47 results about "Visual motion" patented technology

The invention relates to the technical field of numerical control machine tool fault diagnosis, in particular to a remote video-based numerical control machine tool remote fault diagnosis method and system. The method comprises the following steps: collecting video data of the numerical control machine tool under multiple view angles through remote video equipment; motion basic parameters are calculated according to the video data; performing motion feature analysis on the motion basic parameters to obtain an axis motion feature vector; performing axis motion stability calculation on the axis motion characteristic vector to obtain an acceleration stability parameter; performing trajectory curvature analysis according to the acceleration stability parameter and the axis motion feature vector to obtain a visual motion smoothness index; working condition feeding response processing is carried out according to the visual motion smoothness index, and a smoothness-feeding response curve is obtained. According to the method, fault diagnosis is improved from passive response to active prediction through a numerical control machine tool fault diagnosis technology, the non-planned downtime is greatly shortened, and the equipment operation reliability and the production efficiency are improved.

The invention provides a display method, electronic equipment and a storage medium, and relates to the technical field of display. The method comprises the steps that in response to a first operation, the electronic equipment displays a first interface; displaying the target animation in a preset area of the first interface; the target animation comprises animation elements, the animation elements move or are static according to the motion state of the electronic equipment, and the target animation is used for mapping the motion state of the electronic equipment in real time. In the implementation mode, the animation elements in the target animation change according to the motion state of the electronic equipment, so that the displayed target animation can map different motion states of the electronic equipment in real time, the motion information sensed by the eyes of a user is consistent with the motion information sensed by a vestibular system, and the user experience is improved. Namely, the motion state of the target animation in the display interface seen by the eyes of the user is consistent with the motion state sensed by the vestibular system. Therefore, the conflict between the user vision and the motion state sensed by the vestibular system can be relieved, and the motion sickness of the user is relieved.

The invention provides a multi-configuration robot-oriented model training method and device and a storage medium, and relates to the technical field of robot control, and the method comprises the steps: constructing an initial task dynamic representation model, training an initial unified visual motion network according to a mechanical arm operation data set for a multi-configuration robot, and obtaining an initial task dynamic representation model; obtaining a target unified visual action network; performing feature extraction on the mechanical arm operation image sequence and the corresponding mechanical arm state sequence by adopting a target unified visual motion network to obtain a hidden space visual motion representation; and the hidden space visual motion representation serves as an auxiliary supervision signal, the initial VLA strategy network is trained according to the mechanical arm operation image sequence, the corresponding mechanical arm state sequence and the corresponding natural language instruction, and a target VLA strategy network is obtained. The robustness of model strategy learning and the model universality are enhanced, and the execution effect of the configuration robot is improved.

The invention relates to the technical field of artificial intelligence, and discloses a spiking neural network-based robot control method, which comprises the following steps of: acquiring dynamic visual information of an external environment, and encoding the dynamic visual information to generate a dynamic visual pulse stream; collecting motion state information of the robot, and encoding the motion state information to generate a motion state pulse code; through a pulse attention gate control mechanism, computing resources of the pulse neural network are dynamically allocated according to task priorities; inputting the dynamic visual pulse streams and the motion state pulse codes into a pulse neural network, processing the dynamic visual pulse streams and the motion state pulse codes by the pulse neural network according to computing resources, and outputting dynamic visual features and motion state features; fusing the dynamic visual features and the motion state features to generate a visual-motion pulse signal; and converting the vision-motion pulse signal into a motion control instruction, and controlling the robot to execute a corresponding action according to the motion control instruction.

The invention relates to the technical field of auditory body sense training, and discloses an earphone and video auditory body sense training linkage control system. The system comprises a multi-modal data acquisition module, a data fusion processing module, a dynamic feature extraction module, an adaptive weight distribution module, an anomaly detection engine module and a real-time feedback adjustment module. The multi-modal data acquisition module synchronously acquires earphone real-time audio feature data and video training system visual motion feature data; the data fusion processing module aligns the two types of data according to timestamps to generate a space-time incidence matrix; a dynamic feature extraction module extracts auditory sense-somatosensory sense coupling feature vectors from the dynamic feature extraction module; a self-adaptive weight distribution module adjusts a coupling feature vector weighting coefficient according to the target parameter of the current training stage; the anomaly detection engine identifies an abnormal mode based on the deep residual network; and the real-time feedback adjustment module generates an audio and video parameter adjustment instruction according to the detection result. The system realizes linkage control of auditory sense and body sense training, and is suitable for scenes of rehabilitation training, exercise guidance and the like.

