158 results about "Streaming algorithm" patented technology

The embodiment of the invention provides a scene space model adaptive modeling method based on multi-source data fusion. Complete scene information is constructed by fusing a video sequence, point cloud data and IMU (Inertial Measurement Unit) data. A deep learning network is adopted to carry out scene analysis, a scene complexity score is calculated based on a scale coefficient, a shielding degree and a geometric deformation coefficient, and intelligent matching of a modeling strategy is realized. Meanwhile, a scene dynamic area is predicted by using an optical flow algorithm, a geometric deformation field and a camera pose change, the modeling quality is monitored through reconstruction errors, and self-adaptive switching of a modeling algorithm is realized. The method effectively solves the problems of unstable precision and poor adaptability of a traditional modeling scheme in a complex dynamic scene, and improves the quality and efficiency of three-dimensional scene reconstruction.

Disclosed in the present invention are a helium-neon laser therapy method and application based on adaptive beam focusing. The technical method belongs to the technical field of medical lasers, and realizes adaptive beam focusing optimization for helium-neon laser therapy. The process comprises: patient assessment, which involves acquiring an image and data of a treatment region; image processing and analysis, which involves performing feature extraction and lesion segmentation by using a convolutional neural network (CNN) and a U-Net algorithm; laser parameter setting based on wavefront optimization, which involves adjusting laser parameters by means of a feedback control system; therapy execution, which involves tracking and adjusting a beam in real time by means of wavefront optimization and an optical flow algorithm; result assessment and adjustment, which involves assessing a treatment effect by using the CNN and the U-Net algorithm again, and adjusting the laser parameters for repeated treatments if necessary; and finally, assessing a treatment result for the patient after the treatment is completed. The method improves the accuracy, effect and safety of treatment, and increases the degree of satisfaction of a patient.

The invention discloses a few-sample knowledge graph completion method in the field of numerical control programming. The problems that a numerical control programming template generated under the guidance of a numerical control programming knowledge graph is low in reliability and large in limitation in the actual production process, programming personnel still need to repeatedly modify the numerical control programming template, and training entity samples are few are solved. The method comprises the following steps: firstly, designing an entity neighborhood information aggregation network based on a relation path, evaluating and distributing the importance of neighborhood entities through a multi-head self-attention mechanism, and aggregating features; then, according to the semantic similarity of the entity pairs, modeling a semantic map to represent the semantic relationship between the entities, and outputting global semantic features; and finally combining the neighborhood aggregation features and the semantic features to carry out link prediction. On a public data set complemented by a few-sample knowledge graph, the effectiveness and the accuracy of the algorithm are verified by performing a contrast experiment with a mainstream algorithm.

The invention relates to the technical field of virtual local area networks, in particular to a communication system for dividing a virtual local area network (VLAN) into network cards based on a high-speed signal connection switch. The system comprises a network state monitoring module, a flow prediction module, a resource allocation decision module, a flow control execution module, a behavior pattern analysis module, a VLAN configuration adjustment module, a resource allocation optimization module and a network configuration adaptive adjustment module. According to the method, the software-defined network technology and the time sequence analysis are combined, the flow prediction is carried out by using the anomaly detection, the long-short-term memory network and the convolutional neural network, the bandwidth, the priority and the routing are adjusted through the optimization theory and reinforcement learning, the network flexibility is enhanced, and the strategy deployment efficiency is improved through the software-defined network technology; the recurrent neural network and the long short-term memory network deepen behavior recognition, the VLAN configuration is optimized by using the minimum cut maximum flow algorithm, the resource utilization rate is improved, the network adaptive adjustment adopts the genetic algorithm and simulated annealing, the routing is optimized, the QoS strategy is automatically adapted, and the adaptive capability is enhanced.

