Farm livestock data acquisition method, device and equipment and storage medium

Abstract

The invention discloses a farm livestock data acquisition method, device and equipment and a storage medium, and belongs to the technical field of artificial intelligence. Firstly, livestock in a farm is shot through a first unmanned aerial vehicle, and a video frame sequence is obtained; and identifying the position of the livestock in each frame by using a target detection algorithm, and determining the moving path of the livestock. Then, marking paths on the video frames, and training the initial prediction model by using the path data to obtain a livestock moving path prediction model; then, the first unmanned aerial vehicle shoots a current livestock position map according to the instruction and inputs the current livestock position map into the prediction model, and a predicted moving path of the livestock is generated; and finally, the second unmanned aerial vehicle performs shooting according to the predicted path to obtain a second video frame sequence. The invention also relates to the technical field of block chains, and the video frame sequence is stored in a block chain network. According to the method, the image acquisition efficiency is improved, shooting instability caused by cattle body movement is reduced through automatic tracking and prediction, and the quality of cattle face image acquisition is improved.

Description

Technical Field

[0001] The present application belongs to the field of artificial intelligence technology, and specifically relates to a method, device, equipment and storage medium for collecting livestock data on a breeding farm. Background Art

[0002] Farm livestock data collection refers to the process of collecting and recording livestock-related data generated during the breeding process through different means and methods. This process is an important part of the modernization of the breeding industry and is of great significance for improving breeding efficiency and promoting the advancement of breeding technology.

[0003] Taking the collection of cow face images as an example, by collecting cow face images, a unique biometric digital identity database for each cow can be established, providing strong support for breeding management, disease prevention and control, insurance claims, live loans, etc. In addition, cow face image collection can also help improve breeding efficiency, reduce labor costs, and promote the development of the breeding industry towards modernization and intelligence.

[0004] Currently, the collection of cow face images relies on manual photography, which requires farmers to manually operate cameras or mobile phones to take pictures of each cow one by one. However, this manual photography method is inefficient and easily affected by human factors. At the same time, the body movement of the cow will also cause the image to be blurred or deformed, resulting in poor quality of the collected cow face images. Summary of the invention

[0005] The purpose of the embodiments of the present application is to propose a method, device, computer equipment and storage medium for collecting livestock data on a farm, so as to solve the technical problems that the existing cattle face image collection scheme using manual photography is inefficient and easily affected by human factors and factors of cattle body movement, resulting in poor results of the collected cattle face images.

[0006] In order to solve the above technical problems, the present application embodiment provides a method for collecting livestock data in a breeding farm, which adopts the following technical solution:

[0007] A method for collecting livestock data in a breeding farm is applied to an unmanned aerial vehicle (UAV) data collection platform, wherein the UAV data collection platform is configured with a plurality of unmanned aerial vehicles (UAVs), and the method comprises:

[0008] Acquire a first video frame sequence shot by a first drone targeting livestock in a farm;

[0009] In the first video frame sequence, identifying the position information of the livestock in each video frame according to a preset target detection algorithm;

[0010] Based on the position information of the livestock in each video frame, determine the livestock movement path, and mark the livestock movement path in the first video frame sequence;

[0011] Using the first video frame sequence carrying the path mark to train a preset initial prediction model to obtain a moving path prediction model;

[0012] receiving a data collection instruction, instructing the first drone to take photos of livestock in the farm to obtain a current livestock location map;

[0013] Inputting the current livestock position map into the movement path prediction model to obtain the predicted movement path of the livestock;

[0014] The second UAV is instructed to shoot the livestock according to the predicted moving path to obtain a second video frame sequence.

[0015] Furthermore, in the first video frame sequence, before the step of identifying the position information of the livestock in each video frame according to a preset target detection algorithm, the method further includes:

[0016] Obtain a historical video frame sequence, wherein the historical video frame sequence is a video frame sequence carrying a target position label;

[0017] Importing the historical video frame sequence into a preset initial target detection model, wherein the initial target detection model is built based on a target detection algorithm;

[0018] Obtaining an initial detection result output by the initial target detection model, wherein the initial detection result is the position coordinate of the target in each historical video frame predicted by the initial target detection model;

[0019] Compare the initial detection result with the target position label to obtain the position prediction error;

[0020] Based on the position prediction error, the initial target detection model is iterated using the back propagation algorithm until the model is fitted to obtain the target detection model.

[0021] Furthermore, in the first video frame sequence, the step of identifying the position information of the livestock in each video frame according to a preset target detection algorithm specifically includes:

[0022] identifying a location of livestock in each video frame in the first sequence of video frames using an object detection model;

[0023] Get the location information of livestock in each video frame output by the target detection model.

[0024] Furthermore, based on the position information of the livestock in each video frame, the step of determining the livestock movement path and marking the livestock movement path in the first video frame sequence specifically includes:

[0025] Based on the location information of the livestock in each video frame, a target tracker is assigned to each video frame;

[0026] Starting from the first video frame of the first video frame sequence to the last video frame of the first video frame sequence, each target tracker is started in sequence to track the target of the livestock in the first video frame sequence;

[0027] When the target tracker of the last video frame completes tracking, the livestock movement path is generated;

[0028] The livestock movement paths are marked in the first video frame sequence.

[0029] Furthermore, the initial prediction model is built based on the long short-term memory network architecture, and the steps of training the preset initial prediction model using the first video frame sequence carrying the path mark to obtain the moving path prediction model specifically include:

[0030] Extracting features from the first video frame sequence to obtain video frame sequence features;

[0031] Based on the characteristics of video frame sequences, the temporal dependency and spatial relationship between video frames are captured through long short-term memory networks;

[0032] Based on the temporal dependency and spatial relationship between the video frames, predict the movement path of the livestock in the first video frame sequence to obtain the historical predicted movement path of the livestock;

[0033] Compare the path mark with the historical predicted moving path to obtain the path prediction error;

[0034] Based on the path prediction error, the initial prediction model is iterated using the back propagation algorithm until the model is fitted to obtain the mobile path prediction model.

[0035] Furthermore, the step of instructing the second drone to shoot the livestock according to the predicted moving path to obtain a second video frame sequence specifically includes:

[0036] Acquire a photographed image output by the second UAV during the process of photographing the livestock according to the predicted moving path to obtain an initial photographed image;

[0037] Performing image quality evaluation on the initial captured image to obtain image quality parameters;

[0038] Screening the initial captured image according to the image quality parameter to obtain a screened image;

[0039] A second sequence of video frames is constructed based on the filtered image.

