352results about How to "Increase awareness" patented technology

The invention provides an intelligent garden environment monitoring data analysis method and system. Sensors are deployed to collect temperature, humidity and other data in real time to form a data set. Analyzing an environment parameter change mode of each region by adopting a time sequence clustering algorithm, and determining characteristics and trends; and carrying out dimension reduction on the data by using a multi-dimensional scale transformation algorithm to a three-dimensional space for visualization, and generating an environment state map. And evaluating the plant fitness in combination with a plant physiological ecology knowledge base, and generating a scoring matrix to guide the preliminary layout scheme. And dynamically adjusting the layout by applying reinforcement learning, selecting an optimal scheme, and predicting a future environmental evolution path based on the optimal scheme and historical data. An automatic control system framework is constructed, a self-adaptive control algorithm is designed, intelligent regulation and control of plant layout are realized, garden management efficiency is improved, and ecological protection and sustainable development are promoted. Therefore, full-chain intelligent management from environment monitoring to plant layout optimization is realized, and the intelligent level of garden environment monitoring and management is remarkably improved.

The invention provides an urban underground space high-precision detection method, system and device and a storage medium, and relates to the technical field of geological exploration. The method comprises the following steps: acquiring reflection seismic data, high-resolution electromagnetic response data, environmental noise data, well-side seismic data and logging data of a target area, and performing refined processing on the reflection seismic data, the high-resolution electromagnetic response data, the environmental noise data, the well-side seismic data and the logging data to obtain multi-source multi-scale detection data; performing constraint condition extraction based on the multi-source multi-scale detection data to obtain a multi-source constraint condition, and constructing an urban underground space three-dimensional information model based on the multi-source constraint condition and the multi-source multi-scale detection data; and performing data mining based on the urban underground space three-dimensional information model to obtain three-dimensional geological data. According to the scheme, the accuracy of urban underground space detection can be improved.

The invention discloses an establishment method of a rock multi-field coupling model and the model thereof, and belongs to the technical field of engineering geological survey. Through comprehensive application of multi-scale modeling, a nondestructive testing technology, a machine learning algorithm and intelligent optimization, the problems that existing multi-physical field coupling analysis is not deep enough, the precision of an observation method is poor, and a device is prone to being damaged are solved, the advantages of various observations are combined, the building and analysis precision of a rock multi-field coupling model is improved, and the rock multi-field coupling analysis precision is improved. The method has the advantages that the method is simple and convenient to operate, the intelligence and automation level of the model is enhanced, microstructure data and macroscopic response are combined, deeper understanding of rock behaviors is provided, meanwhile, more accurate and efficient decision support can be provided for rock engineering design and construction, and the method is worthy of popularization and application. And a novel analysis and simulation method is provided for the field of rock mechanics research. Through the comprehensive scheme, behaviors of rocks under complex geological conditions can be better predicted and managed, so that the safety and economical efficiency of engineering are improved.

The invention discloses a knowledge enhancement-based neurogenic bladder personalized health management method and system, and relates to the technical field of chronic disease management, and the method comprises the steps: obtaining the illness state information of a neurogenic bladder patient, and obtaining text representation and knowledge graph representation according to the preprocessed illness state information; generating disease condition comprehensive information representation of the patient through splicing and fusion; constructing a chronic disease management scheme library of the neurogenic bladder, performing graph processing on the chronic disease management scheme library, constructing a health management scheme recommendation model based on a graph neural network, taking disease condition comprehensive information representation of the patient as model input, and generating a personalized health management scheme of the neurogenic bladder patient; and according to the personalized health management scheme, regularly performing scheme evaluation, and based on an evaluation score, judging the applicability of the personalized health management scheme. According to the method, the condition characterization of the patient is enhanced by means of massive information knowledge of the knowledge graph, and intelligent recommendation and construction of personalized schemes of chronic diseases of neurogenic bladder patients are enriched.

