A method, device, equipment and medium for determining motion trajectory

Abstract

The present application discloses a motion trajectory determination method, device, equipment and medium for determining the complete motion trajectory of an athlete. The present application inputs the obtained first overhead video frame and the second overhead video frame into a pre-trained position recognition model respectively, and obtains the area where the first athlete is located contained in the first overhead video frame and the corresponding first identification information, and the area where the second athlete is located contained in the second overhead video frame and the corresponding second identification information; and simultaneously inputs the first overhead video frame and the second overhead video frame into a pre-trained twin network model, and for the first athlete corresponding to the first identification information, determines the athlete in the second overhead video frame that is the same as the first athlete. Since the overhead video frame is largely unaffected by the mutual occlusion between athletes, the complete motion trajectory of the first athlete can be accurately obtained to a greater extent.

Description

Technical Field

[0001] The present application relates to the technical field of motion trajectory determination, and in particular to a motion trajectory determination method, apparatus, device, and medium. Background Art

[0002] With the continued advancement of the strategy to build a strong sports nation, sports have played a significant role in building a healthy China, promoting economic transformation and upgrading, and enhancing national cohesion and cultural competitiveness. Sports events are also attracting increasing attention from users, who are gradually shifting their viewing channels to digital media. For example, monthly active users for basketball have exceeded 30 million, and monthly active users for football have exceeded 20 million. Simultaneously, the analysis of sports video is gaining increasing attention. For example, coaches and athletes hope to use this analysis to improve team tactics and player skills, while broadcasters and spectators hope to enhance the viewing experience of sports events through video analysis.

[0003] Taking ball games as an example, when recording sports videos, they are usually recorded based on sideline cameras installed outside the stadium. During the recording, some players (athletes) will be given close-up shots during exciting moments of the game. In addition, athletes inevitably block each other, so it is difficult to ensure that the game video can contain the complete movement trajectory of the athletes on the field (sports field). However, it is very important to analyze the complete movement trajectory of athletes on the sports field (for convenience of description, referred to as the complete movement trajectory of athletes) to improve the overall tactics of the team, the skills of the athletes, and the viewing experience of the game. Therefore, there is an urgent need for a technical solution that can determine the complete movement trajectory of athletes. Summary of the Invention

[0004] The present application provides a motion trajectory determination method, apparatus, device, and medium for determining the complete motion trajectory of an athlete.

[0005] In a first aspect, the present application provides a method for determining a motion trajectory, the method comprising:

[0006] Inputting the acquired first overhead video frame into a pre-trained position recognition model to acquire the region where the first athlete is located and the corresponding first identification information contained in the first overhead video frame, and inputting the second overhead video frame into the pre-trained position recognition model to acquire the region where the second athlete is located and the corresponding second identification information contained in the second overhead video frame, wherein the second overhead video frame is a overhead video frame captured at a time adjacent to that of the first overhead video frame;

[0007] For the first athlete, input the first overhead video frame and the second overhead video frame into a pre-trained Siamese network model to determine which second athletes included in the second overhead video frame are identical to the first athlete;

[0008] The motion trajectory of the first player is determined based on the area where the first player is located in the first overhead video frame and the area where the player identical to the first player is located in the second overhead video frame.

[0009] In a second aspect, the present application provides a motion trajectory determination device, the device comprising:

[0010] an acquisition module, configured to input the acquired first overhead video frame into a pre-trained position recognition model to acquire an area containing a first athlete and corresponding first identification information contained in the first overhead video frame, and to input a second overhead video frame into the pre-trained position recognition model to acquire an area containing a second athlete and corresponding second identification information contained in the second overhead video frame, wherein the second overhead video frame is a overhead video frame captured at a time adjacent to that of the first overhead video frame;

[0011] a tracking module, configured to input the first overhead video frame and the second overhead video frame into a pre-trained Siamese network model for the first player, and determine which second players included in the second overhead video frame are identical to the first player;

[0012] The determination module is configured to determine a motion trajectory of the first player based on an area where the first player is located in the first overhead video frame and an area where the same player as the first player is located in the second overhead video frame.

[0013] In a third aspect, the present application provides a display device, comprising:

[0014] A display, wherein the display is used to display a motion trajectory diagram containing an athlete;

[0015] A controller configured to:

[0016] At a certain moment, the display is controlled to display identity information of all players and position information of all players on the court, wherein the players do not block each other.

[0017] In a fourth aspect, the present application provides an electronic device, which includes at least a processor and a memory, and the processor is used to implement the steps of any of the above-mentioned motion trajectory determination methods when executing a computer program stored in the memory.

[0018] In a fifth aspect, the present application provides a computer-readable storage medium storing a computer program, which, when executed by a processor, implements the steps of any of the above-mentioned motion trajectory determination methods.

[0019] Since the present application can input the obtained first overhead video frame into the pre-trained position recognition model, obtain the area where the first athlete is located contained in the first overhead video frame and the corresponding first identification information; the obtained second overhead video frame is also input into the pre-trained position recognition model, obtain the area where the second athlete is located contained in the second overhead video frame and the corresponding second identification information; and the first overhead video frame and the second overhead video frame can be simultaneously input into the pre-trained twin network model, for the first athlete corresponding to the first identification information, determine the athlete who is the same as the first athlete in the second overhead video frame; and based on the area where the first athlete is located in the first overhead video frame and the area where the athlete who is the same as the first athlete is located in the second overhead video frame, determine the motion trajectory of the first athlete. Compared with the video frames shot at other shooting angles such as the horizontal angle, the overhead video frame can be largely unaffected by the influence of mutual occlusion between athletes, so when determining the motion trajectory of the first athlete based on the area where the first athlete is located in the first overhead video frame and the area where the athlete who is the same as the first athlete is located in the second overhead video frame, the complete motion trajectory of the first athlete can be obtained accurately and completely to a large extent. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] In order to more clearly illustrate the implementation methods in the embodiments of the present application or related technologies, the following is a brief introduction to the drawings required for use in the embodiments or related technology descriptions. Obviously, the drawings described below are some embodiments of the present application. For ordinary technicians in this field, other drawings can also be obtained based on these drawings.

[0021] Figure 1 A schematic diagram of a motion trajectory determination process provided by some embodiments is shown;

[0022] Figure 2 A schematic diagram of setting up cameras in a sports field according to some embodiments is shown;

[0023] Figure 3 A schematic diagram of determining a motion trajectory provided by some embodiments is shown;

[0024] Figure 4 A schematic diagram of another motion trajectory determination process provided by some embodiments is shown;

[0025] Figure 5A schematic diagram of mapping a first actual area into an image of an actual sports field provided by some embodiments is shown;

[0026] Figure 6 A schematic diagram of a motion trajectory determination device provided by some embodiments is shown;

[0027] Figure 7 A schematic structural diagram of an electronic device provided by some embodiments is shown;

[0028] Figure 8 A schematic structural diagram of a display device provided by some embodiments is shown. DETAILED DESCRIPTION

[0029] In order to determine the complete motion trajectory of an athlete, embodiments of the present application provide a motion trajectory determination method, apparatus, device, and medium.