The invention is suitable for the technical field of earth surface monitoring, and provides an earth surface deformation monitoring method and system based on a time sequence InSAR, and the method comprises the following steps: carrying out the SAR image screening and interferogram generation, and collecting corresponding meteorological data and thermal infrared data; performing space-time adaptive atmospheric phase correction, taking the air pressure vertical gradient and the temperature anomaly as driving factors of atmospheric delay, and separating an atmospheric phase through a space-time weighted model; dynamic deformation modeling is carried out, and deformation is decomposed into linear deformation and nonlinear deformation; the method comprises the following steps: taking a mining area road network as a geometric constraint, unwrapping a coherent region by adopting a minimum cost flow algorithm, converting an unwrapping phase into sight-line-direction deformation, and calculating horizontal and vertical deformation components in combination with InSAR sight-line-direction deformation and a digital elevation model. The method adapts to a complex deformation mechanism of a mining area by capturing linear and nonlinear deformation. Through sight line deformation and DEM geometric projection, vertical and horizontal deformation separation is realized, and the deformation direction is determined.

The invention discloses an intelligent building equipment control method and system, and relates to the technical field of intelligent building equipment control, and the method comprises the steps: fusing the multi-mode association topology of two groups of data through employing a Kruskal algorithm, constructing a minimum spanning tree, setting the reciprocal of an edge weight as a pheromone value, and selecting an optimal control path through employing a maximum flow algorithm-ant colony algorithm. The method comprises the following steps: converting phase space data by using a Takens embedding method, randomly selecting a pair of adjacent phase space data from a time sequence as initial points, analyzing the state of building equipment by using a Lyapunov index, constructing a hidden Markov model, and dynamically correcting a transition probability matrix and a transmission probability matrix by using the Lyapunov index to obtain the predicted state of the building equipment. The optimal control path is selected by constructing the minimum spanning tree and combining the maximum flow algorithm-ant colony algorithm, the resource utilization rate and the control precision are improved, and the comprehensiveness and adaptability of equipment control are enhanced based on the combination of the hidden Markov model and the Lyapunov index.

The invention provides a bridge structure damage detection method based on image processing and CNN-LSTM, relates to the field of graphic image processing, and solves the problems that bridge vibration monitoring in the prior art is time-consuming, labor-consuming, high in cost and limited in precision, and adopts the technical scheme that the method comprises the steps of obtaining a recorded vibration video of a to-be-detected bridge structure; extracting the vibration video frame by frame to obtain bridge image data; detecting angular points by using a Harris algorithm, screening image data, and calculating vertical displacement of each feature point one by one; using a Lucas-Kanade optical flow algorithm to calculate the motion vectors of the feature angular points of the adjacent frame images; calculating the actual vibration displacement of the bridge; and carrying out standardization processing on the actual vibration displacement to obtain a two-dimensional matrix, inputting the two-dimensional matrix into the trained CNN-LSTM model, and outputting a final identification result of each type of damage. According to the scheme of the invention, high-precision bridge vibration displacement can be calculated, and efficient, accurate and high-precision bridge damage detection can be realized through the CNN-LSTM damage identification model.

The invention provides a circuit board component accurate positioning and mounting method and system based on visual guidance, and relates to the technical field of mounting, and the method comprises the steps: obtaining the multi-angle image information of a to-be-mounted component, determining a three-dimensional point cloud model and a depth map sequence, and determining the type information; segmenting the point cloud model through a region growing algorithm to extract features for pose estimation; the mounting parameters are called and combined with the posture deviation information to control a mechanical arm; and motion parameters are optimized based on a self-adaptive optical flow algorithm until the precision requirement is met. According to the invention, the component mounting precision and efficiency are improved, and the mounting failure rate is reduced.