[0040] Furthermore, the step of evaluating the image quality of the initial captured image to obtain image quality parameters specifically includes:

[0041] Acquiring image parameters of the initial captured image, wherein the image parameters include resolution, contrast, color saturation, and noise level;

[0042] Calculate the quality evaluation weight of each image parameter;

[0043] Based on the quality evaluation weight, each image parameter is weighted and summed to obtain a comprehensive evaluation index;

[0044] The comprehensive evaluation index is used as the image quality parameter of the initial captured image.

[0045] In order to solve the above technical problems, the present application also provides a livestock data collection device for a breeding farm, which adopts the following technical solution:

[0046] A livestock data collection device for a breeding farm, the device runs on an unmanned aerial vehicle data collection platform, the unmanned aerial vehicle data collection platform is equipped with a plurality of unmanned aerial vehicles, and the device comprises:

[0047] A first shooting module is used to obtain a first video frame sequence shot by a first drone targeting livestock in a breeding farm;

[0048] A target detection module, used to identify the position information of livestock in each video frame in the first video frame sequence according to a preset target detection algorithm;

[0049] A path marking module, used to determine the livestock movement path based on the location information of the livestock in each video frame, and mark the livestock movement path in the first video frame sequence;

[0050] A model training module, used to train a preset initial prediction model using a first video frame sequence carrying a path mark to obtain a moving path prediction model;

[0051] The second shooting module is used to receive the data collection instruction and instruct the first drone to shoot the livestock in the farm to obtain the current livestock position map;

[0052] A path prediction module, used to input the current livestock position map into the movement path prediction model to obtain the predicted movement path of the livestock;

[0053] The third shooting module is used to instruct the second UAV to shoot the livestock according to the predicted moving path to obtain a second video frame sequence.

[0054] In order to solve the above technical problems, the embodiment of the present application further provides a computer device, which adopts the following technical solution:

[0055] A computer device comprises a memory and a processor, wherein the memory stores computer-readable instructions, and when the processor executes the computer-readable instructions, the steps of the method for collecting livestock data on a farm as described in any one of the above items are implemented.

[0056] In order to solve the above technical problems, the embodiment of the present application further provides a computer-readable storage medium, which adopts the following technical solution:

[0057] A computer-readable storage medium having computer-readable instructions stored thereon, wherein the computer-readable instructions, when executed by a processor, implement the steps of any one of the above-mentioned methods for collecting livestock data on a farm.

[0058] Compared with the prior art, the embodiments of the present application have the following beneficial effects:

[0059] The present application discloses a method, device, equipment and storage medium for collecting livestock data in a breeding farm, and belongs to the field of artificial intelligence technology. The present application is applied to a drone data collection platform, which is equipped with several drones. First, the first drone is used to shoot the livestock in the breeding farm to obtain a video frame sequence; the target detection algorithm is used to identify the position of the livestock in each frame and determine its movement path. Then, the path is marked on the video frame, and the initial prediction model is trained using these path data to obtain a livestock movement path prediction model. Subsequently, the first drone shoots the current livestock position map according to the instruction, and inputs it into the prediction model to generate the predicted movement path of the livestock. Finally, the second drone shoots according to the predicted path to obtain a second video frame sequence. The present application obtains a video frame sequence of livestock in a breeding farm through a drone, automatically identifies the livestock position in each video frame using a target detection algorithm, and predicts the movement trajectory of the livestock through a path prediction model, accurately guides the second drone to perform further shooting, reduces manual intervention, improves image acquisition efficiency, and reduces the shooting instability caused by the movement of the cattle body through automatic tracking and prediction, thereby improving the quality of the collected cattle face image. BRIEF DESCRIPTION OF THE DRAWINGS

[0060] In order to more clearly illustrate the scheme in the present application, a brief introduction is given below to the drawings required for use in the description of the embodiments of the present application. Obviously, the drawings described below are some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0061] Figure 1 An exemplary system architecture diagram is shown in which the present application can be applied;

[0062] Figure 2A flow chart showing an embodiment of a method for collecting livestock data in a breeding farm according to the present application is shown;

[0063] Figure 3 A schematic structural diagram of an embodiment of a livestock data collection device for a breeding farm according to the present application is shown;

[0064] Figure 4 A schematic structural diagram of an embodiment of a computer device according to the present application is shown. DETAILED DESCRIPTION

[0065] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as those commonly understood by technicians in the technical field of the present application; the terms used in the specification of the application herein are only for the purpose of describing specific embodiments and are not intended to limit the present application; the terms "including" and "having" and any variations thereof in the specification and claims of the present application and the above-mentioned drawings are intended to cover non-exclusive inclusions. The terms "first", "second", etc. in the specification and claims of the present application or the above-mentioned drawings are used to distinguish different objects, not to describe a specific order.

[0066] Reference to "embodiments" herein means that a particular feature, structure, or characteristic described in conjunction with the embodiments may be included in at least one embodiment of the present application. The appearance of the phrase in various locations in the specification does not necessarily refer to the same embodiment, nor is it an independent or alternative embodiment that is mutually exclusive with other embodiments. It is explicitly and implicitly understood by those skilled in the art that the embodiments described herein may be combined with other embodiments.

[0067] In order to enable those skilled in the art to better understand the solution of the present application, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the accompanying drawings.

[0068] like Figure 1 As shown, the system architecture 100 may include a terminal device 101, a network 102 and a server 103. The terminal device 101 may be a laptop 1011, a tablet computer 1012 or a mobile phone 1013. The network 102 is used to provide a medium for a communication link between the terminal device 101 and the server 103. The network 102 may include various connection types, such as wired, wireless communication links or optical fiber cables.

[0069] The user can use the terminal device 101 to interact with the server 103 through the network 102 to receive or send messages, etc. Various communication client applications can be installed on the terminal device 101, such as web browser applications, shopping applications, search applications, instant messaging tools, email clients, social platform software, etc.

[0070] The terminal device 101 can be various electronic devices with a display screen and supporting web browsing. In addition to the laptop computer 1011, the tablet computer 1012 or the mobile phone 1013, the terminal device 101 can also be an e-book reader, an MP3 player (Moving Picture Experts Group Audio Layer III), an MP4 (Moving Picture Experts Group Audio Layer IV) player, a laptop computer and a desktop computer, etc.

[0071] The server 103 may be a server that provides various services, such as a background server that provides support for web pages displayed on the terminal device 101 .

