15 results about "Object dynamics" patented technology

The invention discloses an attitude estimation method based on multi-class object dynamic key point learning and progressive optimization. The attitude estimation method comprises the following steps: carrying out outlier filtering on dense point clouds converted from depth images by adopting 3D image convolution to obtain filtered point clouds; obtaining key points and feature representation of the object by using the filtered point cloud coordinates, the RGB features of the object and the geometric features of the filtered point cloud through a multi-modal key point querier; dense point cloud geometric features are extracted by using a dynamic graph convolutional network, and in combination with local multi-level features based on key points and a feature reconstruction model, the feature expression ability in a complex scene is enhanced by using attention weighted fusion; and finally, through a progressive optimization module, on the basis of a pose residual prediction network and a parameter dynamic updating mechanism, object pose information is output after three times of iterative computation. According to the method, a dynamic key point learning mechanism based on three-dimensional geometric feature guidance is constructed, and the generalization ability and geometric perception precision of attitude estimation are effectively improved by fusing multi-modal perception information and an iterative optimization strategy.

The embodiment of the application provides a kind of method and system based on the object dynamic generation of reflection mechanism, it is related to the technical field of C++ programming technique.The method comprises the following steps: establishing the mapping relationship between multiple class identifiers and multiple creator functions;Receive configuration file, and determine at least one object definition item based on the configuration file;According to the class identifier contained in the target object definition item, retrieve the corresponding creator function from the mapping relationship;According to the construction parameter, call the creator function to obtain the object instance corresponding to the instance identifier contained in the target object definition item.Through the application, the problem of low efficiency of instance object development is solved, and the effect of improving the efficiency of instance object development is achieved.

The invention discloses an object dynamic deformation monitoring method and system, and relates to the field of deformation monitoring, and the method comprises the steps: collecting multi-frame images of a to-be-measured object at different moments; target segmentation identification is carried out on each frame of image, and the mask position of the to-be-detected object in each frame of image is determined; determining the pixel-level displacement of the to-be-measured object according to the mask position of the to-be-measured object in each frame of image; and converting the pixel-level displacement into displacement under a real coordinate system so as to determine the deformation of the object to be measured. According to the invention, dynamic deformation monitoring is realized through an image processing method, and the precision and real-time performance of dynamic deformation monitoring of an object in a close-range environment can be improved.

Systems and methods for unsupervised multi-object scene decomposition that involve a spatio-temporal amortized inference model for multi-object video decomposition. Systems and methods involve a new spatio-temporal iterative inference framework to jointly model complex multi-object representations and the explicit temporal dependencies between the frames. Those dependencies improve overall quality of decomposition, encode information about object dynamics and can be used to predict future trajectories of each object separately. Additionally, the model can generate precise estimations and output data even without color information. The model has scene decomposition, segmentation and future prediction capabilities. The processor can use the model to simulate future frames of the scene data.

The invention relates to the technical field of abnormal object recognition, in particular to an abnormal object dynamic behavior pattern recognition method based on a graph neural network, a medium and equipment, and the method comprises the steps: enabling the feature representation of each node to be capable of continuously fusing the weighted information of multi-order neighbors through a multi-layer iteration and edge weight-based message transmission mechanism, and enabling the feature representation of each node to be more accurate. Therefore, local and direct association is expanded into global and potential network mode representation, and behavior collaboration of direct / indirect association objects is captured. Similarity calculation and clustering analysis are performed on the Nth-layer intermediate feature vector generated in the graph neural network, and nodes with similar behavior patterns and similar structure roles in the network can be automatically collected into a cluster, so that individual anomaly can be judged, and the probability of individual anomaly can be reduced. And accurate locking and mode classification of the whole hidden collaborative abnormal gang which has no strong direct connection but highly collaborative behaviors can be realized, so that the accuracy and comprehensiveness of an identification result are greatly improved.

The invention provides an object dynamic change monitoring method and system based on an unmanned aerial vehicle image, and relates to the technical field of image analysis. The method comprises the steps of obtaining an image data set of a target area, performing target tracking processing to obtain a plurality of track fragments, determining a plurality of scene structure components, performing rhythm analysis on the target area, and constructing rhythm portraits of the plurality of scene structure components. The method comprises the following steps: acquiring real-time monitoring data of a target area, performing anomaly detection, generating structural behavior anomaly detection data of the target area, constructing an interaction graph and a dynamic flow constraint model of the target area, extracting behavior density data and behavior blocking data of the target area, identifying a plurality of local abnormal flow areas of the target area, and generating a dynamic flow constraint model of the target area; and performing diffusion influence identification on the local abnormal flow area to determine a plurality of flow abnormal areas, and fusing the structure behavior anomaly detection data to generate a dynamic change monitoring result of the target area. According to the method, the real-time sensing capability of progressive risk accumulation in the dynamic behavior of the object is improved.

The present invention discloses a master-slave unified admittance control method for a dual-arm robot for coordinated tasks, including: establishing a reference coordinate system and a transformation matrix between coordinate systems, and obtaining the expected end motion trajectory and the master-slave arm motion model of the dual-arm robot by combining closed-chain and open-chain constraints; decomposing the resultant force received by an object and introducing a grasping matrix to obtain the object dynamic equation during the coordinated motion, and combining with admittance control to obtain the trajectory parameters output by the admittance control algorithm, and designing a master-slave unified admittance control UAMSC algorithm; based on the expected end motion trajectory and the master-slave arm motion model, applying the UAMSC algorithm, establishing a cooperative control strategy for the dual-arm robot for different tasks and executing it to maintain the stability of the internal force of the system under the conditions of human-robot cooperation and the existence of external force interference. The present invention can meet the requirements of human-robot cooperation and suppression of external force interference when the dual-arm robot executes coordinated tasks, and can be applied to scenarios such as cooperative handling by the dual-arm robot.