This invention belongs to the field of image processing technology and discloses a power equipment status recognition method based on multi-source data fusion. The method includes: acquiring multi-source data of a target transmission line segment and performing time synchronization processing to obtain aligned multi-source data; acquiring line structure parameters and, combined with aligned electrical vector data, analyzing excitation intensity indicators reflecting the intensity of periodic electromagnetic interaction between adjacent lines; determining whether preset suspected risk conditions are met; if so, generating a trigger command and collecting corresponding multi-source data to form an event data packet; extracting motion features from the event data packet to obtain the line's visual motion features; performing correlation analysis based on the corresponding electromagnetic excitation features to obtain physical verification results; and analyzing the risk level of the current line's dynamic swaying based on the physical verification results and visual motion features. This invention effectively warns and prevents the risk of insufficient conductor phase-to-phase distance caused by dynamic power flow attraction.

The invention relates to a mobile robot motion estimation method based on pose decoupling. The method comprises the following steps: multi-sensor synchronization and decoupling state modeling; performing attitude-position decoupling pre-integration based on an IMU (Inertial Measurement Unit); visual observation modeling and reprojection constraint construction are carried out; carrying out pose optimization solution based on the decoupling factor graph; pose recovery, motion track output and local re-initialization are carried out; compared with the prior art, the method has the advantages that independent modeling and decoupling estimation are carried out on the pose rotation component and the translation component, the propagation effect of rotation errors to translation solution in traditional visual motion estimation is effectively inhibited, and the stability of translation estimation is remarkably improved. Under visual degradation scenes of weak texture, low parallax, illumination variation, motion blur and the like, estimation drift can be reduced, track continuity and consistency can be improved, and robustness of the system to abnormal matching and noise disturbance can be enhanced; besides, the decoupling mechanism provided by the invention has good universality, can be integrated with various types of visual SLAM or visual-inertial fusion frames, and provides more reliable motion estimation capability for stable autonomous navigation of the mobile robot in a complex environment.

The embodiment of the invention discloses a video linkage tracking method and device, electronic equipment and a storage medium, potential targets and motion trends are analyzed and accurately identified based on a panoramic video frame sequence, target position distribution and the motion trends are combined to dynamically divide and adjust a region of interest in real time, a fixed region division mode is replaced, and the tracking efficiency is improved. The fast moving target is always in an attention range, and the defect that tracking response is not timely is overcome; secondly, configuring a multi-dimensional dynamic triggering condition including a motion speed, a stay time length and an overlapping degree for each dynamic attention area, and adapting to target motion characteristic differences to avoid the limitation of intrusion triggering of a single area; meanwhile, by extracting target vision, motion and category exclusive multi-dimensional features and combining a multi-modal tracking algorithm, precise tracking is achieved, the defect that the precision is poor in a single-device view blind area and a multi-target parallel scene is overcome, false triggering and missed triggering are effectively reduced, and the accuracy and reliability of video linkage tracking are remarkably improved.

Provided is a system, method, and computer program product for analyzing spatio-temporal medical images using an artificial neural network. The method includes capturing a series of medical images of a patient, the series of medical images comprising visual movement of at least one entity, tracking time-varying spatial data associated with the at least one entity based on the visual movement, generating spatio-temporal data by correlating the time-varying spatial data with the series of medical images, and analyzing the series of medical images based on an artificial neural network comprising a plurality of layers, one or more layers of the plurality of layers each combining features from at least three different scales, at least one layer of the plurality of layers of the artificial neural network configured to learn spatio-temporal relationships based on the spatio-temporal data.

The invention relates to the technical field of robot control, and discloses a visual motion joint modeling and prediction method, device, equipment and medium, and the method comprises the steps: constructing a unified input data set, carrying out the preprocessing and time alignment of a visual image and a motion state vector, and generating standardized input data; inputting the visual image into a visual backbone network to obtain a visual feature sequence; the visual feature sequence is sent to a time sequence modeling module to generate feature representation fused with a time sequence context; inputting the feature representation fused with the time sequence context into an action prediction branch to generate an action prediction vector; generating a training loss item set based on the motion prediction vector and the demonstration label, and jointly updating the model to obtain an optimization model; constructing a contrast model based on the optimization model and generating an evaluation record; and determining optimization configuration based on the evaluation record and updating the optimization model. According to the method, through unified perception and prediction and dynamic optimization, the precision and efficiency are improved, and the autonomous operation capability in a complex environment is enhanced.

