156 results about "Point tracking" patented technology

The invention belongs to the field of power distribution network monitoring, and provides a power distribution box damage behavior monitoring method and system based on human body posture estimation, and the method comprises the steps: obtaining a monitoring video of a power distribution box terminal for preprocessing, and obtaining a preprocessed video frame; based on the preprocessed video frame, performing posture recognition by using a pre-trained human body and distribution box posture recognition model to obtain a human body and distribution box posture recognition result; key point tracking and identity association are performed by using a human body and distribution box posture recognition result, a space-time skeleton diagram is constructed, and the space-time skeleton diagram is screened to generate a skeleton space-time sequence; on the basis of the skeleton space-time sequence, a pre-trained space-time diagram convolutional network model is utilized to identify a damage behavior, and a damage behavior identification result is obtained; when the confidence coefficient in the damage behavior recognition result exceeds an alarm threshold value, alarm information is generated, and an alarm process and safety linkage control are triggered. According to the invention, accurate monitoring and real-time early warning of the damage behavior of the distribution box are realized.

The invention discloses a civil engineering foundation pit deformation monitoring system based on image analysis, and relates to the technical field of foundation pit deformation monitoring, and the system collects and enhances a multi-modal image, eliminates interference, corrects image distortion, tracks key points, constructs a panoramic image, generates a point cloud, extracts deformation features, and constructs a deformation map. The method comprises the following steps: identifying a construction disturbance event, constructing a deformation chain and performing causal reasoning, identifying a high-risk area, predicting a deformation trend and outputting an early warning, displaying a monitoring result, collecting user feedback and synchronizing system data. A deformation track is accurately extracted through image correction and key point tracking, a three-dimensional deformation map is constructed in combination with point cloud modeling, causal analysis is carried out based on construction disturbance and a deformation path, a high-risk area is predicted and identified by using a time sequence trend, an early warning is given out, user feedback collection and result synchronization are supported, and monitoring intelligence and response efficiency are improved.

The invention relates to the technical field of smart home control, and particularly discloses a smart home central control system and method based on multi-mode perception, and the system comprises the steps: synchronously collecting a voice audio signal, a gesture image signal, an infrared thermal imaging signal and a millimeter wave radar signal through a plurality of groups of sensors; performing blind source separation processing on the voice and gesture signals, extracting a voice command component and a gesture action component which are independent in statistics, and performing space-time alignment and Kalman filtering fusion on the infrared and radar signals to generate a dynamic environment sensing graph; voice intention features, gesture track features and environment anomaly features are extracted through Mel frequency cepstrum coefficient analysis, skeleton key point tracking and multi-level convolution processing; constructing a three-dimensional decision matrix based on the features, performing weighted evaluation through a fuzzy logic rule base to generate a control instruction priority sequence, and dynamically adjusting an equipment operation mode according to the priority; according to the invention, the problems of control conflict and response delay caused by multi-mode signal coupling are solved.

The invention provides an unmanned aerial vehicle cluster fire point real-time detection and positioning method and system. The method comprises the following steps: acquiring a visible image and an infrared image synchronously shot by a single unmanned aerial vehicle, and recorded shooting time and unmanned aerial vehicle pose data; processing the visible image and the infrared image based on threshold segmentation and morphological filtering to obtain a moment image pair; inputting the moment image pair into the trained double-attention segmentation network, and outputting a fire semantic map; analyzing and processing the fire semantic map based on the connected domain, outputting candidate fire points, and obtaining initial geographic coordinates of the candidate fire points based on the pose data of the unmanned aerial vehicle; based on the shooting time, tracking the initial geographic coordinates of the candidate fire points in real time through Kalman filtering, and outputting a fire point tracking and positioning result of the single unmanned aerial vehicle at the to-be-positioned moment through dynamic compensation of elevation data; and fusing the fire point tracking and positioning results of all the unmanned aerial vehicles in the unmanned aerial vehicle cluster at the to-be-positioned moment to obtain a fire point fusion positioning result at the to-be-positioned moment.