The invention provides an artificial intelligence land biodiversity real-time on-line monitoring evaluation-risk early warning-risk elimination method and system, and the method comprises the steps: collecting and monitoring biological information, geographic and phenological information, environment quality information and other information, carrying out the data comparison and analysis, carrying out the data evaluation and influence factor analysis, and carrying out the risk diagnosis and early warning. A risk elimination method; the system sequentially comprises a data acquisition and monitoring module, a data analysis and diversity evaluation module, a risk diagnosis and early warning module, a risk elimination module and the like. Scientific basis and technical support are provided for land biodiversity protection, management and decision making, and the land biodiversity protection strength is enhanced.

The invention discloses a multi-time-sequence three-dimensional visualization expression method for flood routing, and the method comprises the steps: analyzing NetCDF data of a two-dimensional hydrodynamic model, building an element index library and a space library, and storing the indexes of water level, water depth, flow velocity and the like and the space information of an irregular triangular mesh. And generating an association relationship between the grid points and the element indexes, and creating a time sequence cache attribute table and a spatial data set. And reading three-dimensional point space data and element index data, constructing a three-dimensional space grid, and performing range cutting by using a two-dimensional surface data set. And generating a time sequence service conforming to a three-dimensional model tile data format by adopting a thread pool technology. And loading a three-dimensional service and a terrain service at a browser end, carrying out color and height rendering based on the element values, and dynamically displaying the flood process. And a timer is arranged to realize three-dimensional visual playing moment by moment. The method has the advantages that the dynamic display capability of flood routing is effectively improved, and efficient three-dimensional visual rendering and real-time playing are supported.

The invention provides a fabricated building intelligent management method and system based on a BIM model, and the method comprises the steps: installing a sensor at a collection point, and collecting the position, angle, size and other sensing data in real time; preprocessing the collected sensing data, and carrying out cross validation on the preprocessed sensing data; calculating an actual error of each acquisition point of the preprocessed sensing data according to a cross validation result and a target error range, and generating an error analysis report; the result of the error analysis report is compared with the BIM model, and the difference between the actual component and the design model is displayed; and according to a comparison result, a correction or adjustment measure is formulated for the acquisition points of which the errors of each error analysis report exceed an acceptable range. The system comprises an equipment layout module, an error analysis module and an adjustment and correction module. The method plays an important role in improving the construction efficiency, quality and safety; and a closed loop is formed through data acquisition, analysis and feedback.

The invention discloses a target voice extraction method based on multi-reference clue fusion, and the method comprises the steps: enabling a to-be-extracted multi-channel mixed voice to pass through a feature extraction module, calculating the complex spectrum and spatial features of the mixed voice, and obtaining a mixed voice feature map; enabling the azimuth information and the reference voice of the target speaker to pass through an embedded feature extraction module, and combining the direction features and the voiceprint features to obtain embedded features of the target speaker; inputting the mixed speech feature map and the embedded features of the target speaker into a main extraction network based on a dual-path Fastform neural network to obtain high-dimensional feature representation of the speech to be extracted; the high-dimensional feature representation is converted into a plurality of masks for estimating the separated voice through a mask estimation module, and the masks act on a plurality of spectrums of the mixed voice signals; estimating and extracting a time domain waveform of the speech through inverse short-time Fourier transform reconstruction; and the extracted voice of the target speaker is input into the comparison learning module, so that the correct extracted voice can be output when the mixed voice is absent by the target speaker.

The invention relates to the technical field of traffic engineering, and discloses an urban air traffic vertical take-off and landing field site selection optimization method based on demand maximization, and the method comprises the following steps: S1, obtaining and preprocessing online car-hailing original data; s2, dividing grids for the research area range, and taking the grid center as an alternative address of a vertical take-off and landing field; s3, screening out online car-hailing travels which are impossible to turn to urban air traffic; s4, constructing a vertical take-off and landing field site selection model based on potential demand coverage maximization; s5, outputting the selected vertical take-off and landing field address and the corresponding urban air traffic travel demand; and S6, outputting the current optimal site selection and demand as a final solution of the model. According to the method, under the condition of the specific number of vertical takeoff and landing fields, the vertical takeoff and landing field addresses which enable potential requirements of urban air traffic to achieve maximum coverage are selected, an effective solution is provided for development of the urban air traffic, and construction of a more efficient, convenient and environment-friendly urban traffic system is facilitated.