The invention relates to an adaptive lane detection method based on curvature and edge perception optimization, and belongs to the technical field of automatic driving. According to the method, the adaptability of various scenes is enhanced by introducing an adaptive weight mechanism, complex geometric features are captured by fusing Kolmogorov-Arnold Networks (KAN) convolution, and complex lane shapes and edge details are better processed by combining curvature constraint loss and edge detection loss. Through a series of improvements, good balance between the speed and the detection precision is realized, so that the lane line detection system can efficiently and stably operate in an automatic driving environment with limited resources. In order to verify the effectiveness of the proposed method, wide experiments are performed on a CULane data set. Experimental results show that the F1-score of the KAN-Lane in the curve and shielding scenes is 82.08% and 75.44% respectively, the detection precision and stability are remarkably improved, the method is superior to mainstream algorithms such as CLRNet and the like, and the efficient detection speed is kept. The KAN-Lane provides a more reliable lane detection scheme for automatic driving.

The invention discloses an intelligent data processing method and system for geological disaster early warning, and relates to geological disaster early warning, and the method comprises the steps: collecting multi-source heterogeneous data; extracting topographic features of the geological disaster area according to the optical satellite image data and the laser radar data; performing differential interference processing on the synthetic aperture radar data to obtain differential interference phase data of the geological disaster area; performing phase unwrapping by using an improved minimum cost flow algorithm to obtain a continuous absolute phase value # imgabs0 #; performing nonlinear estimation on the estimation result of the surface deformation quantity to obtain state parameter estimation of the geological disaster area; and performing early warning detection by using a support vector machine (SVM). In order to solve the problem of low geological disaster prediction precision caused by difficult phase unwrapping in the prior art, the unwrapping problem is converted into a space-time network flow graph, space-time continuity constraints are introduced, the incidence relation between data in different periods is considered, a multi-scale strategy is adopted for solving and the like, and the prediction precision is improved.

The invention discloses an underground fluxgate sensor signal processing method, which comprises the following steps of: extracting a magnetic field voltage signal mainly comprising a second harmonic component by adopting a signal detection circuit, and further removing other harmonic components through a digital phase-sensitive rectification and smooth filtering algorithm to realize effective extraction of the magnetic field signal. The second harmonic extraction effect of the phase-sensitive rectification algorithm is verified through a time sequence simulation result, and an open-loop fluxgate sensor designed based on a second harmonic method is experimented. Compared with other types of fluxgate signal detection methods, the second harmonic method effectively improves key performance such as resolution and linearity of the sensor, and the fluxgate sensor designed based on the second harmonic method can realize detection and effective modulation of a detected magnetic field signal and shows a relatively high magnetic measurement level.

The invention provides a sea level rise influence risk assessment method based on multi-source data fusion, a medium and a system, and belongs to the technical field of marine disasters. Basic feature data are obtained by constructing a multi-dimensional data acquisition network, nonlinear time series decomposition is performed by applying multi-scale wavelet analysis, and the risk assessment result is obtained by analyzing the nonlinear time series; establishing a prediction model based on a deep learning neural network; determining key vulnerabilities by using a minimum cut maximum flow algorithm; establishing a regional vulnerability scoring system; generating a comprehensive risk assessment matrix by using a Bayesian network ensemble analysis method; and finally, generating an evaluation report by using the coastline adaptability gating model. According to the model, a multi-head time sequence attention mechanism and bidirectional autoregression coding are fused, parameters are dynamically adjusted by adopting a regional adaptability gating weighting function, and accurate evaluation and dynamic tracking of sea level rising influence risks in a complex geographical environment are realized.

The invention discloses a river flow velocity fluctuation early warning method and system based on an optical flow algorithm. The method comprises the following steps: collecting continuous frame video data; processing the video data through a traditional optical flow algorithm and a deep learning optical flow algorithm, and performing densification processing on the sparse optical flow field data to obtain traditional optical flow densified data; calculating confidence mapping data of the sparse optical flow field data and uncertainty data of the deep learning optical flow field data; obtaining weight map data through a lightweight residual error optimized weight generation network; performing fusion processing on the traditional optical flow dense data and the deep learning optical flow field data according to the weight map data to obtain fused optical flow field data; and calculating flow velocity distribution data of the river surface by fusing the optical flow field data, and outputting flow velocity fluctuation early warning information data according to the flow velocity distribution data. According to the invention, the boundary retention capability of the traditional algorithm and the dense prediction advantage of the deep learning algorithm are fully exerted through the fusion of the two-path optical flow algorithm.