The invention provides an outdoor cross-modal robust positioning and navigation method for extreme severe weather, and relates to the field of online positioning and navigation of mobile robots. The method comprises the following steps: acquiring and preprocessing data; generating a path point set based on a CMR network, perfecting the path point set through local trajectory fitting and error optimization, and constructing a path point relative pose measurement-topology hybrid map; motion prior estimation is carried out based on Doppler velocity, effective matching prior estimation is generated in combination with LiDAR odometer information, cross-modal pose estimation is carried out through a CMR network, and a pure tracking strategy and PD controller path point tracking are adopted; and geometric and intensity alignment is carried out on the forward 4D millimeter wave radar point cloud and the omnidirectional 3D laser radar point cloud through a CMR network, a rotation matrix and a translation vector are output, and accurate positioning is completed. According to the invention, the sensing and positioning defects in extreme weather in the prior art are overcome, and the positioning precision and robustness are improved.

The invention relates to the technical field of 3D visual inspection, in particular to a dynamic shielding area multi-view time sequence point cloud completion method and system in 3D visual inspection. The method comprises the following steps: collecting a multi-view time sequence image of a dynamic detection object for providing a data basis for feature point tracking; and carrying out Lucas-Kanade optical flow method tracking on the feature points, associating the feature points of different visual angles, describing a motion track of the feature points through an optical flow vector, carrying out cross-visual-angle and cross-time feature point matching, and establishing time sequence feature association to obtain time sequence feature points. According to the dynamic occlusion area multi-view time sequence point cloud completion method in 3D visual detection, 360-degree time sequence data of a dynamic detection object is acquired through multi-view time sequence image acquisition, and comprehensive basic data support is provided for subsequent feature tracking.

The invention discloses a monocular vision inertial fusion positioning method based on an optical flow smoothness constraint, which comprises the following steps: designing a visual feature enhancement algorithm based on the optical flow smoothness constraint, improving feature point tracking and elimination in a visual inertial navigation system, and carrying out tight coupling optimization on a visual residual error part so as to improve the quality of observed quantity in positioning; comprising the following steps: generating a clustering result and a support factor for each feature point tracking vector of an acquired image through optical flow smoothness constraint; constructing a compound expectation maximization (EM) algorithm, fusing a support factor into algorithm solution to serve as prior information, and eliminating wrong matching point pairs to obtain a correct matching vector; adaptively adjusting the visual weight of the pose estimator, and performing region shielding; and monocular vision inertial fusion positioning based on optical flow smoothness constraint is realized.

The invention relates to the technical field of AI evaluation, in particular to a post competency AI evaluation system and a post competency AI evaluation method, the post competency AI evaluation system comprises an intelligent scanning layer, a diagnosis decision-making layer, a talent development prediction layer, an education optimization layer and a data closed-loop feedback module, and the intelligent scanning layer comprises a multi-modal biological characteristic acquisition module and a dynamic ability modeling module; the talent development prediction layer comprises a capability-post matching engine and a demission risk early warning unit; the education optimization layer comprises a course reform simulator and a virtual teaching and research room; compared with the defects that in the prior art, single questionnaire or static video analysis is relied on, the evaluation dimension is one-sided, and subjective interference is likely to happen, the scheme synchronously integrates three types of biological characteristics of facial micro-expression (52 key point tracking), writing force waveform (1000 Hz pressure sense) and VR situation response (1200 + scene library), and the evaluation accuracy is improved. Holographic dynamic capture of a candidate cognitive mode, anti-pressure ability and actual combat response is realized, and objectivity and ecological validity of competency evaluation are remarkably improved.

A video processing method and a terminal are provided. In the method, when the terminal records a video, the terminal may shoot (acquire) an image by using a camera (the image shot by using the camera is referred to as an original image below), and determine a focus of the shooting based on the original image. Then, the terminal may implement image focus tracking on a photographed object displayed in a first image region in which the focus is located, and implement audio focus tracking on a photographed object displayed in a second image region in which the focus is located. A focus tracking video is obtained through the image focus tracking and the audio focus tracking.