The invention relates to the technical field of network security, and discloses a deception defense system design method based on heterogeneous fusion, and the method comprises the steps: receiving a network access request sent by a client, carrying out the recognition of the network access request through an attack recognition model, and obtaining a recognition result, wherein the attack recognition model comprises a data processing module and an attack recognition module; when the network access request in the identification result is identified as a normal request, forwarding the network access request to the simulation model; when the network access request in the recognition result is recognized as an attack operation request for service target defect utilization, key parameters and loads are extracted from attack operation to automatically initialize a remote control execution environment, and then an attack behavior is redirected to a high-interaction remote control execution environment unit to be executed; according to the method, automatic and intelligent simulation of a complex service system is realized, deep trapping of highly-hidden unknown attacks is formed, and the problem that highly-hidden unknown network attack threats are difficult to find and capture can be effectively solved.

The invention relates to the technical field of carbon emission evaluation, in particular to a BIM-based carbon emission dynamic evaluation and prediction system, which comprises a data extraction and matching module, a carbon emission calculation module, a full life cycle analysis module and a result visualization module. According to the method, data of the building materials are automatically extracted and are in butt joint with the carbon emission database, so that the capacity of accurately matching the attributes and the number of the materials is achieved, calculation of the carbon emission is more detailed, the method further has the personalized analysis capacity, each material and each construction activity are independently calculated, more detailed data support is provided, and the calculation efficiency is improved. Carbon emission of the whole life cycle is monitored and predicted in real time, the practicability of data is enhanced, environmental changes can be responded more quickly, effective carbon emission reduction measures are implemented, a data visualization technology is applied, carbon emission information becomes more intuitive and understandable, information propagation and application efficiency is enhanced, and the cognition and participation degree of the public and professionals to environmental protection problems is improved.

The invention relates to the technical field of equipment state evaluation, in particular to an evaluation method and system for the health state of offshore operation equipment. The method comprises the following steps: acquiring component data of offshore operation equipment; performing initial life marking on the data of the offshore operation equipment component to generate initial life marking data of the offshore operation equipment component; performing real-time operation data acquisition on the data of the offshore operation equipment component to generate real-time operation data of the offshore operation equipment component; performing component division on the real-time operation data of the components of the offshore operation equipment based on the components of the initial life mark data to generate novel component operation data and old component operation data; performing environment change influence analysis on the operation data of the novel part to generate an external change influence factor of the novel part. According to the invention, through multi-dimensional data acquisition, environment and electrical influence analysis, fatigue damage assessment and intelligent prediction, the accuracy of health state assessment of the offshore operation equipment is improved.

The invention provides a permanent magnet synchronous motor electrical fault diagnosis method, and relates to the technical field of motor fault diagnosis. Comprising the following steps: S1, extracting historical electrical fault key characteristic quantity sets of a plurality of motors; s2, performing corresponding classification division on the plurality of historical electrical fault key characteristic quantity sets and different motor operation state categories to form a plurality of historical data sample sets; s3, calculating the prior probability of each motor operation state category; s4, extracting a newly collected electrical fault key characteristic quantity set to obtain a posterior probability of each motor operation state category; s5, supplementing and updating the characteristic quantity data to obtain the posterior probability of each motor operation state category; and S6, labeling the result, and storing the labeled information into a sample information base to update the historical data sample set. According to the method, the problem that most of current fault diagnosis methods cannot well meet the fault diagnosis requirement under the background of big data is solved.

The invention discloses an underground water level prediction method based on a KAN network architecture. The underground water level prediction method comprises the following steps: collecting underground water level historical data and underground water level influence factor historical data as original data; using LASSO regression to screen out relatively independent variables having influences on the underground water level for predicting the underground water level; obtaining a training set and a verification set; building an underground water level prediction model based on a KAN network architecture, wherein the underground water level prediction model comprises an LSTM layer, a Dropout layer and an output layer of the KAN network; training and optimizing an underground water level prediction model based on a KAN network structure to obtain an optimal parameter combination; and taking the optimal parameter combination as a model parameter to obtain a current underground water level prediction model based on the KAN network structure, inputting the fitted verification set data into the current underground water level prediction model based on the KAN network structure, and using the verified current underground water level prediction model based on the KAN network structure for underground water level prediction. According to the method, the underground water level is predicted.