[0030] In order to make the purpose and implementation of this application clearer, the exemplary implementation of this application will be clearly and completely described below in conjunction with the drawings in the exemplary embodiments of this application. Obviously, the described exemplary embodiments are only part of the embodiments of this application, not all of the embodiments.

[0031] It should be noted that the brief descriptions of terms in this application are only for the purpose of facilitating the understanding of the embodiments described below, and are not intended to limit the embodiments of this application. Unless otherwise specified, these terms should be understood according to their ordinary and usual meanings.

[0032] In the specification and claims of this application and the accompanying drawings, the terms "first," "second," "third," etc. are used to distinguish similar or similar objects or entities, and are not necessarily intended to limit a particular order or sequence, unless otherwise noted. It should be understood that the terms used in this manner are interchangeable under appropriate circumstances.

[0033] The terms "comprise," "include," and "have," and any variations thereof, are intended to cover but not exclude inclusion; for example, a product or device comprising a list of components is not necessarily limited to all the components expressly listed but may include other components not expressly listed or inherent to such product or device.

[0034] The term "module" refers to any known or later developed hardware, software, firmware, artificial intelligence, fuzzy logic, or combination of hardware and / or software code that is capable of performing the functionality associated with that element.

[0035] 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 them. 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 or all of the technical features therein. These modifications or replacements do not deviate the essence of the corresponding technical solutions from the scope of the technical solutions of the embodiments of the present application.

[0036] For ease of explanation, the above description has been made with reference to specific embodiments. However, the above exemplary discussion is not intended to be exhaustive or to limit the embodiments to the specific forms disclosed above. Based on the above teachings, various modifications and variations are possible. The above embodiments are selected and described to better explain the principles and practical applications, so that those skilled in the art can better utilize the embodiments and various different variations of the embodiments suitable for specific use considerations.

[0037] During actual use, the first overhead video frame obtained can be input into a pre-trained position recognition model to obtain the area where the first athlete is located and the corresponding first identification information contained in the first overhead video frame. The second overhead video frame obtained can also be input into the pre-trained position recognition model to obtain the area where the second athlete is located and the corresponding second identification information contained in the second overhead video frame. The first overhead video frame and the second overhead video frame can be simultaneously input into a pre-trained twin network model. For the first athlete corresponding to the first identification information, the athlete who is the same as the first athlete in the second athlete contained in the second overhead video frame is determined, and the movement trajectory of the first athlete is determined based on the area where the first athlete is located in the first overhead video frame and the area where the athlete who is the same as the first athlete is located in the second overhead video frame. Compared with video frames shot at other shooting angles such as a level angle, overhead video frames are less affected by occlusion between athletes, and when determining the motion trajectory of the first athlete based on the area where the first athlete is located in the first overhead video frame and the area where the same athlete as the first athlete is located in the second overhead video frame, the complete motion trajectory of the first athlete can be obtained with greater accuracy and completeness, thereby achieving the purpose of determining the complete motion trajectory of the athlete.

[0038] Figure 1 A schematic diagram of a motion trajectory determination process provided by some embodiments is shown. Figure 1 As shown, the process includes the following steps:

[0039] S101: Input the acquired first overhead video frame into a pre-trained position recognition model to acquire the area where the first athlete is located and the corresponding first identification information contained in the first overhead video frame, and input the second overhead video frame into the pre-trained position recognition model to acquire the area where the second athlete is located and the corresponding second identification information contained in the second overhead video frame, wherein the second overhead video frame is an overhead video frame that is adjacent to the shooting time of the first overhead video frame.

[0040] In a possible implementation, the motion trajectory determination method provided in the embodiment of the present application is applied to an electronic device, which may be, for example, a PC, a mobile terminal, or a server.

[0041] In a possible implementation, a bird's-eye view camera may be provided above a sports field such as a stadium, and a bird's-eye view video frame with a shooting angle overlooking the sports field may be acquired based on the bird's-eye view camera. Figure 2 Schematic diagram of setting up cameras in a sports field according to some embodiments is shown. Figure 2 As shown, a bird's-eye view camera can be set directly above the center of the sports field (court), and a bird's-eye view video frame with a shooting range including the entire sports field can be obtained based on the bird's-eye view camera. In a possible embodiment, multiple bird's-eye view cameras can also be set above the sports field. The present application does not specifically limit the number of bird's-eye view cameras. For example, taking the sports field as a basketball court, a bird's-eye view camera can be set directly above each (two) half of the basketball court with the center line of the basketball court as the boundary. Each bird's-eye view camera only obtains a bird's-eye view video frame of the half of the basketball court where the bird's-eye view camera is located. It can be understood that because the total shooting range of the multiple bird's-eye view cameras arranged above the sports field can cover the entire sports field, the total range of the sports field contained in each bird's-eye view video frame obtained by each bird's-eye view camera for the same shooting time can also include the entire sports field. For example, if multiple overhead cameras are provided, each overhead camera sends the acquired overhead video frame to the electronic device, and the electronic device can, based on the shooting time of the received overhead video frame, splice each overhead video frame of the same shooting time to form an overhead video frame, and execute step S101 based on the spliced ​​overhead video frame. It can be understood that the spliced ​​overhead video frame includes the entire sports field. Of course, the electronic device may not splice each overhead video frame of the same shooting time, but execute step S101 based on each received overhead video frame respectively, which can be flexibly selected according to needs. For the convenience of description, the embodiment of the present application refers to each overhead video frame of the same shooting time as a first overhead video frame, and the total range of the sports field contained in the first overhead video frame may include the entire sports field.

[0042] In one possible implementation, after the electronic device acquires the first overhead video frame, it may input the first overhead video frame into a pre-trained location recognition model. Using this location recognition model, the region where each first athlete contained in the first overhead video frame is located may be acquired. To distinguish each first athlete contained in the first overhead video frame, corresponding first identification information may be assigned to each first athlete. The first identification information may be flexibly configured as needed, for example, using letters, numbers, etc., and is not specifically limited in this application.