The invention discloses a target self-adaptive positioning method and system for low-altitude routing inspection, and aims to solve the problems that target positioning depends on additional hardware, the complex environment adaptability is poor and the real-time performance is insufficient in the existing low-altitude routing inspection. According to the method, an unmanned aerial vehicle collects a target object image and extracts pixel coordinates, Beidou positioning data and IMU attitude information are fused, and normalized coordinates in a camera coordinate system are obtained through coordinate conversion; a dynamic virtual baseline technology is innovatively introduced, depth information of a target object is obtained through feature point tracking, motion solution and depth calculation, and finally three-dimensional longitude and latitude positioning of the target object is achieved by combining a rotation matrix and a coordinate conversion algorithm. The method does not need extra hardware, improves the positioning precision and real-time performance in a complex scene, and is suitable for the dynamic target positioning requirement in the field of low-altitude inspection.

This paper describes a method and system for determining the 3D location of objects within an identified portion of an image. The image processing system receives an image and identifiers of locations within the image. The image can be input into a machine learning model to detect one or more objects within the identified locations. Multiple images can then be used to generate location estimates for those objects. Based on the location estimates, accurate 3D locations can be calculated.

Systems and methods are described for performing a clinical procedure utilizing sentiment analysis. The system includes an imaging system and a control system. The system is configured to obtain image data of a tissue region of a patient from the imaging system; obtain input data indicative of an assessment to perform on the tissue region; analyze the input data and the obtained image data to select one or more target points in the obtained image data to track for performing the indicated assessment; track, via a point tracking model and across image data obtained via a second imaging modality of the imaging system, the one or more target points for an assessment period to determine one or more tissue parameter values for each tracked target point; provide an output to a user based on (i) the input data, and (ii) the one or more tissue parameter values.

The application provides a power distribution network fault disposal simulation training evaluation method based on a seven-layer track node tracking array. First, the simulation training operation result information of seven links, such as a sensing node, in the fault disposal process of a dispatcher is described by establishing a seven-layer track node tracking array model. Then, the fault disposal process of the dispatcher is quantitatively evaluated by establishing a sensing node evaluation index. Secondly, the whole process from the sensing node to the archiving node of the dispatcher in disposing a certain fault is evaluated by establishing a single fault disposal evaluation index. Finally, the comprehensive evaluation of all fault disposal conditions in the simulation training of the dispatcher is performed by establishing a power distribution network fault disposal simulation training comprehensive evaluation index. The scheme can comprehensively reflect the fault disposal process of the dispatcher, better reflect the fault disposal level of the dispatcher, the method is simple and reliable, can meet the field application requirements, and has good popularization value.

The application provides a load action and target behavior recognition method based on a space time sequence image, and the method comprises the following steps: acquiring a time sequence image sequence of a space target to be recognized, and performing target region segmentation on the time sequence image sequence to obtain a segmentation mask; based on the segmentation mask, a plurality of query pixel points are selected, the time sequence image sequence and the query pixel points are input into a dense pixel point tracking model to obtain position coordinates, visibility indication and tracking confidence of the query pixel points in the time sequence image sequence, and a time sequence pixel track set of a load region and a target main body region is formed; according to a monitoring object of interest, inference logic of load action driving or main body motion driving is adopted, the load action driving is used for performing quantitative classification on the load action and the main body motion, and then performing quantitative identification on the target behavior, and the main body motion driving is used for directly performing quantitative identification on the target behavior; and finally, a target behavior category and a corresponding quantitative index are output, and reliable basis is provided for in-orbit state evaluation and abnormal diagnosis.

The invention relates to a penetration test method and system for mobile applications of a swan monk system, and the method comprises the steps: obtaining a current version installation package, extracting static asset data, and capturing dynamic behavior data; constructing a threat model for the gap distributed architecture; generating a penetration test case set through the threat model, and deploying and executing the penetration test case set; inputting the heterogeneous security telemetry data into the risk identification model, and executing a preset corresponding processing strategy; generating a penetration test report based on the risk judgment information and the execution result of the processing strategy; in conclusion, according to the penetration testing method and system for the mobile application of the swan-mong system, the threat model for the swan-mong distributed architecture is constructed, and the dynamic stain tracking and fuzzy testing technology is integrated, so that the cross-equipment safety risk is systematically detected; the method has the effects of effectively identifying the cross-equipment security threats in the swan-gap distributed architecture and improving the comprehensiveness and accuracy of the penetration test.