The invention discloses a construction period landslide disaster advanced prediction and rapid networking monitoring method, and the method comprises the steps: collecting synthetic aperture radar data, digital elevation model data and satellite precision orbit data of a monitoring region, carrying out the data preprocessing, and obtaining a single-view complex data image set; constructing an interference image pair connection graph, performing interference processing, and generating a differential interference graph and a coherence coefficient graph; performing interferometric phase unwrapping processing on the differential interferogram and the coherence coefficient graph according to a minimum cost flow algorithm to obtain earth surface deformation information; ground control points are selected, and the surface accumulated deformation quantity and the surface deformation rate are obtained; interpreting and delineating a potential deformation area, establishing an interpretation mark, selecting a typical deformation area, and extracting a time sequence deformation data set; and analyzing the time sequence deformation data set according to the neural network model to obtain a deformation prediction result and a disaster risk assessment result. According to the invention, the monitoring accuracy and prediction capability of landslide disasters can be improved.

The invention discloses a communication transmission method and device based on multipath delay optimization and a storage medium, and relates to the technical field of communication, and the method comprises the steps: collecting link information in an SRv6 network through a routing protocol; calculating a plurality of preset paths through a maximum flow algorithm according to the link information; generating a segment routing table for each path in the plurality of preset paths; obtaining an SRv6 data packet to be forwarded; determining a target routing path from the segment routing table based on the to-be-forwarded SRv6 data packet; and performing data transmission on the to-be-forwarded SRv6 data packet according to the target routing path. The problems that a traditional end-to-end transmission protocol is poor in performance and high in packet loss rate in a long-distance network are solved, a network with committed quality is introduced again, the compatibility to an existing actual network is high, cross-platform extensibility is achieved, and the network transmission quality is improved.

The invention discloses a graph fraud risk identification method and system based on link mode aggregation. The method comprises the following steps: 1) constructing a relational network graph; the method comprises the following steps of (1) data acquisition, (2) variable-length link sampling, (3) position-associated unified link coding, (4) link information interaction and aggregation and (5) aggregation-associated fraud detection.By means of long-distance link sampling and coding and link information interaction and aggregation, the defects of a mainstream GNN algorithm in the aspect of processing a long-range dependency graph structure are effectively overcome.

The invention discloses a residual current detection system and method based on wireless synchronization and software confluence, and relates to the technical field of residual current detection. A host central processing device receives current sampling data with timestamps from a slave central processing device through a signal receiver to complete centralized receiving of multiple paths of current signals; after the host central processing device receives synchronous current sampling data with timestamps transmitted by the slave central processing device, aligning and synthesizing multiple paths of time-synchronized current signals through a software confluence algorithm, carrying out vector summation on the multiple paths of current signals based on a vector operation principle, and calculating the vector sum of residual current; and judging whether a grounding fault or an insulation defect exists according to whether the vector sum is zero, if the vector sum deviates from the zero value, judging that residual current abnormity exists, and giving an alarm or recording fault information. The detection system effectively improves the accuracy, reliability and applicability of residual current detection.