[0043] In a possible embodiment, for each first athlete in the first overhead video frame, in order to determine the motion trajectory of the first athlete, the overhead video frame adjacent to the first overhead video frame shooting time, that is, the second overhead video frame is also input into the position recognition model completed by pre-training, identical with the first overhead video frame, the total range of the sports field included in the second overhead video frame can also include the entire sports field. Similarly, through this position recognition model, the area where each second athlete included in the second overhead video frame is located and the corresponding second identification information for each second athlete configuration can also be obtained. The second identification information can also be flexibly set according to demand, for example, can be letters, numbers, etc., and the application does not specifically limit this. It is worth noting that, for the unique identification of each first athlete and each second athlete, each first identification information is different from each second identification information.

[0044] S102: For the first athlete corresponding to the first identification information, the first overhead video frame and the second overhead video frame are input into a pre-trained twin network model to determine the second athletes included in the second overhead video frame who are the same as the first athlete.

[0045] In one possible implementation, for each first athlete in the first overhead video frame, in order to determine the runner who is the same as the first athlete in the second overhead video frame, the first overhead video frame and the second overhead video frame can be simultaneously input into a pre-trained twin network model. Based on the twin network model, the athlete who is the same as the first athlete among the second athletes included in the second overhead video frame can be accurately determined.

[0046] In one possible embodiment, for each first athlete in the first overhead video frame, after determining the athlete among the second athletes included in the second overhead video frame who is the same as the first athlete, the second identification information corresponding to the athlete among the second athletes who is the same as the first athlete can be updated to the first identification information corresponding to the first athlete, that is, the second identification information is updated to be the same as the first identification information, so as to indicate that the second athlete corresponding to the second identification information and the first athlete corresponding to the first identification information are the same athlete.

[0047] S103: Determine a motion trajectory of the first player based on an area where the first player is located in the first overhead video frame and an area where the same player as the first player is located in the second overhead video frame.

[0048] In one possible implementation, for each first athlete in the first overhead video frame, after determining the athlete identical to the first athlete in each overhead video frame, the motion trajectory of the first athlete can be determined based on the region in which the first athlete is located in each overhead video frame, the shooting time of each overhead video frame, etc. For example, taking the second overhead video frame as a video frame with the next adjacent shooting time of the first overhead video frame as an example, the motion trajectory of the first athlete at the corresponding shooting times of the first and second overhead video frames is from the region in which the first athlete is located in the first overhead video frame to the region in which the athlete identical to the first athlete is located in the second overhead video frame.

[0049] In one possible implementation, the overhead camera can capture overhead video frames of the sports field in real time and send each captured overhead video frame to the electronic device. The electronic device can then perform steps S101-S103 for each received overhead video frame captured at a specific time. Based on this, the electronic device can obtain the first athlete's complete real-time motion trajectory.

[0050] In a possible implementation, the electronic device can determine the area of ​​the sports field in the first overhead video frame (or the second overhead video frame), and identify the area where the athletes (for ease of description, referred to as the first athlete contained in the first overhead video frame and the second athlete contained in the second overhead video frame) contained in the area of ​​the sports field in the first overhead video frame (or the second overhead video frame) are located. It is understandable that when the athlete enters the field and appears in the area where the sports field is located, the electronic device will obtain the area where the athlete is located, etc. When the athlete leaves the field and exceeds the area where the sports field is located, the electronic device may no longer obtain the area where the athlete is located, etc. The electronic device can determine the complete motion trajectory of the athlete in the sports field.

[0051] Compared to video frames shot at other shooting angles such as a level view angle, the overhead video frames in this application are largely unaffected by the effects of mutual occlusion between athletes, and thus can avoid the problems of tracking failure and mistracking caused by determining (tracking) the athlete's motion trajectory based on the level view video frames in the related art, and can improve the accuracy of the determined athlete's motion trajectory. Therefore, when determining the first athlete's motion trajectory based on the area where the first athlete is located in the first overhead video frame and the area where the same athlete as the first athlete is located in the second overhead video frame, the present application can obtain the first athlete's complete motion trajectory accurately and completely to a greater extent.

[0052] The present application can input the first overhead video frame obtained into a pre-trained position recognition model to obtain the area where the first athlete is located contained in the first overhead video frame and the corresponding first identification information; the second overhead video frame obtained is also input into the pre-trained position recognition model to obtain the area where the second athlete is located contained in the second overhead video frame and the corresponding second identification information; and the first overhead video frame and the second overhead video frame can be simultaneously input into the pre-trained twin network model, for the first athlete corresponding to the first identification information, determine the second athlete contained in the second overhead video frame that is the same as the first athlete, and determine the motion trajectory of the first athlete based on the area where the first athlete is located in the first overhead video frame and the area where the athlete is the same as the first athlete in the second overhead video frame. Compared with video frames shot at other shooting angles such as a horizontal angle, the overhead video frame can be largely unaffected by the influence of mutual occlusion between athletes, and when determining the motion trajectory of the first athlete based on the area where the first athlete is located in the first overhead video frame and the area where the athlete is the same as the first athlete in the second overhead video frame, the complete motion trajectory of the first athlete can be obtained accurately and completely to a large extent.

[0053] In one possible implementation, to generate a motion trajectory of a first athlete, based on the above embodiment, in this embodiment of the present application, determining the motion trajectory of the first athlete based on an area where the first athlete is located in the first overhead video frame and an area where an athlete identical to the first athlete is located in the second overhead video frame includes:

[0054] determining, based on a first area where the first athlete is located in the first overhead video frame, a first actual area where the first area is located in the actual sports field;

[0055] determining a second actual area in the actual sports field where the second area is located based on a second area in the second overhead video frame where the athlete identical to the first athlete is located;

[0056] The first actual area and the second actual area are respectively mapped into the image of the actual sports field to generate a motion trajectory of the first player.

[0057] In one possible implementation, when generating the motion trajectory of each first athlete, for each first athlete in the first overhead video frame, a first actual area in the actual sports field where the first area is located can be determined based on the first area where the first athlete is located in the first overhead video frame. Specifically, when determining the first actual area in the actual sports field where the first area is located based on the first area where the first athlete is located in the first overhead video frame, at least the following two methods can be used:

[0058] The first method is to convert the first area (pixel position in the pixel coordinate system) where the first athlete is located in the first overhead video frame into the first actual area (actual field position in the field coordinate system) where the first area is located in the actual sports field.

[0059] For example, the first area (pixel position in the pixel coordinate system) where the first athlete is located in the first overhead video frame is represented by P ix (u, v) T Indicated by, where u is the horizontal coordinate in the pixel coordinate system, and v is the vertical coordinate in the pixel coordinate system. The first actual area where the first area is located in the actual sports field (the actual field position in the field coordinate system) is represented by P w (x w ,y w , z w ) T It indicates that the first area where the first athlete is located in the first overhead video frame can be converted into the first actual area where the first area is located in the actual sports field by the following conversion formula;

[0060] The conversion formula is: Among them, z c is the optical axis of the top-view camera, M1 is the intrinsic parameter matrix of the top-view camera, and M2 is the extrinsic parameter matrix of the top-view camera. The conversion of pixel positions in the pixel coordinate system to actual field positions in the corresponding court coordinate system is a conventional technique and will not be further elaborated here.