[0072] It should be noted that the livestock data collection method for a breeding farm provided in the embodiment of the present application is generally executed by a server / terminal device, and accordingly, the livestock data collection device for a breeding farm is generally arranged in the server / terminal device.

[0073] It should be understood that Figure 1 The numbers of terminal devices, networks and servers in the embodiment are merely illustrative. The above system may have any number of terminal devices, networks and servers according to implementation requirements.

[0074] Continue to refer Figure 2 , shows a flow chart of an embodiment of a method for collecting livestock data on a farm according to the present application. The livestock data collection method of the present application is applied to a drone data collection platform, and the drone data collection platform is configured with several drones, and the method includes:

[0075] S201, obtaining a first video frame sequence shot by a first drone targeting livestock in a farm.

[0076] Specifically, the system controls the first drone to cover the key areas of the farm through a pre-set flight path or automatic navigation system, and automatically shoots herds of cattle, sheep or other livestock. The solution of this application is specifically for shooting herds of cattle in the farm to obtain images of cattle faces. The drone's camera can record high-resolution videos and supports night infrared shooting to ensure all-weather operation. During the shooting process, the drone will maintain stable flight to avoid shaking caused by wind speed or obstacles, ensuring clear images.

[0077] By using drone photography to replace traditional manual operations, the reliance on manual operations is reduced, while also improving the efficiency of data collection in large-scale farms.

[0078] S202: In a first video frame sequence, identifying the position information of livestock in each video frame according to a preset target detection algorithm.

[0079] Specifically, the system inputs the first video frame sequence acquired from the drone into a preset target detection algorithm, such as YOLOv5, Faster R-CNN or other deep learning models. The algorithm can automatically identify the livestock in each video frame, mark the livestock, and extract the location coordinate information of the livestock.

[0080] The target detection algorithm outputs the location data of all livestock in each frame of the image. These data will be used for trajectory calculation and path marking in subsequent steps. Automated detection greatly reduces the errors and workload of manual identification.

[0081] It should be noted that in order to improve the robustness of detection, the algorithm needs to consider different environmental factors, such as lighting changes, occlusion, camera angle, etc., and implement corresponding measures. For example, lighting normalization processing methods can be used to deal with lighting changes.

[0082] S203, determining a livestock movement path based on the position information of the livestock in each video frame, and marking the livestock movement path in the first video frame sequence.

[0083] Specifically, after identifying the position of the livestock in each frame, the system calculates the movement path of the livestock by analyzing the position changes in these consecutive frames. This can be achieved through tracking algorithms, such as Kahlman filtering, SORT (Simple Online and Realtime Tracking) or Deep SORT, which are multi-target tracking algorithms. These algorithms predict the position in the next frame based on the position coordinates in consecutive frames, and match these predicted positions with the actual detected positions to form a smooth movement trajectory. The system then visually marks the movement path of each livestock in the video frame, such as using different colored lines to represent the trajectories of different livestock.

[0084] This not only can visually display the movement trajectory of livestock, but also provide clear labeling information for subsequent model training. Through automated path analysis, it improves the accuracy and efficiency of data processing.

[0085] S204: Use the first video frame sequence carrying the path mark to train a preset initial prediction model to obtain a moving path prediction model.

[0086] Specifically, the system uses the first video frame sequence carrying the livestock movement path mark to train an initial path prediction model. This prediction model uses deep learning-based techniques, such as recurrent neural networks (RNN), long short-term memory networks (LSTM), or temporal convolutional networks (TCN) to learn the movement patterns of livestock. By inputting historical path data, the model gradually learns the movement patterns of livestock and can predict future movement trends.

[0087] The mobile path prediction model is used in drone mission planning to achieve more intelligent and automated management and monitoring of livestock.

[0088] In order to improve the prediction accuracy of the model, when there are few training samples, data enhancement technology can be used to expand the samples during training to increase the diversity of data, and cross-validation and other methods can be used to prevent overfitting. After the training is completed, the obtained movement path prediction model can be used to predict the movement trajectory of livestock in real time.

[0089] S205, receiving a data collection instruction, instructing the first drone to take photos of the livestock in the farm to obtain a current livestock location map.

[0090] Specifically, after receiving a new data collection instruction, the system will instruct the first drone to take off again and go to the designated area in the farm to shoot. The purpose of this shooting is to obtain the current livestock location map, that is, the latest livestock distribution. When flying, the drone will cover key areas in the farm according to the previous flight path and shooting plan, and record the latest location of the livestock with high accuracy. After the shooting is completed, the drone will automatically return and upload the video data, from which the system extracts key frames and generates a static location map of the current livestock. This location map will be used as input data, combined with the previously trained path prediction model, to predict the future movement path of the livestock.

[0091] Continue to use drones for tracking and shooting to achieve a fully automated shooting process without human intervention, ensuring the efficiency and consistency of data collection.

[0092] S206, inputting the current livestock position map into a movement path prediction model to obtain a predicted movement path of the livestock.

[0093] Specifically, the system will input the current livestock location map into the previously trained movement path prediction model. The model will predict the movement trajectory of livestock in the future based on historical data and current location information. This prediction is based on previously learned movement patterns and current environmental conditions, and can effectively infer the livestock's next possible movement path. The model may generate multiple possible movement paths and assign a probability to each path to indicate the possibility of the path. The system will select the optimal path based on these prediction results as the basis for subsequent drone photography.

[0094] By using artificial intelligence technology to make predictions, the system can have a certain degree of foresight, adjust the flight plan of the drone in advance, and achieve dynamic tracking and management of livestock.

[0095] S207, instructing the second drone to shoot the livestock according to the predicted moving path to obtain a second video frame sequence.

[0096] Specifically, after obtaining the predicted movement path of the livestock, the system will instruct the second drone to shoot according to the predicted path. The second drone will plan the flight route in advance based on the path provided by the prediction model, and maintain continuous shooting of the target livestock to ensure that the key behaviors and status of the livestock can be captured. Unlike the mission of the first drone, the mission of the second drone is more targeted tracking shooting, which may focus on specific individuals or groups to ensure all-round monitoring of their behavior.

[0097] Through this precise shooting based on predicted paths, the drone data collection platform can more efficiently obtain dynamic data of livestock and generate more accurate and detailed second video frame sequences to obtain high-quality livestock images.