The invention discloses a YOLOv8-based SAR (Synthetic Aperture Radar) image ship target detection method, and aims to solve the problems that a part of ship targets in a synthetic aperture radar image are small in size, large in target size change, complex in background noise and the like. Wherein the multi-scale extraction backbone network uses an optimized C2f module to enhance multi-scale feature extraction and fusion, and designs an SPA module to enrich gradient flow information, thereby improving the detection effect of different scale targets. And the neck is replaced by lighter dynamic up-sampling, so that the calculation overhead and the model complexity are reduced, and meanwhile, the recognition of small and medium-sized ships with complex backgrounds is optimized. And a multi-dimensional attention mechanism and pruning optimization are added in the design of the detection part, so that the sensitivity to key features in a complex background of the model is enhanced, and the detection accuracy is improved. Experiments are carried out on two public data sets of HRSID and SSDD, the mAP50 of 92.2% and 98.5% are realized respectively and are improved by 2% and 2.2% respectively compared with a baseline, and the complexity of a DD-YOLO model is lower than that of a current mainstream algorithm.

The invention relates to the technical field of power systems, in particular to a flexibility resource evaluation method, device and equipment of a power distribution network, a medium and a program, and the method comprises the steps: obtaining a network topology structure and network electrical characteristics of the power distribution network; determining node flexibility resources and network flexibility resources according to the network topology structure; determining an allowable adjustment range of flexibility resources of various devices in the power distribution network according to the node flexibility resources and the network electrical characteristics; determining network transmission capacities of various devices in the power distribution network according to the network flexibility resources and the network electrical characteristics; and inputting the allowable adjustment range of each device and the network transmission capacity into a maximum flow algorithm, and evaluating the target adjustment range of the flexibility resource of each node of the power distribution network based on the maximum flow algorithm. Therefore, the problems of large resource and time consumption, relatively low processing efficiency, relatively low accuracy of an evaluation result and the like of quantitative calculation of the network flexibility of the power distribution network in related technologies are solved.

The invention relates to the technical field of electric connector weak light assembly vibration compensation alignment methods, particularly discloses an electric connector weak light assembly vibration compensation alignment method, and aims to solve the problems of low electric connector assembly positioning precision and high pseudo soldering rate caused by vibration in a weak light environment. The method comprises the following steps: monitoring the thermal radiation change of a vibration source in real time through an infrared thermal imaging sensor, and pre-judging a vibration acceleration vector in combination with a lightweight long-short-term memory neural network; synchronously triggering dual polarization angle imaging to respectively capture a housing reference mark and an enhanced terminal metal feature; the vibration data and the polarization image flow are fused in the field programmable gate array, a self-adaptive optical flow algorithm is adopted to resolve the displacement compensation amount, and a piezoelectric ceramic actuator is driven to carry out submicron order reverse displacement compensation; when the compensation amount exceeds a threshold value, a low-delay correction instruction is sent to the robot through an Ethernet control automation technology bus, and the steady-state error is ensured to be smaller than 0.3 micrometer by combining a strain gauge fine tuning feedback loop. According to the technical scheme, the assembling precision, efficiency and production line adaptability are remarkably improved.

The embodiment of the present application provides a method for adaptive modeling of scene space model by fusion of multi-source data, which constructs complete scene information by fusing video sequences, point cloud data and IMU data. A deep learning network is used for scene analysis, and the scene complexity score is calculated based on the scale coefficient, occlusion degree and geometric deformation coefficient to achieve intelligent matching of modeling strategies. At the same time, the dynamic area of ​​the scene is predicted by using the optical flow algorithm, geometric deformation field and camera pose change, and the modeling quality is monitored by reconstruction error to achieve adaptive switching of the modeling algorithm. This method effectively solves the problems of unstable accuracy and poor adaptability of traditional modeling solutions in complex dynamic scenes, and improves the quality and efficiency of three-dimensional scene reconstruction.