[0061] The second method is to predetermine the correspondence between the pixel size in the overhead video frame and the size of the actual sports field (the scale of the overhead video frame), and determine the first actual area (actual field position) where the first area is located in the actual sports field based on the scale and the first area (pixel position) where the first athlete is located in the first overhead video frame.

[0062] In one possible implementation, taking the standard basketball court size of 15*28 meters as an example, the standard basketball court size can be gridded, for example, divided into 1500*2800 grids, and the length and width of each grid are 1 cm. That is, the standard basketball court size can be refined to the centimeter level, so that the position of the first actual area where the first athlete is located in the actual sports field can be accurately determined to the centimeter level.

[0063] The process of determining the second actual area where the second area is located in the actual sports venue based on the second area where the athlete identical to the first athlete is located in the second overhead video frame is the same as the process of determining the first actual area where the first area is located in the actual sports venue based on the first area where the first athlete is located in the first overhead video frame in the above-mentioned embodiment, and will not be repeated here.

[0064] After determining the first actual area and the second actual area, the motion track of the first athlete can be determined based on the first actual area and the second actual area. Specifically, when determining the motion track of the first athlete based on the first actual area and the second actual area, the first actual area and the second actual area can be mapped to the image of the actual sports field respectively, thereby generating the motion track of the first athlete. Exemplary, taking the video frame of the next adjacent shooting time of the first overhead video frame as an example, the motion track of the first athlete is that the first athlete is located in the first actual area in the first overhead video frame, and moves to the second actual area in the second overhead video frame. The process wherein the first actual area and the second actual area are mapped to the image of the actual sports field can adopt prior art, which will not be repeated here.

[0065] See Figure 3 , Figure 3 A schematic diagram of determining a motion trajectory provided by some embodiments is shown. The left side of the figure (left and right as shown in the figure) is a first overhead video frame containing multiple first athletes. The right side of the figure (left and right as shown in the figure) is a schematic diagram of determining the first actual area where each first athlete in the first overhead video frame is located in the left figure, and mapping the first actual area to the image of the actual sports field. Figure 3 It can also be seen that compared with video frames shot based on other shooting angles such as a level angle, the overhead video frame can be largely unaffected by the influence of mutual occlusion between athletes. By mapping the first actual area and the second actual area to the image of the actual sports field respectively, the motion trajectory of the first athlete generated can accurately and completely obtain the complete motion trajectory of the first athlete to a large extent.

[0066] Based on the above embodiments, in an embodiment of the present application, after obtaining the area where the second player is located and the corresponding second identification information contained in the second overhead video frame, the method further includes:

[0067] It is determined whether there is a target second player in the second overhead video frame whose distance from the first player is not greater than a set first distance threshold. If yes, the target second player is determined to be the same player as the first player.

[0068] In one possible implementation, for each first athlete in the first overhead video frame, when determining which second athletes included in the second overhead video frame are identical to the first athlete, the distance between each second athlete and the first athlete in the second overhead video frame may be first determined. Specifically, when determining the distance between each second athlete and the first athlete in the second overhead video frame, the second actual area corresponding to each second athlete and the first actual area corresponding to the first athlete in the second overhead video frame may be first determined based on the method provided in the above embodiment. Then, based on the first actual area and the second actual area, the distance between each second athlete and the first athlete in the second overhead video frame may be determined.

[0069] In a possible embodiment, in view of the fact that the speed of movement of the athlete is no more than certain speed threshold value usually, as 10 meters per second etc., can according to the frame rate of overlooking camera, determine the time difference (for convenience of description, be called duration) of the shooting time interval of adjacent two overlooking video frames, can according to the product of this duration and speed threshold value, determine the first distance threshold value of setting, and think that the distance difference of same athlete in adjacent two overlooking video frames is not more than (less than or equal to) this first distance threshold value usually.Exemplary, suppose that the speed of movement of the athlete is no more than 10 meters per second usually, when the frame rate of overlooking camera is 60Hz, the first distance threshold value of setting can be 17 centimetres, namely thinks that the distance difference of same athlete in adjacent two overlooking video frames is not more than 17 centimetres usually.For another example, suppose that the speed of movement of the athlete is no more than 10 meters per second usually, when the frame rate of overlooking camera is 30Hz, the first distance threshold value of setting can be 33 centimetres, namely think that the distance difference of same athlete in adjacent two overlooking video frames is not more than 33 centimetres usually. Based on this, in one possible implementation, a second athlete in the second overhead video frame whose distance from the first athlete in the first overhead video frame is no greater than a set first distance threshold (i.e., the second athlete in the second overhead video frame that is closer to the first athlete in the first overhead video frame) can be regarded as the same athlete as the first athlete.

[0070] Specifically, after determining the distance between each second player and the first player in the second overhead video frame, it can be determined whether there is a target second player in the second overhead video frame whose distance from the first player is no greater than a set first distance threshold. If so, the target second player can be determined to be the same player as the first player. It is understood that the number of target second players is 1.

[0071] In one possible embodiment, if it is determined that the distance between the first athlete and at least two second athletes included in the second overhead video frame is no greater than a set first distance threshold, that is, there are at least two second athletes in the second overhead video frame whose distance from the first athlete is no greater than the set first distance threshold, then it can be considered that the distance between the athletes is relatively close at this time, and based on the distance from the first athlete, it is difficult to distinguish the second athletes included in the second overhead video frame as the same athlete as the first athlete. In order to accurately determine the second athletes included in the second overhead video frame that are the same as the first athlete, the first overhead video frame and the second overhead video frame can be simultaneously input into a pre-trained twin network model, and based on the twin network model, the second athletes included in the second overhead video frame that are the same as the first athlete are determined.

[0072] In a possible implementation, based on the above embodiments, in the embodiment of the present application, the process of training the twin network model includes:

[0073] Obtaining a first sample overhead video frame containing a fourth athlete and a second sample overhead video frame containing a fifth athlete from the sample set, wherein for the fourth athlete contained in the first sample overhead video frame, the first sample overhead video frame and the second sample overhead video frame have corresponding sample labels, and the sample labels are used to identify whether the fifth athlete contained in the second sample overhead video frame and the fourth athlete are the same athlete;

[0074] Determining, using the original Siamese network model, for a fourth athlete included in the first sample overhead video frame, an identification tag corresponding to the first sample overhead video frame and the second sample overhead video frame, the identification tag being used to identify whether a fifth athlete included in the second sample overhead video frame is the same athlete as the fourth athlete;

[0075] The original twin network model is trained according to the sample label and the identification label to obtain a trained twin network model.