[0098] In the above embodiments, the present application solves the problem of low efficiency and influence of human and cattle movement factors in manually shooting cow face images through automatic shooting by drones and intelligent algorithms. The video frame sequence of livestock in the farm is obtained by drones, the position of livestock in each video frame is automatically identified using the target detection algorithm, and the movement trajectory of livestock is predicted by the path prediction model, so as to accurately guide the second drone for further shooting. This process reduces manual intervention, improves image acquisition efficiency, and reduces shooting instability caused by cattle movement through automatic tracking and prediction, thereby improving the quality of the collected cow face images.

[0099] Furthermore, in the first video frame sequence, before the step of identifying the position information of the livestock in each video frame according to a preset target detection algorithm, the method further includes:

[0100] Obtain a historical video frame sequence, wherein the historical video frame sequence is a video frame sequence carrying a target position label;

[0101] Importing the historical video frame sequence into a preset initial target detection model, wherein the initial target detection model is built based on a target detection algorithm;

[0102] Obtaining an initial detection result output by the initial target detection model, wherein the initial detection result is the position coordinate of the target in each historical video frame predicted by the initial target detection model;

[0103] Compare the initial detection result with the target position label to obtain the position prediction error;

[0104] Based on the position prediction error, the initial target detection model is iterated using the back propagation algorithm until the model is fitted to obtain the target detection model.

[0105] In this embodiment, the system first provides basic data for the target detection model by obtaining a sequence of historical video frames with target position labels. These historical video frame sequences have been annotated and contain the precise position of the target (such as livestock) in each frame. Next, these labeled data will be imported into a preset initial target detection model, which is built based on existing target detection algorithms (such as YOLO, Faster R-CNN, etc.). The initial target detection model predicts the input historical video frames and outputs the position coordinates of the target in each frame. These prediction results are called initial detection results. Subsequently, the system compares the initial detection results of the model with the true labels in the video frames to calculate the position prediction error. The error calculation process can be performed by distance measurement or intersection over union (IoU). In order to further improve the prediction accuracy of the model, the system will iteratively train the model using a back propagation algorithm and adjust the weight parameters of the model to reduce the position prediction error. This process will continue until the model converges and the error reaches an acceptable range, and finally an optimized target detection model is obtained. Through this step-by-step optimization method, the system can achieve high-precision detection of the target position and provide reliable model support for future real-time detection.

[0106] Through the above steps, the accuracy and generalization ability of the target detection model are effectively improved. Through repeated iterative optimization, the position prediction error is reduced, so that the model can more accurately detect the target position in the video frame.

[0107] Furthermore, in the first video frame sequence, the step of identifying the position information of the livestock in each video frame according to a preset target detection algorithm specifically includes:

[0108] identifying a location of livestock in each video frame in the first sequence of video frames using an object detection model;

[0109] Get the location information of livestock in each video frame output by the target detection model.

[0110] In this embodiment, the system first processes each frame in the first video frame sequence using a trained object detection model, which has been trained and optimized using a large amount of historical data and can accurately identify the location of livestock in the frame.

[0111] Taking the target detection model trained based on the deep convolutional network structure as an example, in actual operation, the video frames are input into the target detection model frame by frame, and the model extracts features in the image through its deep convolutional network structure to identify and locate livestock. The core steps in the detection process include feature extraction, candidate region generation, classification, and regression. The model outputs the location information of all detected livestock in each frame, which is usually expressed in the form of a bounding box, that is, the coordinates of each livestock (such as the upper left and lower right pixel points).

[0112] Through the above steps, the target detection model is used to realize automatic identification of livestock locations, which greatly improves the efficiency and accuracy of livestock monitoring. It not only solves the problem of low efficiency of traditional manual monitoring, but also maintains a high recognition accuracy rate in complex environments.

[0113] Furthermore, based on the position information of the livestock in each video frame, the step of determining the livestock movement path and marking the livestock movement path in the first video frame sequence specifically includes:

[0114] Based on the location information of the livestock in each video frame, a target tracker is assigned to each video frame;

[0115] Starting from the first video frame of the first video frame sequence to the last video frame of the first video frame sequence, each target tracker is started in sequence to track the target of the livestock in the first video frame sequence;

[0116] When the target tracker of the last video frame completes tracking, the livestock movement path is generated;

[0117] The livestock movement paths are marked in the first video frame sequence.

[0118] In this embodiment, the system first assigns a target tracker to each video frame based on the position information of the livestock in each video frame. These trackers usually use target tracking algorithms such as KCF (Kernelized Correlation Filters), MOSSE (Minimum Output Sum of Squared Error), DeepSORT (Deep Simple Online Real-time Tracking), etc. Each tracker will track the movement of the target livestock in subsequent video frames based on the detected initial position information. The tracking process starts from the first video frame of the first video frame sequence, starts the target tracker frame by frame, and updates the livestock's position information in real time according to the changes between frames until the last video frame is processed. Each tracker generates the livestock's movement path based on the livestock's trajectory information, and continuously accumulates the trajectory changes between frames to finally generate a complete movement path.

[0119] Through the above steps, by using the target tracker, the system can maintain continuous tracking of the target livestock, avoiding the problem of detection interruption caused by changes in the livestock's position.

[0120] Furthermore, the initial prediction model is built based on the long short-term memory network architecture, and the steps of training the preset initial prediction model using the first video frame sequence carrying the path mark to obtain the moving path prediction model specifically include:

[0121] Extracting features from the first video frame sequence to obtain video frame sequence features;

[0122] Based on the characteristics of video frame sequences, the temporal dependency and spatial relationship between video frames are captured through long short-term memory networks;

[0123] Based on the temporal dependency and spatial relationship between the video frames, predict the movement path of the livestock in the first video frame sequence to obtain the historical predicted movement path of the livestock;

[0124] Compare the path mark with the historical predicted moving path to obtain the path prediction error;

[0125] Based on the path prediction error, the initial prediction model is iterated using the back propagation algorithm until the model is fitted to obtain the mobile path prediction model.

[0126] In this embodiment, the system first performs feature extraction on the first video frame sequence. The feature extraction process includes extracting visual information from each frame, such as pixel features, edge features, motion vectors, etc., which will be processed into feature vectors suitable for model input. These features reflect the spatial relationship of the video frames and record the state changes of each livestock at different time points. Then, the system uses a long short-term memory network (LSTM) to process these features. The LSTM network is a neural network structure that can capture the dependencies between long time series data and is particularly suitable for processing time series data. Here, LSTM is used to capture the temporal dependencies and spatial relationships between video frames in order to predict the movement path of livestock in the video sequence. The LSTM network gradually predicts the possible movement trajectory of livestock by inputting the features of consecutive frames, and generates a historical predicted movement path.