The invention discloses a video call forgery detection method based on a high-frequency sound signal, and the method comprises the steps: enabling a terminal device to actively transmit a high-frequency sound signal which is imperceptible to human ears in a video call process, and embedding a random sequence code in the high-frequency sound signal; the method comprises the following steps: synchronously acquiring an audio signal and a video signal of a video call, verifying a random sequence carried in the audio, extracting a human body motion feature of a target user reflected by high-frequency sound response, and extracting a corresponding visual motion feature from a video picture; and judging whether the video call is a real-time real call or not by analyzing the time sequence consistency between the acoustic motion feature and the visual motion feature. According to the method, a high-frequency sound active detection mechanism of random sequence coding and cross-modal motion consistency check are utilized, and playback and synthesis class forgery in a video call are defended from the physical level. The system can be deployed in terminal equipment such as a computer, a smart phone and the like with a loudspeaker, a microphone and a camera.

The present application relates to a kind of vehicle speed detection methods based on outer pole face image, belong to the field of automatic driving.The method includes: S1: periodically capture the image of the road in front of vehicle, obtain time-space distance image;S2: the curve cluster surface of the time-space distance image obtained is divided into several parts;S3: the curve corresponding to each part obtained;S4: by its analytic differential, the speed curve of each part is obtained;S5: the generation of full interval speed curve, according to the displacement of feature point, the speed of vehicle is calculated.The present application is based on the principle of visual kinematics, using the image obtained by vehicle-mounted camera and distance sensor, the speed of vehicle is calculated by observing the motion of feature point in image.Compared with the existing speed detection method based on physical sensor, the method is lower in cost, but the principle is simple, and the accuracy is also higher.The method can be used alone, can also be combined with the data of other sensors, improve detection accuracy and stability.

The invention discloses a generative AI training scene synthesis method for vestibular rehabilitation. The invention belongs to the technical field of medical artificial intelligence, and the method comprises the steps: obtaining vestibular function defect features of a subject, the features comprising a cold and hot test asymmetry ratio, a vHIT gain value or spontaneous nystagmus intensity; the features are coded into scene control vectors, the scene control vectors comprise the visual motion direction, the angular velocity amplitude, the acceleration change rate and the space complexity, and the angular velocity amplitude is dynamically calculated according to the patient compensation threshold value and the function retention rate and covers 60%-90% of the patient compensation threshold value and the function retention rate; inputting the vector into a pre-trained conditional diffusion model or a conditional generative adversarial network (cGAN) or other generative AI models to generate a customized VR scene; and rendering the scene to a head-mounted display device for rehabilitation training. According to the invention, one person has one scene and accurate stimulation is realized, the problems of fixed scene, uncontrollable stimulation, high development cost and the like of the existing VR rehabilitation scene are solved, and the training safety and effectiveness are improved.

The invention discloses a mobile robot formation finite time control method based on visual servo. According to the method, firstly, a relative kinematics model between a navigation robot and a following robot is established under a world coordinate system, the relative kinematics model is converted into a relative vision kinematics model in an image space in combination with a pinhole camera model, and an image error vector is obtained through normalization; for the error model, a non-linear observer is designed to estimate the unknown speed of the piloting robot, a parameter adaptive updating law is provided to process the uncertain feature height, and a finite time controller based on the linear speed and the angular speed is further designed to realize accurate tracking of the piloting robot. The information required by the method can be acquired from the image, the method does not depend on local communication or GPS between the robots, the communication and sensor loads are small, the tracking control precision is good, the system response speed is high, and the method is suitable for efficient cooperative control in the multi-robot formation task.

This invention discloses a method, system, and robot for generating 3D visual motion strategy training data based on task decomposition. The method includes: constructing a Task-Edit fully synthetic framework; firstly, parsing and reconstructing single-source human demonstration data to recreate the operation process in a simulation environment and generate a baseline template; then, decoupling the task into three major components: scene, skill, and object; generating massive and diverse "scene-skill-object" configuration data through independent editing and flexible combination of the attributes of each component; subsequently, completing Sim2Real data adaptation through depth map repair, point cloud denoising and fusion to narrow the gap between simulation and real-world scenes; finally, using high-quality structured data for 3D visual motion strategy training. This invention overcomes the limitations of traditional object center transformation, significantly improves model generalization ability, effectively enhances adaptability to complex scenes and long-term tasks, reduces data generation and deployment costs, and provides an end-to-end solution for 3D visual motion strategy training for complex robot operations.