[0076] In one possible implementation, when training the original twin network model, a first sample overhead video frame containing a fourth athlete and a second sample overhead video frame containing a fifth athlete in the sample set can be first obtained, wherein for each fourth athlete contained in the first sample overhead video frame, the first sample overhead video frame and the second sample overhead video frame correspond to a common sample label, and the sample label is used to identify whether the fifth athlete contained in the second sample overhead video frame is the same athlete as the fourth athlete.

[0077] When training the original twin network model, the first sample overhead video frame and the second sample overhead video frame can be simultaneously input into the original twin network model. Through the original twin network model, for the fourth athlete contained in the first sample overhead video frame, the identification labels corresponding to the first sample overhead video frame and the second sample overhead video frame can be determined, where the identification label can be used to identify whether the fifth athlete and the fourth athlete contained in the second sample overhead video frame are the same athlete.

[0078] In a specific implementation, after determining the identification labels corresponding to the first sample bird's-eye view video frame and the second sample bird's-eye view video frame input, since the sample labels corresponding to the first sample bird's-eye view video frame and the second sample bird's-eye view video frame are pre-saved, the accuracy of the recognition result of the twin network model can be determined based on whether the sample labels and the identification labels are consistent. In a specific implementation, if they are inconsistent, it means that the recognition result of the twin network model is inaccurate, and the parameters of the twin network model need to be adjusted to train the twin network model.

[0079] In a possible implementation, the above operation can be performed on each sample overhead video frame in the sample set, and when the preset convergence condition is met, it is determined that the training of the twin network model is completed.

[0080] The preset convergence condition can be satisfied when the sample bird's-eye view video frames in the sample set pass through the original twin network model, the number of correctly identified sample bird's-eye view video frames is greater than a set number, or the number of iterations of training the twin network model reaches a set maximum number of iterations, etc. This setting can be flexibly made in the specific implementation and is not specifically limited here.

[0081] In one possible implementation, when training the original twin network model, the sample overhead video frames in the sample set can be divided into training sample video frames and test sample video frames. The original twin network model is first trained based on the training sample video frames, and then the reliability of the trained twin network model is verified based on the test sample video frames.

[0082] In order to identify the identity information of the first athlete, based on the above embodiments, in the embodiment of the present application, the method further includes:

[0083] Inputting the acquired video frame into a pre-trained position recognition model to acquire an area where a third athlete is located contained in the video frame; and identifying identity information of the third athlete contained in the video frame; wherein the video frame and the first overhead video frame are shot at the same time;

[0084] Determine whether there is a target third athlete in the video frame whose distance from the first athlete is not greater than a set second distance threshold; if so, determine the target third athlete as the same athlete as the first athlete; and update the first identification information corresponding to the first athlete based on the identity information of the target third athlete.

[0085] In a possible implementation, the identification information of the first athlete determined based on the above embodiment is the first identification information. In order to accurately determine the identity information such as the name, team, uniform color, uniform number, etc. of the first athlete corresponding to the first identification information, a sideline camera can be set in advance on the periphery of a sports venue such as a stadium, and a video frame of the athlete in the sports venue can be obtained based on the sideline camera. Wherein the present application does not specifically limit the shooting angle of the video frame. For example, the shooting angle of the video frame can be a shooting angle parallel to the horizon (straight angle), or a shooting angle at a certain angle to the horizon (non-parallel to the horizon), as long as the identity information of the athlete contained in the video frame can be identified based on the video frame. For the convenience of description, the video frame that can identify the identity information of the athlete contained in the video frame will be referred to as a straight video frame. It is understandable that the shooting angle of the straight video frame can be parallel to the horizon (straight angle), or can be at a certain angle to the horizon (non-parallel to the horizon), wherein the size of the angle can be flexibly set according to demand, and the present application does not specifically limit this.

[0086] See Figure 2 This application does not specifically limit the number of side field cameras. The total shooting range of each side field camera can cover the entire sports field. Similar to the overhead camera, after each side field camera acquires a video frame, it can send the acquired video frame to the electronic device.

[0087] The electronic device can, based on the shooting time of the received video frames, splice each video frame of the same shooting time to form a video frame, and based on the spliced ​​video frames, perform the step of inputting the acquired video frames into a pre-trained position recognition model. It can be understood that the spliced ​​video frames include the entire sports field. Of course, the electronic device may also not splice each video frame of the same shooting time, but instead perform the step of inputting the acquired video frames into a pre-trained position recognition model based on each received video frame, which can be flexibly selected according to needs. For the convenience of description, the embodiment of the present application refers to each video frame of the same shooting time as a video frame, and the total range of the sports field contained in the video frame may include the entire sports field.

[0088] In one possible implementation, the electronic device may input the acquired video frame into a pre-trained location recognition model, and use the location recognition model to obtain the region where the third athlete is located within the video frame. Simultaneously, the electronic device may also identify the identity information of the third athlete within the video frame. The process of identifying the identity information of the third athlete within the video frame may include at least the following two identification methods:

[0089] The first identification method: based on face recognition, determining the identity information of the third player contained in the video frame.

[0090] The process of determining the identity information of the third player contained in the video frame based on face recognition can adopt existing technologies, which will not be described in detail here.

[0091] The second identification method is to determine the uniform number and uniform color of the third athlete contained in the video frame; and determine the identity information of the third athlete based on the stored correspondence between the uniform color, uniform number and the athlete's identity information.

[0092] Specifically, when determining the uniform number and uniform color of the third athlete contained in the video frame, the video frame can be input into a pre-trained uniform color and number recognition model. The uniform color and uniform number of the third athlete contained in the video frame can be obtained using the uniform color and number recognition model. Alternatively, optical character recognition (OCR) can be used to identify the uniform number of the third athlete contained in the video frame, or related color recognition techniques can be used to identify the uniform color of the third athlete contained in the video frame. These techniques are not further described here.

[0093] In order to determine the identity information of an athlete, the electronic device may pre-store a correspondence between the uniform color, uniform number, and the athlete's identity information, wherein the correspondence between the uniform color, uniform number, and the athlete's identity information may be set based on the uniform color, uniform number, and identity information of the participating teams in the sports event, and this application does not specifically limit this. When determining the identity information of a third athlete contained in a video frame, after determining the uniform number and uniform color of the third athlete contained in the video frame, the target identity information corresponding to the uniform number and uniform color of the third athlete contained in the video frame may be determined as the identity information of the third athlete based on the stored correspondence between the uniform color, uniform number, and identity information of the athlete.