The invention provides a reservoir water rain condition monitoring system and a monitoring method, the system constructs a dual monitoring system of physical observation reference and electronic automatic acquisition, a customized stainless steel water gauge and a multi-element acquisition device are deployed on key sections of a reservoir dam and a drainage basin, and a front-end edge calculation module is utilized to invert the flow in real time. And after the data is transmitted to the cloud platform, intelligent management and control are carried out through the four-pre functional module. The system displays the watershed situation based on a GIS graph, integrates an online flow measurement algorithm to generate a water and rain condition process line in real time, and realizes visual supervision by combining video monitoring and OSD data superposition. The core multi-target optimization scheduling model comprehensively considers flood control safety and power generation benefits, generates an optimal scheduling scheme through simulation deduction, and provides a closed-loop control interface. The problems of low monitoring precision, scheduling decision lag and the like in the prior art are solved, and the automatic management level and the water resource utilization rate of the reservoir group are remarkably improved.

The invention provides a time sequence data fragment master copy selection method, device, equipment, medium and product, and the method comprises the steps: mapping a cluster node set and a cluster fragment set into vertexes of a flow network, adding a source point and a sink point into the flow network, and constructing the flow network; wherein the cluster fragment set comprises a plurality of time sequence data fragments, and each time sequence data fragment comprises a master copy and a slave copy; solving a minimum cost and maximum flow result in the flow network by using a minimum cost and maximum flow algorithm; the minimum-cost maximum-flow result comprises a node where each time sequence data fragment primary copy in the cluster fragment set is located. According to the scheme provided by the invention, the most balanced cluster master copy distribution is realized, so that the demand that an industrial Internet of Things production environment bears dense computing loads is met.

The invention relates to the technical field of production scheduling and intelligent manufacturing, and provides an assembly line scheduling optimization method considering generalized priority constraints, which comprises the following steps: constructing an integer linear programming model, performing generalized priority relation enhancement and propagation on an input instance of the integer linear programming model, and improving the capacity of a workstation; calculating a simple lower bound based on task processing time, a lower bound based on the longest path of the priority graph and a lower bound based on a maximum flow algorithm, and taking the maximum value of the three as a final lower bound; defining an action space, defining a state space to represent the change of a current solution on an objective function value and a distribution balance degree relative to a global optimal solution and a local optimal solution, and designing a reward function to adaptively select a neighborhood operation; constructing an initial scheme based on the final lower bound and a Q-learning dynamic selection module, and performing batch movement iteration through neighborhood operation adaptively selected by the Q-learning dynamic selection module; and outputting an assembly line scheduling scheme meeting the generalized priority constraint based on a result of the batch movement iteration local search.

The invention provides a source network load storage cooperative control optimization method based on an improved selective transmission algorithm, and belongs to the technical field of power system dispatching. An energy storage device equivalent circuit model considering the temperature effect is established, and the state of charge is estimated in real time through Kalman filtering; a power system is modeled into a directed graph network flow model to set power balance constraints, an improved genetic algorithm based on tent chaotic mapping is adopted to solve an optimization model, a network flow algorithm is embedded in an iteration process to check feasibility, and premature convergence is prevented by adaptively adjusting crossover probability and mutation probability. And finally, outputting a Pareto optimal solution set, and selecting a scheduling scheme with optimal comprehensive performance according to an ideal point method. The technical problems of high system operation cost and insufficient new energy consumption rate caused by the fact that multi-objective optimization solution is easy to fall into local optimum in the source network load storage cooperative control optimization process are solved.

The invention discloses a non-linear constraint coupling driven implicit modeling optimization method and system and a memory. The method comprises the steps of firstly obtaining nonlinear constraint data of a target geological interface, then generating initial implicit function representation through a Hermite radial basis function based on an initial control point of the target geological interface, and then obtaining a nonlinear coupling driving optimization objective function about implicit function representation of the target geological interface according to the data type of nonlinear constraint, and optimizing through an implicit gradient flow algorithm to obtain a geological interface reconstruction model and a global optimization target geological interface model. The invention further provides a computer readable memory, the memory contains a computer program for realizing the implicit modeling optimization method for nonlinear constraint coupling driving, the computer program can be read and executed, and the computer program, a programmable logic controller and a processor form an optimization system. According to the system, nonlinear constraint data of a large number of geological interfaces are fully coupled, an implicit modeling framework is constrained, and therefore high-precision modeling of the three-dimensional geological interfaces is achieved.