The application relates to the technical field of blind guiding robot interactive navigation, and particularly discloses a blind guiding robot interactive navigation system combining vision, inertial navigation and voice, wherein the blind guiding robot interactive navigation system synchronously acquires environment vision, motion and sound data through a multi-modal perception data acquisition module, analyzes the data through a context-user joint state feature extraction module, combines a user potential intention deduction module, predicts the user intention, generates an adaptive guiding strategy, adjusts a navigation path according to the strategy through an adaptive guiding path optimization module, ensures safety and efficiency, and finally provides voice and tactile feedback through a situational multi-modal interactive instruction output module, thereby enhancing the interactive experience and navigation autonomy. The application solves the travel problem of the visually impaired in a complex environment, improves the individualization, situational perception ability and safety of blind guiding, and brings more natural and efficient navigation assistance to the visually impaired.

A technique of camera exposure control for vision-based navigation of an unmanned aerial vehicle (UAV) includes acquiring an aerial image of a ground area below the UAV with an onboard camera system of the UAV, estimating a visual motion factor based on a speed of the UAV and an altitude of the UAV, and adjusting an exposure control setting of the onboard camera system based on the visual motion factor.

This application discloses a method, device, storage medium, and vehicle for controlling a motion sickness-preventing seat, relating to the field of vehicle seat adjustment and control technology. The method includes: detecting that an in-vehicle display screen is turned on; extracting visual motion acceleration from the content being played on the display screen and determining the user's perceived motion acceleration; identifying a first user viewing the display screen and determining a first seat occupied by the first user; and controlling and adjusting the first seat based on the visual motion acceleration and the perceived motion acceleration. This method can actively align visual perception of motion with bodily sensation feedback while the user is watching in-vehicle entertainment content using visual motion acceleration. This eliminates sensory conflict, ensuring consistent sensory signals, significantly reducing the incidence of motion sickness. The enhanced immersion in the content through motion feedback improves user satisfaction. The system automatically starts and stops, making it convenient to use, and the hardware cost is controllable.

The application discloses a kind of robot simulation scene and trajectory automatic generation method and system, receive natural language operation instruction, and through semantic analysis, visual asset generation, three-dimensional reconstruction and physical credible layout generation three-dimensional simulation scene that can be used for robot operation training;Utilize video generation model to synthesize the continuous image sequence of target operation task, and extract two-dimensional visual motion trajectory in target object area;By depth estimation, three-dimensional motion fusion, grasping posture generation and inverse kinematics mapping, two-dimensional visual motion trajectory is converted into robot end effector trajectory and corresponding robot joint trajectory, the simulation scene and expert trajectory generated are paired, form training data set for robot imitation learning strategy training, and support the strategy obtained by training is deployed to simulation environment or real robot hardware.The application can reduce the cost of robot training data acquisition and artificial trajectory design, improve the degree of automation of simulation scene and operation trajectory generation.

A gap measuring device for elevator inspection and detection relates to the technical field of elevator detection tools, a threaded rod is screwed and inserted at the front end of a mounting sleeve through a bearing, the rear end of the threaded rod is screwed at the rear end of the mounting sleeve through a bearing, a mounting seat is screwed and sleeved on the threaded rod through a thread, a connecting sleeve is fixedly arranged on the mounting seat, and the connecting sleeve is fixed on the mounting seat. The rear end of the connecting sleeve is arranged at the rear end of the mounting sleeve in a penetrating manner, the connecting bolt is fixedly arranged on the probe, and the connecting bolt is screwed and inserted into the rear end of the connecting sleeve through a thread; by arranging a probe with adjustable standard parameters and an elastic telescopic support structure, the probe can be simply inserted into a gap and touches the bottom when the probe is inserted for detection, and meanwhile, the device is also provided with a visual motion conversion structure, so that the movement of the probe can be converted into a simpler swing display structure, and the detection accuracy is improved. Therefore, the visual depth detection of the gap of the elevator workpiece can be realized through the device.