[0127] After that, the system compares the predicted path with the actual path mark and calculates the path prediction error. This error reflects the deviation of the model in path prediction. In order to optimize the model, the system will iteratively adjust the model based on the path prediction error using the back propagation algorithm. During each iteration, the model weights will be continuously updated based on the error feedback until the model output is highly consistent with the actual path. Finally, the system obtains an optimized model that can accurately predict the movement path of livestock.

[0128] Through the above steps, by using the LSTM network to capture spatiotemporal dependencies and combining the back-propagation algorithm for model optimization, the system can more accurately predict the movement trajectory of livestock, reduce path prediction errors, and effectively improve the accuracy of the movement path prediction model.

[0129] Furthermore, the step of instructing the second drone to shoot the livestock according to the predicted moving path to obtain a second video frame sequence specifically includes:

[0130] Acquire a photographed image output by the second UAV during the process of photographing the livestock according to the predicted moving path to obtain an initial photographed image;

[0131] Performing image quality evaluation on the initial captured image to obtain image quality parameters;

[0132] Screening the initial captured image according to the image quality parameter to obtain a screened image;

[0133] A second sequence of video frames is constructed based on the filtered image.

[0134] In this embodiment, the system first obtains the initial captured images output by the second drone during the process of photographing the livestock according to the predicted moving path. These images may include the results of shooting at multiple angles, lighting conditions and different time points, so the quality of the images may vary. In order to ensure that the images used for analysis have sufficient clarity and effective information, the system will evaluate the image quality of these initial captured images.

[0135] The image quality evaluation process includes the evaluation of image clarity, contrast, exposure, noise level and other parameters, which reflect whether the image is clear enough, whether it is affected by environmental factors, whether it is blurry or distorted, etc. After the evaluation is completed, the system will screen the initial captured images according to the preset image quality threshold, exclude those images that do not meet the quality standards, and only retain the screened images that meet the requirements.

[0136] In order to further improve the quality of the data, the screening criteria may also include whether the key livestock in the image is clearly visible, whether it appears completely in the picture, etc. Finally, based on the screened high-quality images, the system combines them into a second video frame sequence.

[0137] Through the above steps, low-quality images are excluded through the image quality evaluation and screening process, thereby improving the overall reliability and effectiveness of the second video frame sequence and ensuring the quality of image data used for analysis.

[0138] Furthermore, the step of evaluating the image quality of the initial captured image to obtain image quality parameters specifically includes:

[0139] Acquiring image parameters of the initial captured image, wherein the image parameters include resolution, contrast, color saturation, and noise level;

[0140] Calculate the quality evaluation weight of each image parameter;

[0141] Based on the quality evaluation weight, each image parameter is weighted and summed to obtain a comprehensive evaluation index;

[0142] The comprehensive evaluation index is used as the image quality parameter of the initial captured image.

[0143] In this embodiment, the system first obtains various image parameters of the initial captured image, including resolution, contrast, color saturation, and noise level. Resolution reflects the clarity of the image. The higher the resolution, the richer the image details. Contrast measures the distinction between bright and dark areas in the image. Good contrast helps to enhance the visual effect of the image. Color saturation reflects the vividness of the color in the image. Higher saturation makes the image more attractive. Noise level evaluates unnecessary interfering pixels in the image. Too high noise will affect the clarity and detail performance of the image.

[0144] In order to comprehensively evaluate the image quality, the system calculates the quality evaluation weight of each image parameter. The quality evaluation weight can be determined based on the needs of the actual application scenario. For example, in some cases, resolution may be the most critical factor, while in other cases, contrast or noise level may dominate. Then, the system weights and sums each image parameter based on these quality evaluation weights to generate a comprehensive evaluation index. This comprehensive evaluation index can fully reflect the overall quality performance of the image, and the system uses it as the image quality parameter of the initial captured image. The calculation formula for the comprehensive evaluation index Q is as follows:

[0145] Q=αR+βC+γS+δN

[0146] Where R, C, S, and N represent resolution, contrast, color saturation, and noise level respectively, and α, β, γ, and δ are the quality evaluation weights corresponding to R, C, S, and N respectively.

[0147] Through the above steps and the weighted method, the system can flexibly adjust the evaluation criteria according to the needs of different scenarios, thereby ensuring the best image quality for analysis and decision-making, and ensuring the comprehensiveness and accuracy of image quality evaluation.

[0148] In the above embodiment, the present application discloses a method for collecting livestock data in a farm, which belongs to the field of artificial intelligence technology. The present application is applied to a drone data collection platform, which is equipped with several drones. First, the livestock in the farm are photographed by the first drone to obtain a video frame sequence; the target detection algorithm is used to identify the position of the livestock in each frame and determine its movement path. Then, the path is marked on the video frame, and the initial prediction model is trained using these path data to obtain a livestock movement path prediction model. Subsequently, the first drone photographs the current livestock position map according to the instruction, and inputs it into the prediction model to generate a predicted movement path of the livestock. Finally, the second drone shoots according to the predicted path to obtain a second video frame sequence. The present application obtains a video frame sequence of livestock in a farm through a drone, automatically identifies the livestock position in each video frame using a target detection algorithm, and predicts the movement trajectory of the livestock through a path prediction model, accurately guides the second drone to perform further shooting, reduces manual intervention, improves image acquisition efficiency, and reduces the shooting instability caused by the movement of the cattle body through automatic tracking and prediction, thereby improving the quality of the collected cattle face image.

[0149] In this embodiment, the livestock data collection method of the farm is executed on the electronic device (eg Figure 1The server shown in the figure can receive instructions or obtain data through a wired connection or a wireless connection. It should be noted that the above wireless connection method may include but is not limited to 3G / 4G connection, Wi-Fi connection, Bluetooth connection, Wi MAX connection, Zigbee connection, UWB (Ultra Wi-Fi Band) connection, and other wireless connection methods currently known or to be developed in the future.

[0150] It should be emphasized that in order to further ensure the privacy and security of the above-mentioned video frame sequence information, the above-mentioned video frame sequence information can also be stored in a blockchain node.

[0151] The blockchain referred to in this application is a new application model of computer technologies such as distributed data storage, peer-to-peer transmission, consensus mechanism, encryption algorithm, etc. Blockchain is essentially a decentralized database, a string of data blocks generated by cryptographic methods. Each data block contains a batch of network transaction information, which is used to verify the validity of its information (anti-counterfeiting) and generate the next block. Blockchain can include the underlying blockchain platform, platform product service layer, and application service layer.