[0094] Specifically, when determining the identity information of the first athlete corresponding to the first identification information, an athlete identical to the first athlete in the video frame can be determined, and the identity information of the athlete identical to the first athlete in the video frame shot at the same time as the first overhead video frame can be used as the identity information of the first athlete corresponding to the first identification information. Specifically, when determining the identity information of the first athlete identical to the first athlete in the video frame, a determination can be made as to whether there is a target third athlete in the video frame whose distance from the first athlete is no greater than a set second distance threshold. If there is a target third athlete whose distance from the first athlete is no greater than the set second distance threshold, the target third athlete can be determined as the athlete identical to the first athlete. Furthermore, the first identification information of the first athlete can be updated based on the identity information of the target third athlete, i.e., the first identification information can be updated to the identity information of the target third athlete.

[0095] For ease of understanding, the motion trajectory determination process provided by this application is described below through a specific embodiment. Figure 4 FIG. 4 shows another schematic diagram of a motion trajectory determination process provided by some embodiments, such as Figure 4 As shown, the process includes the following steps:

[0096] S401: Input the acquired first overhead video frame into a pre-trained position recognition model to acquire the area where the first athlete is located and the corresponding first identification information contained in the first overhead video frame, and input the second overhead video frame into the pre-trained position recognition model to acquire the area where the second athlete is located and the corresponding second identification information contained in the second overhead video frame, wherein the second overhead video frame is an overhead video frame that is adjacent to the shooting time of the first overhead video frame.

[0097] S402: For the first player corresponding to the first identification information, the first overhead video frame and the second overhead video frame are input into a pre-trained twin network model to determine the second players included in the second overhead video frame who are the same as the first player.

[0098] S403: Input the acquired video frame into a pre-trained position recognition model to obtain the area where the third athlete is located in the video frame; and identify the identity information of the third athlete contained in the video frame; wherein the video frame is shot at the same time as the first overhead video frame; determine whether there is a target third athlete in the video frame whose distance from the first athlete is not greater than a set second distance threshold; if so, determine the target third athlete as the same athlete as the first athlete; and update the first identification information corresponding to the first athlete based on the identity information of the target third athlete.

[0099] S404: Determine a first actual area in the actual sports field where the first area is located based on the first area where the first athlete is located in the first overhead video frame; determine a second actual area in the actual sports field where the second area is located based on the second area where the athlete identical to the first athlete is located in the second overhead video frame; map the first actual area and the second actual area respectively to an image of the actual sports field to generate a motion trajectory of the first athlete.

[0100] It is worth noting that the steps included in the above-mentioned motion trajectory determination process are only an example. This application does not limit the order between S403 and S404, that is, after determining that the second athlete included in the second overhead video frame is the same as the first athlete, the order between the two steps of first determining the target third athlete and updating the first identification information corresponding to the first athlete based on the identity information of the target third athlete, or first generating the motion trajectory of the first athlete is not limited.

[0101] See Figure 5 , Figure 5 FIG. 4 shows a schematic diagram of mapping a first actual area into an image of an actual sports field provided by some embodiments, such as Figure 5As shown, after the first identification information corresponding to the first athlete is updated based on the identity information of the third target athlete, such as the uniform color and uniform number, the user can further obtain the athlete's identity information, thereby better allowing the user to know which athlete each motion trajectory belongs to. In this application, the electronic device can not only determine the complete motion trajectory of the first athlete based on the overhead video frame obtained by the overhead camera, but also determine the identity information of the first athlete based on the video frame obtained by the sideline camera, thereby obtaining the complete motion trajectory of the athlete including the identity information of the first athlete, allowing the user to have a macro understanding of the athlete's position and motion trajectory, which is beneficial for coaches and athletes to analyze tactics, etc., and improves the viewing experience of the event and the user experience.

[0102] In a possible implementation, based on the above embodiments, in the embodiment of the present application, the process of training the location recognition model includes:

[0103] Obtain any first sample image containing an athlete in the sample set, the first sample image corresponding to a sample region position label of the athlete contained in the first sample image;

[0104] Determining, by means of an original position recognition model, a recognition region position label of the athlete contained in the first sample image;

[0105] The original position recognition model is trained according to the sample area position label and the recognition area position label to obtain a trained position recognition model.

[0106] In an embodiment of the present application, a sample set includes multiple first sample images, wherein the first sample image can be a bird's-eye view video frame, or a head-on view video frame or other video frame. In order to obtain the regional position information of the athlete, each first sample image also corresponds to a sample region position label of each athlete contained in the first sample image. In a possible implementation, the sample region position label may include the coordinate position of the center pixel point of the athlete's target frame in the first sample image, the coordinate position of the bottom midpoint pixel point of the target frame in the first sample image, etc. Exemplarily, when the first sample image is a bird's-eye view video frame, the sample region position label corresponding to the first sample image may be the coordinate position of the center pixel point of the athlete's target frame contained in the first sample image in the first sample image. When the first sample image is a head-on view video frame or other video frame, the sample region position label corresponding to the first sample image may be the coordinate position of the bottom midpoint pixel point of the athlete's target frame contained in the first sample image in the first sample image.

[0107] When training the original position recognition model, a first sample image containing an athlete in the sample set can be obtained, and the first sample image corresponds to a sample region position label. The obtained first sample image is input into the original position recognition model, and the original position recognition model is used to obtain the recognition region position label of the athlete contained in the first sample image.

[0108] In a specific implementation, after determining the recognition region position label of the input first sample image, since the sample region position label of the first sample image is pre-stored, the accuracy of the recognition result of the position recognition model can be determined based on whether the sample region position label is consistent with the recognition region position label. In a specific implementation, if they are inconsistent, it indicates that the recognition result of the position recognition model is inaccurate, and the parameters of the position recognition model need to be adjusted to train the position recognition model.

[0109] In a specific implementation, when adjusting the parameters in the location recognition model, a gradient descent algorithm may be used to back-propagate the gradients of the parameters of the location recognition model, thereby training the location recognition model.

[0110] In a possible implementation, the above operation may be performed on each first sample image in the sample set, and when a preset convergence condition is met, it is determined that the training of the position recognition model is completed.

[0111] The preset convergence condition may be satisfied by the first sample image in the sample set passing through the original position recognition model, the number of correctly recognized first sample images being greater than a set number, or the number of iterations of training the position recognition model reaching a set maximum number of iterations. These settings may be flexibly made in specific implementations and are not specifically limited here.

[0112] In one possible implementation, when training the original position recognition model, the first sample image in the sample set can be divided into a training sample image and a test sample image. The original position recognition model is first trained based on the training sample image, and then the reliability of the trained position recognition model is verified based on the test sample image.