According to the intraoperative hemorrhagic spot trajectory determination method and device, the electronic equipment and the storage medium, the method comprises the steps that firstly, the endoscope video is obtained, then frame-by-frame detection is conducted on the endoscope video according to the hemorrhagic spot detection model, and the target image frame containing the hemorrhagic spot and the hemorrhagic spot coordinates of the hemorrhagic spot in the target image frame are obtained; and processing at least one image frame after the target image frame by using a bleeding point tracking algorithm according to the bleeding point coordinates, updating the bleeding point coordinates of the bleeding point in each image frame in real time, and finally processing the bleeding point coordinates updated in real time according to a trajectory processing algorithm to obtain a bleeding point trajectory of the bleeding point. According to the scheme, the whole process from recognition and positioning to dynamic tracking and complete track generation of the hemorrhagic spot in the operation is achieved, the accuracy and real-time performance of the position information of the hemorrhagic spot are ensured, the development situation of the hemorrhagic spot is visually presented through the complete track, and reliable data support is provided for rapid hemostasis decision making in the operation.

The invention discloses a physical field uniformity control method and system in the die forging forming process of a complex-shaped component, and the method comprises the steps: building a finite element model in the forging forming process, and obtaining a physical field at any moment through simulation calculation; the forming process is divided into a plurality of stages, and different forging speeds are adopted in each stage; designing a plurality of groups of different forging process parameter combinations, and performing finite element numerical simulation calculation on each group of parameters; establishing a coordinate system, and selecting data sampling points on the forge piece according to a Poisson disc sampling method; constructing a forming process data set of data sampling points by applying a finite element simulation calculation result and a finite element software point tracking function; a machine learning algorithm is selected, and a forming process prediction model is established by applying the data set; and a comprehensive optimization function is constructed, and forging process parameters for improving the uniformity of the physical field are rapidly determined, so that simultaneous rapid prediction and online accurate regulation and control of the space uniformity and the time-dependent uniformity of the physical field are realized.

This application discloses a multi-target point tracking and obstacle avoidance method and apparatus for unmanned aerial vehicles (UAVs). The method determines one or more path tracking points based on the UAV's global path and current position. The global path is the planned path for the UAV to fly from the path's starting point to its endpoint. Based on the path's endpoint and each path tracking point, state input data is acquired, including LiDAR data, the UAV's current action state, the relative position of the UAV to the path's endpoint, and the relative position of the UAV to each path tracking point. A trained deep reinforcement learning model is then acquired. The state input data is input into the trained deep reinforcement learning model to obtain action commands output by the model. These action commands are used to guide the UAV to avoid obstacles. Therefore, this application can, with limited computing resources, guide the UAV to avoid obstacles while simultaneously ensuring the UAV accurately reaches its path's endpoint.

This invention discloses an overload alarm device, method, and storage medium for a circulating vertical parking garage. The device includes: a vehicle platform weight detection module installed at the garage entrance, a signal processing module, a main control core, and a safety interlock module. The main control core embeds two weight limit thresholds (yellow warning threshold T1 and red shutdown threshold T2) and has dynamic zero-point tracking functionality. The alarm method of this invention eliminates errors through static zero-point tracking, performing static weighing and multi-level judgment after the vehicle has come to a complete stop; when a level two overload is detected (W > T2), an audible and visual alarm is triggered, and the starting power of the main motor drive device is forcibly cut off through the independent hardware circuit of the safety interlock module, while simultaneously locking the entrance door in an open state. This invention achieves graded early warning and physical forced blocking, effectively improving the safety of garage operation.