[0152] The embodiments of the present application can acquire and process relevant data based on artificial intelligence technology. Artificial Intelligence (AI) is the theory, method, technology and application system that uses digital computers or machines controlled by digital computers to simulate, extend and expand human intelligence, perceive the environment, acquire knowledge and use knowledge to obtain the best results.

[0153] AI basic technologies generally include sensors, dedicated AI chips, cloud computing, distributed storage, big data processing technology, operation / interaction systems, mechatronics, etc. AI software technologies mainly include computer vision technology, robotics technology, biometrics technology, speech processing technology, natural language processing technology, and machine learning / deep learning.

[0154] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiments can be implemented by instructing the relevant hardware through computer-readable instructions, and the computer-readable instructions can be stored in a computer-readable storage medium. When the program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, the aforementioned storage medium can be a non-volatile storage medium such as a disk, an optical disk, a read-only memory (ROM), or a random access memory (RAM).

[0155] It should be understood that, although the steps in the flowchart of the accompanying drawings are displayed in sequence as indicated by the arrows, these steps are not necessarily executed in sequence in the order indicated by the arrows. Unless otherwise specified herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least a part of the steps in the flowchart of the accompanying drawings may include multiple sub-steps or multiple stages, and these sub-steps or stages are not necessarily executed at the same time, but can be executed at different times, and their execution order is not necessarily sequential, but can be executed in turn or alternately with other steps or at least a part of the sub-steps or stages of other steps.

[0156] Further references Figure 3 , as a response to the above Figure 2 In order to realize the method shown in the figure, the present application provides an embodiment of a livestock data collection device for a breeding farm, and the device embodiment is Figure 2 Corresponding to the method embodiment shown, the device can be specifically applied to various electronic devices.

[0157] like Figure 3 As shown, the livestock data collection device 300 of the farm described in this embodiment runs on an unmanned aerial vehicle data collection platform, and the unmanned aerial vehicle data collection platform is equipped with a plurality of unmanned aerial vehicles. The livestock data collection device 300 of the farm includes:

[0158] The first shooting module 301 is used to obtain a first video frame sequence shot by a first drone targeting livestock in a breeding farm;

[0159] The target detection module 302 is used to identify the position information of the livestock in each video frame in the first video frame sequence according to a preset target detection algorithm;

[0160] A path marking module 303 is used to determine the livestock movement path based on the location information of the livestock in each video frame, and mark the livestock movement path in the first video frame sequence;

[0161] A model training module 304 is used to train a preset initial prediction model using a first video frame sequence carrying a path mark to obtain a moving path prediction model;

[0162] The second shooting module 305 is used to receive a data collection instruction and instruct the first drone to shoot livestock in the farm to obtain a current livestock position map;

[0163] A path prediction module 306 is used to input the current livestock position map into the movement path prediction model to obtain the predicted movement path of the livestock;

[0164] The third shooting module 307 is used to instruct the second drone to shoot the livestock according to the predicted moving path to obtain a second video frame sequence.

[0165] Furthermore, the farm livestock data collection device 300 also includes a target detection model training module, which is specifically used for:

[0166] Obtain a historical video frame sequence, wherein the historical video frame sequence is a video frame sequence carrying a target position label; import the historical video frame sequence into a preset initial target detection model, wherein the initial target detection model is built based on a target detection algorithm; obtain an initial detection result output by the initial target detection model, wherein the initial detection result is the position coordinate of the target in each historical video frame predicted by the initial target detection model; compare the initial detection result with the target position label to obtain a position prediction error; and iterate the initial target detection model using a back propagation algorithm based on the position prediction error until the model is fitted to obtain a target detection model.

[0167] Furthermore, the target detection module 302 is specifically used for:

[0168] The target detection model is used to identify the position of the livestock in each video frame in the first video frame sequence; and the position information of the livestock in each video frame output by the target detection model is obtained.

[0169] Furthermore, the path marking module 303 is specifically used for:

[0170] Based on the position information of the livestock in each video frame, a target tracker is assigned to each video frame; starting from the first video frame of the first video frame sequence to the last video frame of the first video frame sequence, each target tracker is started in sequence to track the livestock in the first video frame sequence; when the target tracker of the last video frame completes tracking, a livestock movement path is generated; and the livestock movement path is marked in the first video frame sequence.

[0171] Furthermore, the initial prediction model is built based on the long short-term memory network architecture, and the model training module 304 is specifically used for:

[0172] The first video frame sequence is subjected to feature extraction to obtain features of the video frame sequence; based on the features of the video frame sequence, the time dependency and spatial relationship between the video frames are captured through a long short-term memory network; based on the time dependency and spatial relationship between the video frames, the movement path of the livestock in the first video frame sequence is predicted to obtain the historical predicted movement path of the livestock; the path mark and the historical predicted movement path are compared to obtain the path prediction error; based on the path prediction error, the initial prediction model is iterated using a back propagation algorithm until the model is fitted to obtain a movement path prediction model.

[0173] Furthermore, the third shooting module 307 is specifically used for:

[0174] Acquire captured images output by the second UAV during the process of photographing livestock according to the predicted moving path to obtain an initial captured image; perform image quality evaluation on the initial captured image to obtain image quality parameters; filter the initial captured image according to the image quality parameters to obtain a filtered image; and construct a second video frame sequence based on the filtered image.

[0175] Furthermore, the third shooting module 307 is further used for:

[0176] Obtain image parameters of the initial captured image, wherein the image parameters include resolution, contrast, color saturation and noise level; calculate the quality evaluation weight of each image parameter; perform weighted summation on each image parameter based on the quality evaluation weight to obtain a comprehensive evaluation index; and use the comprehensive evaluation index as the image quality parameter of the initial captured image.

[0177] In the above embodiment, the present application discloses a livestock data collection device for a farm, which belongs to the field of artificial intelligence technology. The present application is applied to a drone data collection platform, which is equipped with several drones. First, the livestock in the farm are photographed by the first drone to obtain a video frame sequence; the target detection algorithm is used to identify the position of the livestock in each frame and determine its movement path. Then, the path is marked on the video frame, and the initial prediction model is trained using these path data to obtain a livestock movement path prediction model. Subsequently, the first drone photographs the current livestock position map according to the instruction, and inputs it into the prediction model to generate a predicted movement path of the livestock. Finally, the second drone shoots according to the predicted path to obtain a second video frame sequence. The present application obtains a video frame sequence of livestock in a farm through a drone, automatically identifies the livestock position in each video frame using a target detection algorithm, and predicts the movement trajectory of the livestock through a path prediction model, accurately guides the second drone to perform further shooting, reduces manual intervention, improves image acquisition efficiency, and reduces the shooting instability caused by the movement of the cattle body through automatic tracking and prediction, thereby improving the quality of the collected cattle face image.