[0113] In a possible implementation, based on the above embodiments, in the embodiment of the present application, the process of training the uniform color and number recognition model includes:

[0114] Obtain any second sample image containing an athlete in the sample set, the second sample image corresponding to a sample uniform color label and a sample uniform number label of the uniform worn by the athlete contained in the second sample image;

[0115] Determining, by using the original uniform color and number recognition model, the identification uniform color label and the identification uniform number label of the uniform worn by the athlete contained in the second sample image;

[0116] The original uniform color and number recognition model is trained according to the sample uniform color label and the identified uniform color label, the sample uniform number label and the identified uniform number label to obtain a trained uniform color and number recognition model.

[0117] In the embodiments of the present application, the sample set includes multiple second sample images. In one possible implementation, the second sample images may be video frames captured at a horizontal angle or at a specific angle to the horizon. To obtain the uniform color and uniform number of the athlete's uniform, each second sample image also corresponds to a sample uniform color label and a sample uniform number label for the athlete's uniform contained in the second sample image.

[0118] When training the original uniform color and number recognition model, a second sample image containing an athlete in the sample set can be obtained. The second sample image corresponds to a sample uniform color label and a sample uniform number label of the uniform worn by the athlete contained in the second sample image. The obtained second sample image is input into the original uniform color and number recognition model, and the original uniform color and number recognition model is used to obtain the identification uniform color label and identification uniform number label of the uniform worn by the athlete contained in the second sample image.

[0119] In a specific implementation, after determining the identification uniform color label and identification uniform number label of the athlete's uniform contained in the input second sample image, because the sample uniform color label and sample uniform number label of the athlete's uniform contained in the second sample image are pre-saved, it is possible to determine whether the recognition result of the uniform color and number recognition model is accurate based on whether the sample uniform color label is consistent with the identification uniform color label and whether the sample uniform number label is consistent with the identification uniform number label. In a specific implementation, if they are inconsistent, it means that the recognition result of the uniform color and number recognition model is inaccurate, and it is necessary to adjust the parameters of the uniform color and number recognition model to train the uniform color and number recognition model.

[0120] In a specific implementation, when adjusting the parameters in the uniform color and number recognition model, a gradient descent algorithm can be used to backpropagate the gradients of the parameters of the uniform color and number recognition model, thereby training the uniform color and number recognition model.

[0121] In a possible implementation, the above operation may be performed on each second sample image in the sample set. When a preset convergence condition is met, it is determined that the training of the uniform color and number recognition model is completed.

[0122] The preset convergence condition may be satisfied if the number of correctly recognized second sample images in the sample set exceeds a set number, or the number of iterations of training the uniform color and number recognition model reaches a set maximum number of iterations. These settings can be flexibly adjusted in practice and are not specifically limited here.

[0123] In one possible implementation, when training the original team uniform color and number recognition model, the second sample image in the sample set can be divided into a training sample image and a test sample image. The original team uniform color and number recognition model is first trained based on the training sample image, and then the reliability of the trained team uniform color and number recognition model is verified based on the test sample image.

[0124] Based on the same technical concept, the present application also provides a motion trajectory determination device, which can implement the process executed by the electronic device in the aforementioned embodiment. Figure 6 Schematic diagram of a motion trajectory determination device provided by some embodiments is shown. Figure 6 As shown, the device includes:

[0125] An acquisition module 61 is configured to input the acquired first overhead video frame into a pre-trained position recognition model to acquire the area where the first athlete is located and the corresponding first identification information contained in the first overhead video frame, and to input the second overhead video frame into the pre-trained position recognition model to acquire the area where the second athlete is located and the corresponding second identification information contained in the second overhead video frame, wherein the second overhead video frame is a overhead video frame captured at a time adjacent to that of the first overhead video frame;

[0126] a tracking module 62 configured to input the first overhead video frame and the second overhead video frame into a pre-trained Siamese network model for the first athlete, and determine which second athletes included in the second overhead video frame are identical to the first athlete;

[0127] The determination module 63 is configured to determine the motion trajectory of the first athlete based on the area where the first athlete is located in the first overhead video frame and the area where the same athlete as the first athlete is located in the second overhead video frame.

[0128] In one possible embodiment, the tracking module 62 is also used to determine, for the first identification information corresponding to the first athlete, whether the distance between the first athlete and at least two second athletes contained in the second overhead video frame is not greater than a set first distance threshold before inputting the first overhead video frame and the second overhead video frame into the pre-trained twin network model; if so, proceed to subsequent steps.

[0129] In a possible implementation, the tracking module 62 is further configured to determine whether there is a target second athlete in the second overhead video frame whose distance from the first athlete is no greater than a set first distance threshold; if so, determining the target second athlete as the same athlete as the first athlete.

[0130] In a possible implementation, the device further includes:

[0131] The recognition module is configured to input the acquired video frame into a pre-trained position recognition model to acquire an area where a third athlete contained in the video frame is located; and to identify the identity information of the third athlete contained in the video frame; wherein the video frame is shot at the same time as the first overhead video frame; determine whether there is a target third athlete in the video frame whose distance from the first athlete is not greater than a set second distance threshold, and if so, determine the target third athlete as the same athlete as the first athlete; and update the first identification information corresponding to the first athlete based on the identity information of the target third athlete.

[0132] In a possible implementation, the recognition module is specifically configured to determine the identity information of the third athlete contained in the video frame based on face recognition; or

[0133] Determine the uniform number and uniform color of the third athlete contained in the video frame; and determine the identity information of the third athlete based on the stored correspondence between the uniform color, uniform number and the athlete's identity information.

[0134] In a possible implementation, the recognition module is specifically configured to input the video frame into a pre-trained uniform color and number recognition model to obtain the uniform color and uniform number of the third athlete contained in the video frame.

[0135] In one possible embodiment, the determination module 63 is specifically used to determine a first actual area where the first area is located in the actual sports venue based on the first area where the first athlete is located in the first overhead video frame; determine a second actual area where the second area is located in the actual sports venue based on the second area where the athlete identical to the first athlete is located in the second overhead video frame; and map the first actual area and the second actual area respectively to the image of the actual sports venue to generate a motion trajectory of the first athlete.

[0136] Based on the same technical concept, the present application also provides an electronic device. Figure 7 A schematic diagram of the structure of an electronic device provided by some embodiments is shown. Figure 7 As shown, it includes: a processor 71, a communication interface 72, a memory 73 and a communication bus 74, wherein the processor 71, the communication interface 72, and the memory 73 communicate with each other through the communication bus 74;

[0137] The memory 73 stores a computer program. When the program is executed by the processor 71, the processor 71 completes the steps of the electronic device performing the corresponding functions in the above method.

[0138] The communication bus mentioned in the electronic device mentioned above may be a Peripheral Component Interconnect (PCI) bus or an Extended Industry Standard Architecture (EISA) bus. This communication bus can be divided into an address bus, a data bus, a control bus, etc. For ease of illustration, only one thick line is used in the figure, but this does not mean that there is only one bus or only one type of bus.