The invention provides an e-commerce brand promotion system based on social network influence propagation, and belongs to the technical field of e-commerce brand promotion systems. The system comprises a data acquisition module, a node identification module, a propagation dynamics model construction module, a promotion strategy generation module and an execution and feedback module. The data acquisition module obtains social and e-commerce data in a compliance manner and carries out standardization processing. The node identification module screens and classifies KOL through a multi-dimensional index, and determines potential propagation nodes in combination with K-means clustering; the propagation model module optimizes a'social network-SEIR 'model based on SEIR, and selects an optimal path by using a Dijkstrar + maximum flow algorithm; the strategy module generates a hierarchical KOL cooperation and user incentive scheme; the feedback module collects data per hour, dynamic adjustment is performed within 24 hours, and the method is executed in five steps corresponding to five modules. 30 days after application, the brand popularity is improved by more than or equal to 60%, the cost waste rate of small and medium-sized brands is reduced to 15% or below, the market sinking reach rate is improved to 38%, and popularization requirements of industries such as beauty makeup and the like are met.

The invention discloses a wire harness coating defect detection method and system based on visual identification, and relates to the technical field of visual detection. According to the system, aiming at the defect positioning problem caused by bending, torsion and illumination variation and shielding in detection of a wire harness in a complex form, an enhanced feature map capable of capturing wire harness space information and potential defects is generated by synchronously acquiring a multi-view image sequence, adopting a convolutional neural network to extract features and combining with normalization processing. According to the method, the central axis of the wire harness is tracked through iterative fitting, a unified axis model is constructed by fusing multi-view data through an optical flow algorithm, the radial offset of a defect area is calculated based on the model so as to carry out preliminary positioning, and finally, accurate three-dimensional reconstruction is carried out on the defect by adopting multi-view geometric constraint. According to the method, through multi-view information fusion and three-dimensional geometric optimization, the precision and robustness of wire harness defect detection in a complex environment are remarkably improved, and reliable spatial position data support is provided for automatic processing.

The invention discloses a video analysis-based early warning and monitoring method for children hyperactivity behavior and a storage medium, and the method comprises the steps: firstly, guaranteeing the comprehensive monitoring of children behaviors through employing a multi-channel video collection device, and providing a sufficient continuous data source for subsequent analysis; furthermore, the quality of monitoring data is ensured through the use of image noise reduction and segmentation technologies, and accurate motion features are effectively extracted through the combination of an optical flow algorithm and a background subtraction algorithm; further, the movement behaviors of the children are comprehensively analyzed by calculating behavior parameters such as movement amplitude, movement direction change and continuous movement time, and potential behavior abnormity is revealed; the multi-dimensional behavior feature vector obtained through behavior parameter analysis is compared with the standard template, the behavior type of the child is automatically recognized, early-stage symptoms of abnormal behaviors such as hyperactivity and the like are found in time, and a basis for early diagnosis and intervention is provided for behavior problems such as hyperactivity and the like of the child.

The application discloses a kind of based on dynamic receptive field and adaptive feature fusion plate defect detection deep learning method, to improve the detection precision and robustness of complex background and small scale defect under industrial environment.The proposed method is based on YOLOv10, dynamic receptive field module and adaptive feature fusion module are introduced in its architecture.Dynamic receptive field module uses local features and context information to dynamically adjust the size of convolution receptive field, to enhance the perception ability of different scale target;Adaptive feature fusion module realizes the weighted fusion of multi-layer features through multi-layer perception mechanism, improves feature expression ability and target recognition performance.The experimental results show that the method of the application is superior to existing mainstream algorithms in detection accuracy and real-time performance, has good engineering adaptability and industrial application prospect, and is suitable for furniture manufacturing, building material processing and other plate detection scenarios.