The invention relates to a fish respiratory rate detection method based on YOLO segmentation and optical flow tracking, and belongs to the technical field of aquatic organism physiological status monitoring. The method comprises the steps of video data input and initialization, video preprocessing and frame rate dynamic calibration, gill segmentation by a YOLO model and target screening, gill region-of-interest extraction, optical flow feature point tracking and dynamic maintenance, gill motion displacement calculation and signal optimization, respiratory rate calculation and smooth optimization, and result output and visualization. Under complex scenes such as water wave interference and fish swimming, the respiratory rate calculation error is smaller than or equal to + / -2 BPM, the effective data rate is larger than or equal to 95%, and the method is superior to the prior art (for example, the patent CN120599516A) and can be applied to scenes such as aquaculture health early warning and aquatic organism physiological experiments.

The application discloses a kind of large composite component autoclave process monitoring thermocouple arrangement and control methods, method includes the following steps: obtaining the temperature field data of whole curing process curve of part in autoclave;Obtain the low-temperature score sum and high-temperature score sum of each node;Obtain high-temperature score set and low-temperature score set;Screen out high-score low-temperature node and high-score high-temperature node;Obtain the node after optimization;According to the position and quantity of the node after optimization, arrange and control thermocouple.The application locks key monitoring point by unified scoring logic and dynamic weight mechanism, avoids the problems of part replacement failure and difficulty in capturing real temperature difference of experience scheme, overcomes the contingency and instability of single feature point tracking, effectively reduces process abnormal misjudgment and missed judgment risk, ensures accurate coverage of temperature key area while avoiding redundant arrangement. Ensure that monitoring target is always focused on temperature key area, avoid monitoring failure caused by feature point position distortion.

The application discloses a kind of ultra-high speed imaging method, device and system of event camera, belong to image processing technical field;The present application obtains the superpixel segmentation mask of motion area according to event trigger position and image apparent-position feature, selects event point in superpixel segmentation area to carry out continuous time point tracking, combines point tracking time sequence track and space superpixel mask to calculate continuous time and space dense optical flow field, uses optical flow field to calculate the imaging result between two frames at any imaging time t, can realize the high-precision imaging of ultra-high speed motion scene.

This invention discloses a cascaded multi-target tracking method and system with point tracking verification, relating to the fields of computer vision and target tracking technology. The method includes the following steps: Step 1, frame detection and trajectory initialization; Step 2, construction and execution of the point tracking verification module; Step 3, dynamic two-stage cascaded matching based on verification results; and Step 4, trajectory filtering by the short trajectory suppression module. This invention employs the aforementioned cascaded multi-target tracking method and system with point tracking verification. It achieves refined verification of trajectory-detection matching through feature point extraction and optical flow tracking, dynamically optimizes the matching cost matrix based on the verification results, and designs a dual-threshold short trajectory filtering mechanism to accurately eliminate pseudo-trajectories. Supplemented by a lightweight computational optimization strategy, it improves tracking robustness and accuracy while ensuring real-time performance. The overall solution can be non-intrusively integrated into existing tracking frameworks.

The invention belongs to the technical field of target detection, and discloses a video jitter removal method for unmanned aerial vehicle overhead transmission line inspection, the method uses a feature point method to track and carry out electronic image stabilization, and the method comprises the following steps: firstly, based on GFTT and other feature point detection algorithms, extracting and analyzing feature points of continuous video frames in an original video; secondly, performing feature point tracking by using an optical flow method to obtain estimated values of feature points, then describing relative motion among the feature points by adopting affine transformation, and further performing linear processing to obtain an original motion track of the camera; performing motion smoothing processing on the original motion trail of the camera to obtain a stable motion trail, namely a smooth motion trail, representing the motion of the camera; and finally, calculating a compensation transformation matrix by comparing the original motion trail and the smooth motion trail of the camera, and compensating the frame of the original video so as to generate a video with stable motion.

Methods and systems are described herein for determining three-dimensional locations of objects within identified portions of images. An image processing system may receive an image and an identification of location within an image. The image may be input into a machine learning model to detect one or more objects within the identified location. Multiple images may then be used to generate location estimations of those objects. Based on the location estimations, an accurate three-dimensional location may be calculated.