The invention provides a psychological semantic scene image generation method, which comprises the following steps of: 1, processing data according to a psychological scale filled by a user to obtain a three-level level score of a psychological factor; 2, constructing a basic semantic concept group through hierarchical semantic mapping; 3, constructing structured cue words, wherein the structured cue words comprise a text cue word project and an image cue word project; 4, generating an image note through a large language model according to the text cue word, and generating a background image and page elements through a text-to-graph model according to the image cue word; 5, generating a semantic scene graph from the background picture and the page elements generated in the step 4 through a differentiable graph synthesizer; and step 6, generating a final psychological report in combination with the generated graph annotation and the semantic scene graph.

The invention relates to the technical field of equipment state monitoring, in particular to a data prediction model subdivision method and system based on hydropower station equipment states, and the method comprises the following steps: collecting the operation data of hydropower station equipment, sorting the operation data according to a time sequence, and removing the noise and abnormal values of the data; and obtaining a processed time sequence data set. According to the method, trend and periodic change in data and statistical characteristic data input into the hidden Markov model can be effectively identified by calculating the average value and the standard deviation of equipment operation, the dynamic relation between the operation states of the equipment is disclosed through state transition analysis, cognition of equipment early warning and fault stages is enhanced, and the reliability of the equipment early warning and fault stages is improved. The calculation of the real-time state probability not only predicts the health condition of the equipment in the next week, but also can accurately evaluate the severity of the alarm stage, provides a feasible maintenance plan, prolongs the service life of the equipment, and facilitates the formulation of an energy-saving operation strategy.

The invention discloses a multi-dimensional police and civilian joint defense fraud data analysis method and platform, and the method comprises the following steps: training an analysis model, constructing an automatic abnormality detection and early warning mechanism, and building a real-time monitoring and early warning system; carrying out regular fraud model display, and simulating the fraud model to carry out directional training test; obtaining an abnormal access behavior of sensitive data of the real-time monitoring and early warning system to obtain abnormal data; continuously adjusting analysis model parameters based on directional training test results and abnormal data; according to the method, firstly, the model is trained by analyzing the data of the fraud cases, test analysis is carried out, so that more effective public education and propaganda strategies are formulated, and public anti-fraud cognition and consciousness are accurately improved by regularly carrying out training test and accurate positioning training on the public; the development of anti-fraud propaganda is facilitated, the anti-fraud workload is reduced, and the propaganda effect and efficiency are improved.

The invention relates to a GPU-based high-IO perceptibility long-sequence text injection method, medium and equipment, and the method comprises the steps: obtaining and coding a long-sequence text input, processing a coded matrix vector through a multi-head attention mechanism, and realizing text injection, the matrix vector comprises a query Q, a key K and a value V. The method is characterized in that the value V is equal to the query Q, the key K and the value V. In the multi-head attention mechanism, the similarity is calculated by using an L2 distance, the matrix vector is segmented into blocks to realize thread block parallel processing during forward propagation, and the similarity-based result matrix and the gradients of the query Q, the key K and the value V are recalculated by using a recalculation strategy during back propagation. Compared with the prior art, the method has the advantages of reducing the overhead of attention calculation memory access, improving the operation efficiency and the like.

The invention relates to the technical field of security and protection monitoring, in particular to a method and system for detecting abnormal work of a network camera, and the method comprises the steps: training a basic data set to form a large model, collecting normal and abnormal pictures, carrying out the blocking processing, labeling, preprocessing, and carrying out the multi-task learning training of a CNN network structure model. SSIM is used as a loss function, simulated abnormal data is generated to enhance generalization ability, and model performance is evaluated; after video stream key frames or fragment blocks are obtained, a result R0 is recognized through the large model, reviews regularly, if abnormity occurs, the result is output, and after a certain result is accumulated, the large model is trained. The system comprises a data acquisition module, a model training module, a detection data acquisition module, a data partitioning module, an anomaly detection module, an examination module, a model optimization module and the like. According to the method, the detection efficiency and accuracy are improved, manual intervention is reduced, the problems of low video multi-channel detection performance, tedious threshold setting and low detection accuracy in the prior art are effectively solved, and the method is suitable for various security and protection monitoring scenes.