The invention discloses a training method of an end-to-end visual motion model, a trajectory planning method and an application thereof to meet the requirements of high-level automatic driving application, and the training method comprises the steps: inputting a sample image at a T moment into a first model, and obtaining a perception task query and an image Token corresponding to an image perception result at the T moment; based on the perception task Query at the T moment and a second model, obtaining a predicted trajectory of the target object and a planned trajectory of the target vehicle at the T moment; based on the predicted trajectory of the target object at the T moment, the planned trajectory of the target vehicle and a third model, obtaining a predicted trajectory of the target vehicle at the T + 1 moment; based on the predicted trajectory of the target vehicle at the T + 1 moment, the planned trajectory of the target vehicle at the T moment, the image Token and a fourth model, obtaining a predicted image at the T + 1 moment; and executing back propagation of the model gradient based on the predicted image at the T + 1 moment so as to update parameters of the model and construct an end-to-end visual action model.

The application provides a visual motion function test system and method based on full-view surround synchronous visual stimulation, and applies to the technical field of medical data processing. The visual stimulation parameters and the rotating rod motion mode parameters are combined, the strip grid rolling direction and the rotating rod rotating direction are synchronously matched, the synchronous presentation scene of the visual stimulation and the motion task is established based on the four-screen linkage control technology and the rotating rod-screen adaptive design, and the full-view stimulation constraint condition is generated; the visual stimulation parameters, the rotating rod reference data, the synchronous control logic and the dynamic parameter combination constraint condition are processed, the rotating rod time, the falling rotating speed and the motion trajectory data are collected; the collected kinematic data are processed, the function parameters including the visual state grouping and the brain region neuron activation count are combined, the motion balance ability and brain function correlation data are generated; and the visual-motor function evaluation result is generated according to the combination of the normality test and the inter-group statistical method.

The invention relates to a movable vehicle-mounted parallel mechanism for target tracking and response based on vision and a control method.The vehicle-mounted parallel mechanism comprises a trolley shell (3), a trolley lower panel (2), a trolley upper panel (5) and a trolley bottom plate (14), and three parallel mechanism supports (9) are detachably and evenly connected to the trolley bottom plate (14) in the circumferential direction through bolts; each parallel mechanism support (9) is provided with a parallel mechanism module (7). The parallel mechanism module (7) comprises a monocular camera (37) and a depth camera (42); four Mecanum wheel modules (1) are symmetrically arranged on the two sides of the bottom face of the trolley bottom plate (14). According to the invention, full-automatic juggling in a moving state is realized, the requirements of dynamic physical principle demonstration and robot control teaching in an education scene are met, and the robot can be used as interactive equipment in an entertainment scene and a low-cost verification platform of a vision-motion closed-loop algorithm in a scientific research scene; and the blank in the moving juggling scene in the prior art is filled.

The invention provides a police moving target continuous tracking system and method based on artificial intelligence. The system comprises a skynet video recording data retrieval unit, an artificial intelligence target analysis unit and a moving track map drawing unit. The skynet video recording data retrieval unit comprises a server; the artificial intelligence target analysis unit is used for video frame picture segmentation, single frame picture analysis and object similarity calculation; the moving track map drawing unit is provided with a map interface and is used for geographic coordinate marking and dynamic track generation; the method comprises the following steps: video retrieval; analyzing a suspected target; drawing a moving track; an artificial intelligence target analysis unit is adopted to be connected with a skynet video recording data retrieval unit, the problems of high-speed processing of massive videos, suspected target recognition and historical track drawing are solved through intelligent analysis and by means of a computer concurrency technology, and finally a visual motion track is directly generated, so that investigation work is effectively assisted.

The present application relates to a kind of lower extremity exoskeleton motion mode prediction method and system based on VIO system, prediction method includes according to the input depth camera field of view environmental information and motion information, user state information is estimated, and constructs three-dimensional point cloud model diagram, constructs topographic information by extracting plane information from the depth map under the current perspective of depth camera and mapping to three-dimensional point cloud model diagram, and user current topography is judged in combination with user state information, accurately predict the motion mode of lower extremity exoskeleton and send switching instruction.System includes visual integrated wearing module, lower extremity sensing module and VIO system operation module, respectively for collecting visual motion data, gait cycle phase and the step of executing exoskeleton motion mode prediction method.The present application can effectively solve the problem of insufficient terrain recognition accuracy, user positioning, motion state estimation is not accurate, to realize the accurate prediction of the switching state of lower extremity exoskeleton motion mode.