[0178] To solve the above technical problems, the present application also provides a computer device. Figure 4 , Figure 4 This is a basic structural block diagram of the computer device in this embodiment.

[0179] The computer device 4 includes a memory 41, a processor 42, and a network interface 43 that are interconnected and communicated through a system bus. It should be noted that the figure only shows a computer device 4 having a memory 41, a processor 42, and a network interface 43, but it should be understood that it is not required to implement all the components shown, and more or fewer components can be implemented instead. Among them, those skilled in the art can understand that the computer device here is a device that can automatically perform numerical calculations and / or information processing according to pre-set or stored instructions, and its hardware includes but is not limited to a microprocessor, an application specific integrated circuit (ASIC), a field-programmable gate array (FPGA), a digital processor (DSP), an embedded device, etc.

[0180] The computer device may be a computing device such as a desktop computer, a notebook, a PDA, a cloud server, etc. The computer device may interact with a user through a keyboard, a mouse, a remote controller, a touch pad, or a voice control device.

[0181] The memory 41 includes at least one type of readable storage medium, and the readable storage medium includes flash memory, hard disk, multimedia card, card-type memory (for example, SD or DX memory, etc.), random access memory (RAM), static random access memory (SRAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), programmable read-only memory (PROM), magnetic memory, disk, optical disk, etc. In some embodiments, the memory 41 can be an internal storage unit of the computer device 4, such as a hard disk or memory of the computer device 4. In other embodiments, the memory 41 can also be an external storage device of the computer device 4, such as a plug-in hard disk, a smart memory card (Smart Med ia Card, SMC), a secure digital (Secure Digital, SD) card, a flash card (Flash Card), etc. equipped on the computer device 4. Of course, the memory 41 can also include both the internal storage unit of the computer device 4 and its external storage device. In this embodiment, the memory 41 is generally used to store the operating system and various application software installed on the computer device 4, such as computer-readable instructions of a livestock data collection method in a farm, etc. In addition, the memory 41 can also be used to temporarily store various types of data that have been output or are to be output.

[0182] The processor 42 may be a central processing unit (CPU), a controller, a microcontroller, a microprocessor, or other data processing chips in some embodiments. The processor 42 is generally used to control the overall operation of the computer device 4. In this embodiment, the processor 42 is used to run computer-readable instructions or process data stored in the memory 41, such as computer-readable instructions for running the farm livestock data collection method.

[0183] The network interface 43 may include a wireless network interface or a wired network interface. The network interface 43 is generally used to establish a communication connection between the computer device 4 and other electronic devices.

[0184] In the above embodiment, the present application discloses a computer device, which belongs to the field of artificial intelligence technology. The present application is applied to a drone data acquisition platform, which is equipped with several drones. First, the livestock in the farm are photographed by the first drone to obtain a video frame sequence; the target detection algorithm is used to identify the position of the livestock in each frame and determine its movement path. Then, the path is marked on the video frame, and the initial prediction model is trained using these path data to obtain a livestock movement path prediction model. Subsequently, the first drone photographs the current livestock position map according to the instruction, and inputs it into the prediction model to generate the predicted movement path of the livestock. Finally, the second drone is photographed according to the predicted path to obtain a second video frame sequence. The present application obtains a video frame sequence of livestock in the farm through a drone, automatically identifies the livestock position in each video frame using a target detection algorithm, and predicts the movement trajectory of the livestock through a path prediction model, accurately guides the second drone to perform further shooting, reduces manual intervention, improves image acquisition efficiency, and reduces the shooting instability caused by the movement of the cattle body through automatic tracking and prediction, thereby improving the quality of the collected cattle face image.

[0185] The present application also provides another embodiment, namely, providing a computer-readable storage medium, wherein the computer-readable storage medium stores computer-readable instructions, and the computer-readable instructions can be executed by at least one processor to enable the at least one processor to perform the steps of the above-mentioned farm livestock data collection method.

[0186] In the above embodiment, the present application discloses a computer-readable storage medium, which belongs to the field of artificial intelligence technology. The present application is applied to a drone data acquisition platform, which is equipped with several drones. First, the livestock in the farm are photographed by the first drone to obtain a video frame sequence; the target detection algorithm is used to identify the position of the livestock in each frame and determine its movement path. Then, the path is marked on the video frame, and the initial prediction model is trained using these path data to obtain a livestock movement path prediction model. Subsequently, the first drone photographs the current livestock position map according to the instruction, and inputs it into the prediction model to generate a predicted movement path of the livestock. Finally, the second drone shoots according to the predicted path to obtain a second video frame sequence. The present application obtains a video frame sequence of livestock in the farm through a drone, automatically identifies the livestock position in each video frame using a target detection algorithm, and predicts the movement trajectory of the livestock through a path prediction model, accurately guides the second drone to perform further shooting, reduces manual intervention, improves image acquisition efficiency, and reduces the shooting instability caused by the movement of the cattle body through automatic tracking and prediction, thereby improving the quality of the collected cattle face image.

[0187] Through the description of the above implementation methods, those skilled in the art can clearly understand that the above-mentioned embodiment methods can be implemented by means of software plus a necessary general hardware platform, and of course by hardware, but in many cases the former is a better implementation method. Based on such an understanding, the technical solution of the present application, or the part that contributes to the prior art, can be embodied in the form of a software product, which is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk), and includes a number of instructions for a terminal device (which can be a mobile phone, computer, server, air conditioner, or network device, etc.) to execute the methods described in each embodiment of the present application.

[0188] The present application can be used in many general or special computer system environments or configurations. For example: personal computers, server computers, handheld or portable devices, tablet devices, multiprocessor systems, microprocessor-based systems, set-top boxes, programmable consumer electronic devices, network PCs, minicomputers, mainframe computers, distributed computing environments including any of the above systems or devices, etc. The present application can be described in the general context of computer executable instructions executed by a computer, such as program modules. Generally, program modules include routines, programs, objects, components, data structures, etc. that perform specific tasks or implement specific abstract data types. The present application can also be practiced in distributed computing environments, in which tasks are performed by remote processing devices connected through a communication network. In a distributed computing environment, program modules can be located in local and remote computer storage media including storage devices.