The invention discloses a method for predicting a riverbed siltation height and a sediment siltation range after sediment supply change, and the method comprises the following steps: S1, employing a variable rate equation to describe a dynamic adjustment process of a river channel characteristic variable under the sediment supply change, and the river channel characteristic variable comprises the riverbed siltation height and the sediment siltation range; s2, solving the variable rate equation to obtain a calculation expression of the characteristic variables; s3, determining a riverbed siltation height balance value and a sediment siltation range balance value of the sediment supply point position according to an incoming flow condition and an incoming sand condition of the target river channel; and S4, according to the balance value obtained in the step S3, in combination with the calculation expression of the characteristic variables, calculating and obtaining the riverbed siltation height and the sediment siltation range of the target river channel. The method can effectively predict the deposition height and the deposition range of the riverbed deposition area under different sediment supply conditions, provides key parameters for dangerous area division in flood control practice, and provides important theory and method support for mountain river water and sediment disaster prevention and control.

The invention discloses a model training method and system for archaeological cultural relics, and mainly relates to the technical field of artificial intelligence. Comprising the following steps: processing original archaeological cultural relic data, constructing an archaeological cultural relic knowledge base, and establishing a multi-modal data set; according to the established multi-modal data set and a public cultural relic multi-modal data set, training a multi-modal large model; and reasoning according to the user question in combination with an archaeological cultural relic knowledge base. The beneficial effects of the invention are that the method can better process and analyze archaeological cultural relic data, and provides a sustainable model training method for archaeological cultural relic research.

The invention discloses a shortcut operation method of equipment, electronic equipment and a readable storage medium, relates to the technical field of terminals, and can solve the problems of tedious operation and more time consumption during equipment operation. In the application, the electronic equipment can support a user to quickly trigger the operation on the electronic equipment through simpler and more convenient double-touch operation or sliding operation after double-touch aiming at the key, and the operation comprises but is not limited to at least one of the following operations: opening an application program, opening an application function, opening an application card, opening an application applet, switching a function state and the like. Therefore, since the operation range of the touch operation for the key is concentrated on the key, the touch operation can be completed through one-hand operation, and quicker and more convenient equipment operation experience can be provided for a user.

The invention discloses a head-mounted nerve stimulator capable of delaying and improving a cognitive function, and belongs to the technical field of brain nerve stimulation, the head-mounted nerve stimulator comprises a head-mounted earmuff worn on the head of a patient, one side of the middle part of the head-mounted earmuff is provided with an 808nm infrared light therapy device which is used for stimulating the brain of the patient and is adjustable in angle, and the other side of the head-mounted earmuff is provided with an infrared light source. A 40Hz light flickering device used for conducting visual stimulation on eyes of a patient is arranged in the eye protection device, and the device can stimulate the brain of the patient in the modes of 808nm infrared therapy, eye visual stimulation 40Hz light flickering, bone conduction accurate transmission 40Hz audio waves, transcranial magnetic stimulation and nasal olfactory stimulation. Gamma wave activity of the brain is promoted, cognitive performance, emotion balance and overall psychological health are improved, different stimulation effects are obtained through different stimulation modes, the clearing function of a brain lymphatic system is enhanced, brain performance of a patient is optimized, and sleep and cognition are greatly improved.

The invention discloses a visual experimental device for simulating the migration process of plugging agent particles in rock fracture.The visual experimental device comprises a fluid injection system, a fracture fluid simulation system and a data collection system, the fluid injection system is connected with the fracture fluid simulation system, and the fluid injection system injects experimental base fluid into the fracture fluid simulation system; the crack fluid simulation system is used for placing a rock sample with cracks and through holes and enabling the rock sample to be soaked in the experimental base fluid, and the data collection system is used for recording the migration, accumulation and bridging process of particles of the experimental base fluid in the cracks of the rock sample. The migration, accumulation and bridging mechanism of the plugging agent particles in the crack under the real rock crack condition is explored, the understanding of the mechanical behavior of the plugging agent particles in the natural crack is improved, and an important theoretical basis is provided for drilling plugging site construction operation.