The application discloses a high-speed motion three-dimensional point tracking and trajectory reconstruction method and system, and relates to the technical field of three-dimensional point tracking. The method comprises the following steps: synchronously collecting an RGB image sequence and an event stream through a binocular multi-modal observation device; performing time alignment on the RGB image sequence and the event stream, extracting an RGB feature map based on the RGB image sequence, and extracting an event feature map based on the event stream; mapping the RGB feature map and the event feature map to a three-dimensional space; supplementing a three-dimensional feature point cloud through a linear interpolation algorithm; balancing the spatial distribution of the supplemented three-dimensional feature point cloud through an anchor point sampling algorithm to obtain a fused three-dimensional feature point cloud; and generating high-speed three-dimensional point trajectory data based on the fused three-dimensional feature point cloud through a time correlation transformer. The application effectively fuses the texture fidelity of the RGB image and the high time resolution of the event data, and can realize high-frame-rate and high-precision three-dimensional point trajectory reconstruction in a high-speed motion scene.

The utility model provides a direction-adjustable camera structure which comprises a camera mechanism and a main machine body, one side of the camera mechanism is connected with a rotary supporting unit, the other side of the camera mechanism is connected with a first driving unit, and the first driving unit drives the camera mechanism to rotate around a first axis of the rotary supporting unit. Side plates on the two sides of the main machine body respectively comprise a first extending part and a second extending part, the first driving unit is installed on the first extending part, and the rotary supporting unit is installed on the second extending part. The main machine body can further comprise a second driving unit, and the second driving unit is suitable for driving the main machine body to horizontally rotate around a second axis. Based on the camera shooting structure provided by the invention, the camera shooting structure can obtain a great pitching angle and a free horizontal angle at the same time, and the camera shooting structure is suitable for fixed-point tracking shooting of large-range cultural and sports or life scenes.

The application provides a human pelvis pose recognition method and device, electronic equipment and medium, and relates to the field of medical image registration. The method comprises the following steps: acquiring the pelvis 2D data and model pose of a patient in a minimally invasive pelvis surgery; inputting the pelvis 2D data into a trained interest point tracking network for processing to obtain the first positions of multiple interest points; processing the first positions of the interest points based on the triangulation method to obtain the second positions of the interest points, and then determining a pose offset matrix; and adjusting the model pose based on the pose offset matrix. The method proposed in the application can be understood as a multi-view 2D / 3D image registration method in pelvis minimally invasive surgery navigation based on transfer learning. A lightweight interest point tracking network combining U-Net and an attention mechanism is designed, and a style transfer module is introduced to effectively solve the problem of insufficient data and improve the performance of the network on real X-rays.

The application discloses a full-coverage unmanned aerial vehicle surveying and mapping path planning method and system, relates to the technical field of surveying and mapping path planning, and comprises the following steps: constructing a neural network energy model after carrying out inflation processing on original obstacles, rasterizing a surveying and mapping area, calculating total energy values of each grid center point, forming a three-dimensional energy field matrix, generating an initial skeleton path node sequence based on surveying and mapping start point coordinates, surveying and mapping end point coordinates and a preset segmentation number of the surveying and mapping area, adjusting the positions of internal nodes in the initial skeleton path node sequence, obtaining an optimized skeleton path node sequence, carrying out encryption processing, generating a dense flight point sequence, constructing a three-dimensional unmanned aerial vehicle motion model, constructing a feasible speed space, generating a candidate speed combination set, selecting optimal control instructions from the candidate speed combination set, tracking local target points through the optimal control instructions, and generating a final flight trajectory by complementing uncovered areas through iteration.

Methods, systems, and apparatus, including computer programs encoded on computer storage media, for tracking query points in videos using generation of tokens and point tracking neural networks. In one aspect, a method includes obtaining a video sequence including a plurality of video frames; obtaining a set of one or more query points that each specify a spatial position in a corresponding video frame; generating a plurality of video tokens representing the video sequence; generating a query token for each query point for the corresponding video frame that characterizes the spatial position of the query point; and processing the video tokens and the query tokens for the query points using a neural network to generate a network output that includes a predicted spatial position of each query point in video frames other than the corresponding video frame for the query point.