and the backbone feature extraction network of the YOLO series model is Darknet53.

3. The method according to claim 1, characterized in that Determining the pixel points within the border of the region of interest from the subsequent group based on the corner detection model includes: The Harris corner detection model is used to determine the border of the pixel.

4. The method according to claim 1, characterized in that: Determining the motion vector of the pixel point based on the optical flow motion estimation model includes: Determining the centroid of a plurality of said pixels; A motion vector of the centroid is determined based on the optical flow motion estimation model.

5. The method according to claim 4, characterized in that The optical flow motion estimation model includes a Lucas-Kanade estimation model.

6. The method according to claim 4 or 5, characterized in that: Tracking the target using the motion vector includes: Determine the position of the centroid in the current image frame using an IoU alignment algorithm; Based on the motion vector and the position of the centroid in the current image frame, the position of the target in the current image frame is determined.

7. A drone tracking device, characterized in that: The device comprises: An acquisition module, adapted to acquire a plurality of continuous image frames containing a target to be tracked, and divide the plurality of continuous image frames into a plurality of groups, each group including an inference frame and a plurality of moving frames; A recognition module, adapted to recognize a target from the reasoning frames in the previous groups based on a target tracking model, and determine a bounding box of an area of ​​interest of the target on the reasoning frame; A determination module, adapted to determine the pixel points within the border of the region of interest from the subsequent group based on a corner detection model; A tracing module adapted to re-determine the pixel points in the border of the region of interest from the running frames in the subsequent group based on a tracing algorithm of motion interpolation of object detection in a video sequence; A tracking module, adapted to determine the motion vector of the pixel point based on an optical flow motion estimation model, and track the target using the motion vector; The device is also suitable for: After the pixel association between groups fails, a preset number of image frames are saved, which can be extracted and identified as targets; When the border of the region of interest cannot be identified from the new image frame in the group, the saved image frame is called and the border of the region of interest therein is identified, and the border of the region of interest is used for the next step of identification and tracking, thereby improving the confidence of target identification.

8. A drone, characterized in that: The invention comprises a processor and a memory arranged to store computer executable instructions, wherein when the executable instructions are executed, the processor performs the drone tracking method according to any one of claims 1 to 6.

9. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores one or more programs, which, when executed by a processor, implement the drone tracking method according to any one of claims 1 to 6.

Images