The present invention provides a method, device, equipment and storage medium for dynamic solution of an object. The method obtains the model elements of the object, determines the locking target point of the object based on the model elements, creates multiple nodes based on the locking target point and the world coordinate system, and constructs a first node tree associated with the locking target point at the level of the multiple nodes and a second node tree associated with the world coordinate system. When the object moves, the transformation data of the third node in the second node tree at each time point is collected, and key frames are baked based on each transformation data, and animation data of the object in motion is generated based on each key frame. In this way, after creating multiple levels of nodes, it is only necessary to bake the data of the child nodes to achieve complete locking of the transformation data of the moving object points, which solves the problem of character special effects solution being exposed when the existing point locking solution solves character special effects with excessive changes.

The application relates to the technical field of abnormal object recognition, in particular to an abnormal object dynamic behavior pattern recognition method based on a graph neural network, a medium and equipment, through a multi-layer iteration and edge weight-based message passing mechanism, feature representation of each node can continuously fuse weighted information of multi-order neighbors, thereby expanding local and direct correlation into global and potential network mode representation, and behavior coordination of directly / indirectly correlated objects can be captured; through similarity calculation and clustering analysis on the Nth layer intermediate feature vector generated by the graph neural network, nodes with similar behavior patterns and similar structure roles in the network can be automatically collected into a clustering cluster, so that not only individual abnormalities can be determined, but also the whole implicit cooperative abnormal gang without strong direct connection but with high behavior coordination can be accurately locked and pattern classified, and the accuracy and comprehensiveness of the recognition result are greatly improved.

The invention discloses an intelligent parking lot safety monitoring method and system based on machine vision, and relates to the technical field of intelligent safety monitoring. The method comprises the steps that a parking lot image set on a time sequence is collected firstly, an effective image set capable of reflecting key changes is obtained through effective image extraction, redundant data are reduced, and the follow-up processing efficiency is improved; preprocessing the effective image set, and optimizing the image quality; identifying an object by using an enhanced YOLOv8 model to obtain an identified object information set; then determining a time sequence behavior sequence of each identified object to facilitate comprehensive tracking of object dynamics; abnormal judgment is carried out according to the behavior sequence, and potential safety risks can be found in time; and the abnormal object is labeled and compared with the database to obtain a monitoring result, so that efficient safety early warning can be realized. According to the scheme, data value mining and processing efficiency are considered through effective image set construction, and risk identification accuracy is enhanced according to the enhancement model and behavior analysis, so that safety monitoring efficiency is improved.

The invention relates to a fluid dynamic system high-precision prediction method fusing a differential equation and a neural network, belongs to the technical field of fluid dynamic system prediction, and solves the problem that in the prior art, high-precision prediction of fluid dynamic system behaviors cannot be realized by purely depending on the neural network. The method comprises the following steps: analyzing dynamic behaviors of a fluid dynamic system, and constructing an object dynamic behavior simplified model in the fluid dynamic system; initializing system parameters and external excitation parameters, and carrying out numerical solution solving on the object dynamic behavior simplified model to obtain numerical solution time sequence data of object displacement and speed in a fluid dynamic system of a sampling time sequence; and constructing training data of the neural network model for predicting the dynamic behavior of the object based on the numerical solution time sequence data of the displacement and the speed and the external excitation parameters, and predicting the displacement and the speed of the object in the fluid dynamic system at the to-be-predicted time based on the trained neural network model for predicting the dynamic behavior of the object.

The application discloses a kind of multi-target operation object dynamic selection method and system, it is related to computer vision and intelligent video analysis technical field, specifically includes the following steps: S1, obtains the continuous video frame sequence that industrial assembly line monitoring equipment gathers;S2, object detection and tracking are carried out to the video frame, and the object set containing object detection frame, category label and track ID is output;S3, based on the spatial position relationship between object, contact state and function association, construct combination relation graph, identify single body object and the combination target formed by multiple objects.The multi-target operation object dynamic selection method and system, by object combination relationship modeling, multiple object combinations formed operation unit (such as "cover+box", "bolt+nut") can be identified, avoid to combine target error split into multiple independent objects;In automobile parts assembly scene test, the combination target identification accuracy reaches 94.2%.

The invention relates to an anomaly detection method and device, electronic equipment and a storage medium, and relates to the technical field of artificial intelligence, and the method comprises the steps: receiving an anomaly detection instruction for an object recognition operation of a target object under the condition that an object recognition anomaly occurs in the object recognition process of the target object; based on the anomaly detection instruction, obtaining an object key point data packet corresponding to the object recognition operation; based on the key point position information, performing object identification dynamic reduction on the target object to obtain object dynamic data corresponding to the target object; the object dynamic data is used for indicating the action execution condition of the target object in the object recognition operation process; based on the identification dynamic prompt information and the object dynamic data, generating an action execution result corresponding to the target object; and generating an abnormal attribution recognition result corresponding to the target object according to the action execution result. According to the embodiment of the invention, the reasons for abnormal object recognition can be accurately distinguished, so that the success rate of object recognition can be improved.

The embodiments of this specification provide an object generation method, a 4D object generation method, and a video generation method. The object generation method includes: determining object generation parameters and multiple sets of object images corresponding to multiple time points, where the multiple time points are arranged in chronological order, and each set of object images includes object images of a target object from multiple perspectives at the corresponding time point; inputting the object generation parameters and the multiple sets of object images into an object generation model to obtain a dynamic object corresponding to the target object, where the object dimension of the target object is smaller than the object dimension of the dynamic object, and the object generation model processes the multiple sets of object images according to the object generation parameters, determines multiple object dynamic data corresponding to the multiple perspectives, and generates the dynamic object according to the multiple object dynamic data.