[0189] Obviously, the embodiments described above are only some embodiments of the present application, rather than all embodiments. The preferred embodiments of the present application are given in the accompanying drawings, but they do not limit the patent scope of the present application. The present application can be implemented in many different forms. On the contrary, the purpose of providing these embodiments is to make the understanding of the disclosure of the present application more thorough and comprehensive. Although the present application is described in detail with reference to the aforementioned embodiments, for those skilled in the art, it is still possible to modify the technical solutions recorded in the aforementioned specific implementation methods, or to perform equivalent replacement of some of the technical features therein. Any equivalent structure made using the contents of the specification and drawings of this application, directly or indirectly used in other related technical fields, is similarly within the scope of patent protection of this application.

Claims

1. A method for collecting livestock data in a breeding farm, characterized in that: The method is applied to a drone data collection platform, wherein the drone data collection platform is equipped with a plurality of human drones, and the method comprises: Acquire a first video frame sequence shot by a first drone targeting livestock in the farm; In the first video frame sequence, identifying the position information of the livestock in each video frame according to a preset target detection algorithm; Determine a livestock movement path based on the location information of the livestock in each video frame, and mark the livestock movement path in the first video frame sequence; Using the first video frame sequence carrying the path mark to train a preset initial prediction model to obtain a moving path prediction model; receiving a data collection instruction, instructing the first drone to take photos of the livestock in the farm to obtain a current livestock location map; Inputting the current livestock position map into the movement path prediction model to obtain the predicted movement path of the livestock; Instruct a second drone to shoot the livestock according to the predicted moving path to obtain a second video frame sequence.

2. The method for collecting livestock data in a breeding farm according to claim 1, characterized in that: Before the step of identifying the position information of the livestock in each video frame according to a preset target detection algorithm in the first video frame sequence, the method further includes: Acquire a historical video frame sequence, wherein the historical video frame sequence is a video frame sequence carrying a target position label; Importing the historical video frame sequence into a preset initial target detection model, wherein the initial target detection model is built based on the target detection algorithm; Acquire an initial detection result output by the initial target detection model, wherein the initial detection result is the position coordinates of the target in each historical video frame predicted by the initial target detection model; Compare the initial detection result with the target position label to obtain a position prediction error; The initial target detection model is iterated based on the position prediction error using a back propagation algorithm until the model is fitted to obtain a target detection model.

3. The method for collecting livestock data in a breeding farm according to claim 2, characterized in that: The step of identifying the position information of the livestock in each video frame according to a preset target detection algorithm in the first video frame sequence specifically includes: identifying a location of livestock in each video frame in the first sequence of video frames using the object detection model; The position information of the livestock in each video frame output by the target detection model is obtained.

4. The method for collecting livestock data in a breeding farm according to claim 1, characterized in that: The step of determining the livestock movement path based on the position information of the livestock in each video frame, and marking the livestock movement path in the first video frame sequence specifically includes: assigning a target tracker to each video frame based on the position information of the livestock in each video frame; Starting from the first video frame of the first video frame sequence to the last video frame of the first video frame sequence, starting each target tracker in sequence to track the livestock in the first video frame sequence; When the target tracker of the last video frame completes tracking, generating the livestock movement path; The livestock movement path is marked in the first sequence of video frames.

5. The method for collecting livestock data in a breeding farm according to claim 1, characterized in that: The initial prediction model is built based on a long short-term memory network architecture, and the step of using the first video frame sequence carrying the path mark to train the preset initial prediction model to obtain the moving path prediction model specifically includes: Extracting features from the first video frame sequence to obtain video frame sequence features; Based on the video frame sequence features, the temporal dependency and spatial relationship between video frames are captured by the long short-term memory network; Based on the temporal dependency and spatial relationship between the video frames, predicting the movement path of the livestock in the first video frame sequence to obtain the historical predicted movement path of the livestock; Compare the path mark with the historical predicted moving path to obtain a path prediction error; The initial prediction model is iterated based on the path prediction error using a back propagation algorithm until the model is fitted to obtain a mobile path prediction model.

6. The method for collecting livestock data in a breeding farm according to claim 1, characterized in that: The step of instructing the second drone to shoot the livestock according to the predicted moving path to obtain a second video frame sequence specifically includes: Acquire a photographed image output by the second UAV during the process of photographing the livestock according to the predicted moving path to obtain an initial photographed image; Performing image quality evaluation on the initial captured image to obtain image quality parameters; Screening the initial captured image according to the image quality parameter to obtain a screened image; The second sequence of video frames is constructed based on the filtered image.

7. The method for collecting livestock data in a breeding farm according to claim 6, characterized in that: The step of evaluating the image quality of the initial captured image to obtain image quality parameters specifically includes: Acquiring image parameters of the initial captured image, wherein the image parameters include resolution, contrast, color saturation and noise level; Calculating the quality evaluation weight of each of the image parameters; Performing weighted summation on each of the image parameters based on the quality evaluation weight to obtain a comprehensive evaluation index; The comprehensive evaluation index is used as an image quality parameter of the initial captured image.

8. A livestock data collection device for a breeding farm, characterized in that: The device is operated on a drone data collection platform, and the drone data collection platform is equipped with several drones. The device includes: A first shooting module is used to obtain a first video frame sequence shot by a first drone targeting the livestock in the farm; A target detection module, used for identifying the position information of the livestock in each video frame in the first video frame sequence according to a preset target detection algorithm; A path marking module, used for determining a livestock movement path based on the position information of the livestock in each video frame, and marking the livestock movement path in the first video frame sequence; A model training module, used to train a preset initial prediction model using the first video frame sequence carrying the path mark to obtain a moving path prediction model; A second shooting module is used to receive a data collection instruction and instruct the first drone to shoot the livestock in the farm to obtain a current livestock position map; A path prediction module, used for inputting the current livestock position map into the movement path prediction model to obtain the predicted movement path of the livestock; The third shooting module is used to instruct the second drone to shoot the livestock according to the predicted moving path to obtain a second video frame sequence.

9. A computer device, characterized in that: The method comprises a memory and a processor, wherein the memory stores computer-readable instructions, and the processor implements the steps of the method for collecting livestock data in a farm as claimed in any one of claims 1 to 7 when executing the computer-readable instructions.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer-readable instructions, and when the computer-readable instructions are executed by a processor, the steps of the method for collecting livestock data on a farm according to any one of claims 1 to 7 are implemented.

Images