The invention discloses a visual motion capture method and device, equipment and a storage medium. The method comprises the following steps: acquiring a scene capture image, and determining motion capture images of at least two human body parts according to the scene capture image; respectively inputting each motion capture image into a recognition model matched with each human body part, and determining motion capture parameters of each human body part; according to the topological connection relation of the human body parts, correcting the motion capture parameters of the human body parts to obtain motion capture correction parameters of the human body parts; and determining a motion capture result according to the motion capture correction parameter of each human body part. According to the technical scheme, topology correction is carried out on the motion capture parameters of all the human body parts, the problem of poor motion linkage caused by direct splicing of the motion capture results of all the human body parts is solved, and the consistency and visual effect of human body motion capture are improved.

The invention relates to a sports motion error correction method and device based on computer vision, equipment and a medium. The method comprises the following steps: firstly, acquiring visual motion video data of a target user in a sports process, extracting a visual spatial-temporal feature vector through a preset visual feature extraction network, further generating a disturbance visual feature vector, calculating a joint feature drift vector and an error perception statistical magnitude through a micro-error disturbance model, and calculating the visual motion video data according to the joint feature drift vector and the error perception statistical magnitude. And fusing the visual spatio-temporal feature vector and the error perception statistic to generate a fused feature vector, and generating an error correction instruction according to the fused feature vector. By adopting the method, through active evaluation and integration of stability information of visual features, negative conduction of underlying perception uncertainty to a final decision instruction is effectively inhibited, and the reliability of error correction guidance in a complex real scene is improved.

The utility model provides a device for simulating and drawing solar apparent moving tracks at different latitudes and different times, and belongs to the technical field of teaching aids. The device comprises a bottom support, a lamp strip switch, the earth, a connecting shaft, a celestial sphere, a lamp strip, a declination circle, a sky, a sky support and a lighting mechanism, the lamp strip switch is installed on the bottom support, the earth is fixed in the celestial sphere through the connecting shaft, and the connecting shaft is connected with the connecting shaft. According to the equipment for simulating the moving tracks of the sun at different latitudes and different time, the moving tracks of the sun at different latitude positions can be conveniently described on the sky, abstract problems are visualized, students can conveniently observe the moving tracks of the sun, the students can conveniently learn and understand the content of the sun apparent motion, and the teaching efficiency is improved. Students can conveniently transfer and draw a three-dimensionally presented annual solar apparent motion track into a graph, further understand abstract knowledge of solar apparent motion, and improve the cosmic view of a learning concept.

The application discloses a kind of self anti-off-plane disturbance micron-level precision visual motion tracking methods, method includes: S1: constructing micron-level precision visual motion tracking measurement system;S2: the motion image of observation object is acquired, and real-time detection off-plane disturbance;S3: for the visual precision degradation caused by off-plane disturbance, self anti-off-plane disturbance device and self anti-off-plane disturbance program are optimized to self anti-interference scheme;S4: according to the fuzzy rule of preestablished, for low frequency high amplitude type off-plane disturbance, precision repair is realized to self anti-off-plane disturbance device in hardware level optimization;For high frequency low amplitude type off-plane disturbance, precision repair is realized to self anti-off-plane disturbance program in software level optimization;For wide frequency or cross amplitude hybrid type off-plane disturbance, precision repair is realized to self anti-off-plane disturbance device and self anti-off-plane disturbance program synchronous linkage;The application realizes the precision optimization of micro visual motion tracking under different off-plane displacement disturbances, improves the anti-off-plane disturbance ability of system.

The invention belongs to the technical field of sow management, and particularly relates to a sow oestrus recognition method and system based on a neural network, and the method comprises the steps: firstly, synchronously collecting a dynamic posture video, three-axis acceleration and skin resistance data in a pigsty, so as to cover the posture, activity and physiological information related to sow oestrus; thirdly, video frames are synchronously corrected, sensor data are filtered and denoised, time-space registration is achieved through timestamp alignment, and data consistency is guaranteed; respectively extracting visual, motion and physiological features by using a three-dimensional convolutional neural network, a bidirectional long-short-term memory network and a one-dimensional convolutional neural network with dynamic weights, and fusing the features through a double-stage attention mechanism; inputting the fusion features into a classifier based on a converter, optimizing a model by a mixed loss function, and outputting an oestrus probability P; and finally, when P exceeds a preset threshold value of 0.7-0.9, the early warning module triggers acousto-optic and terminal prompting, and the whole process from data acquisition to oestrus early warning is completed.