[0139] The communication interface 72 is used for communication between the electronic device and other devices.

[0140] The memory may include random access memory (RAM) or non-volatile memory (NVM), such as at least one disk memory. Alternatively, the memory may be at least one storage device located away from the processor.

[0141] The above-mentioned processor can be a general-purpose processor, including a central processing unit, a network processor (NP), etc.; it can also be a digital signal processing processor (DSP), an application-specific integrated circuit, a field programmable gate array or other programmable logic device, a discrete gate or transistor logic device, a discrete hardware component, etc.

[0142] Based on the same technical concept, the present application also provides a computer-readable storage medium, which stores a computer program that can be executed by an electronic device, and the computer-executable instructions are used to enable the computer to execute the process executed by the aforementioned method part.

[0143] The above-mentioned computer-readable storage medium can be any available medium or data storage device that can be accessed by the processor in the electronic device, including but not limited to magnetic storage such as floppy disks, hard disks, magnetic tapes, magneto-optical disks (MO), etc., optical storage such as CDs, DVDs, BDs, HVDs, etc., and semiconductor storage such as ROMs, EPROMs, EEPROMs, non-volatile memories (NANDFLASH), solid-state drives (SSDs), etc.

[0144] Based on the same technical concept, the present application also provides a display device, Figure 8 A schematic diagram of a display device structure provided by some embodiments is shown. Figure 8 As shown, the display device includes:

[0145] A display 81, wherein the display 81 is used to display a motion trajectory diagram including an athlete;

[0146] A controller 82 configured to:

[0147] At a certain moment, the display 81 is controlled to display the identity information of all players and the position information of all players on the court, wherein the players do not block each other.

[0148] In a possible implementation, the display device can complete the steps of the electronic device performing corresponding functions in the above method, which will not be described in detail here.

[0149] Those skilled in the art will appreciate that the embodiments of the present application can be provided as methods, systems, or computer program products. Therefore, the present application can adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment in combination with software and hardware. Moreover, the present application can adopt the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to magnetic disk storage, CD-ROM, optical storage, etc.) that contain computer-usable program code.

[0150] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the present application. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 a process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0151] These computer program instructions may also be stored in a computer readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 a process or multiple processes and / or boxes Figure 1 The function specified in one or more boxes.

[0152] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operational steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing the instructions executed on the computer or other programmable device for implementing the process. Figure 1 a process or multiple processes and / or boxes Figure 1 The steps of the functions specified in one or more blocks. Obviously, those skilled in the art may make various changes and modifications to this application without departing from the spirit and scope of this application. Thus, if such modifications and variations of this application fall within the scope of the claims of this application and their equivalents, this application is intended to include such modifications and variations.

Claims

1. A method for determining a motion trajectory, characterized in that: The method comprises: Inputting the acquired first overhead video frame into a pre-trained position recognition model to acquire the region where the first athlete is located and the corresponding first identification information contained in the first overhead video frame, and inputting the second overhead video frame into the pre-trained position recognition model to acquire the region where the second athlete is located and the corresponding second identification information contained in the second overhead video frame, wherein the second overhead video frame is a overhead video frame captured at a time adjacent to that of the first overhead video frame; For the first player corresponding to the first identification information, input the first overhead video frame and the second overhead video frame into a pre-trained Siamese network model to determine which second players included in the second overhead video frame are identical to the first player; determining a motion trajectory of the first athlete based on an area where the first athlete is located in the first overhead video frame and an area where the same athlete as the first athlete is located in the second overhead video frame; the total range of the sports field included in the first overhead video frame and the second overhead video frame includes the entire sports field; The determining of the motion trajectory of the first athlete based on the area where the first athlete is located in the first overhead video frame and the area where the same athlete as the first athlete is located in the second overhead video frame includes: determining a first actual area in the actual sports field where the first area is located based on a first area in the first overhead video frame where the first athlete is located and a pre-stored scale between pixel sizes in the overhead video frame and sizes of the actual sports field; determining a second actual area in the actual sports field where the second area is located based on a second area in the second overhead video frame where the athlete identical to the first athlete is located and the scale; The first actual area and the second actual area are respectively mapped into the image of the actual sports field to generate a motion trajectory of the first player.

2. The method according to claim 1, characterized in that After obtaining the area where the second player is located and the corresponding second identification information contained in the second overhead video frame, and before inputting the first overhead video frame and the second overhead video frame into the pre-trained Siamese network model for the first identification information corresponding to the first player, the method further includes: Determine whether the distance between the first player and at least two second players included in the second overhead video frame is not greater than a set first distance threshold; if so, proceed to subsequent steps.

3. The method according to claim 1, characterized in that After obtaining the area where the second player is located and the corresponding second identification information contained in the second overhead video frame, the method further includes: It is determined whether there is a target second player in the second overhead video frame whose distance from the first player is not greater than a set first distance threshold. If yes, the target second player is determined to be the same player as the first player.

4. The method according to claim 1, wherein The method further comprises: Inputting the acquired video frame into a pre-trained position recognition model to acquire an area where a third athlete is located contained in the video frame; and identifying identity information of the third athlete contained in the video frame; wherein the video frame and the first overhead video frame are shot at the same time; Determine whether there is a target third athlete in the video frame whose distance from the first athlete is not greater than a set second distance threshold; if so, determine the target third athlete as the same athlete as the first athlete; and update the first identification information corresponding to the first athlete based on the identity information of the target third athlete.

5. The method according to claim 4, characterized in that The identifying the identity information of the third player included in the video frame includes: Determining identity information of a third athlete contained in the video frame based on face recognition; or Determine the uniform number and uniform color of the third athlete contained in the video frame; and determine the identity information of the third athlete based on the stored correspondence between the uniform color, uniform number and the athlete's identity information.

6. The method according to claim 5, characterized in that Determining the uniform number and uniform color of the third athlete included in the video frame includes: The video frame is input into a pre-trained team uniform color and number recognition model to obtain the team uniform color and team uniform number of the third athlete contained in the video frame.

7. A display device, characterized in that: The display device comprises: A display, wherein the display is used to display a motion trajectory diagram containing an athlete; A controller, wherein the controller is configured to: execute the steps of the motion trajectory determination method according to any one of claims 1 to 6.

8. An electronic device, characterized in that: The electronic device includes at least a processor and a memory, and the processor is configured to implement the steps of the motion trajectory determination method according to any one of claims 1 to 6 when executing a computer program stored in the memory.

9. A computer-readable storage medium, characterized in that It stores a computer program, which, when executed by a processor, implements the steps of the motion trajectory determination method according to any one of claims 1 to 6.

Images