The utility model provides a cold-proof clothing waterproofness detection device, and relates to the clothing detection technology field, the cold-proof clothing waterproofness detection device comprises a support wall and a detection mechanism, the inner side of the support wall is provided with the detection mechanism, through the arrangement of the detection mechanism, the stability and consistency of clothing in the detection process can be ensured through the arrangement of a transmission assembly in use, and the detection efficiency is improved. The detection assembly accurately detects moisture permeation through a liquid sensor, moisture can be rapidly removed and clothes drying can be accelerated through the arrangement of the drying assembly, a fan and a heat pipe, the detection efficiency is improved, water flow can be accurately controlled through the arrangement of the water spraying assembly and a pressure adjusting nozzle, the comprehensiveness of the test is ensured, and the recovery assembly is arranged, so that the detection efficiency is improved. The water utilization rate is improved, the water level is automatically monitored through the liquid level sensor, the overflow risk is avoided, the environmental protection property and safety of the equipment are guaranteed, an efficient, accurate and user-friendly detection platform is provided, and evaluation of the waterproof performance of the clothes is more scientific and standardized.

The invention discloses an electric power production safety simulation training system and method based on AR virtual reality. The system comprises AR glasses, a simulation teaching resource cloud platform and a 3D modeling real-time interaction subsystem. The method comprises the following steps: firstly, constructing a virtual reality scene of a power production environment by using an AR virtual reality technology; a plurality of training modules are arranged in the virtual reality scene, and each training module corresponds to one kind of power production safety knowledge or skill; the employee wears AR glasses, enters a virtual reality scene, and selects a corresponding training module for training according to own needs; the training system evaluates the operation of the employees and provides feedback and suggestions according to the evaluation result; the training system records the training progress and the training effect of the employees and generates a training report; and the training system periodically updates the training content. According to the invention, the immersion and reality of training are improved, the safety awareness of employees is enhanced, the safety awareness of the employees is improved, the training cost is reduced, and the training effect is improved.

The invention discloses a lung cancer early screening database management system, which relates to the technical field of database management and comprises a data acquisition module, a data arrangement module, a calculation module and an analysis module. The system integrates detection results of multiple biomarkers and individual clinical characteristics, provides a more accurate and more comprehensive lung cancer risk assessment database management system, can significantly improve the screening accuracy, reduces unnecessary excessive diagnosis and treatment, and improves the screening efficiency. The system considers clinical feature information such as age, gender, smoking history and family history, and constructs a comprehensive risk scoring system, so that the screening accuracy is further improved, the lung cancer risk of an individual can be comprehensively reflected, and the possibility of missed diagnosis and misdiagnosis is reduced.

The invention discloses a railway line environment potential safety hazard intelligent identification system, which mainly comprises a sensor module, a data processing unit and an alarm and feedback system, and is characterized in that the sensor module comprises a video monitoring camera, an environment sensor (such as a temperature sensor, a humidity sensor, a vibration sensor and a noise sensor), a laser radar and a gas detection sensor; the system is used for collecting multi-dimensional data along the railway, the safety monitoring efficiency and the early warning capability of the environment along the railway are remarkably improved through the application of the system, and high efficiency of informatization management and hidden danger rectification is achieved. Through an intelligent recognition algorithm, the system can automatically detect and recognize potential safety hazards along the railway, the dependence on manual inspection is reduced, and the emergency response speed and optimization of resource allocation are improved. In addition, the data storage and history backtracking functions of the system support long-term data analysis, and continuous improvement of railway safety management is facilitated.

The invention belongs to the technical field of bioinformatics, and particularly relates to a single-cell multi-omics data clustering framework based on deep collaborative contrast learning. According to the framework, the association relationship between cells is fully utilized, and meanwhile, the feature interaction of the same cell between different omics data is further enhanced, so that efficient clustering of multiple omics data is realized. In order to make full use of topological information between cells, a graph self-encoder and a feature information enhancement module under different omics are designed, and extraction and information enhancement of cell features are achieved. Meanwhile, a comparative learning technology is introduced to enhance interaction among different omics characteristics of the same cell, and finally single-cell multi-omics clustering is realized through a multi-omics deep collaborative clustering module.

The invention relates to an object point cloud reconstruction method based on a guided cross-modal robot, and the method comprises the steps: obtaining a depth image and a corresponding RGB image of an object under a single visual angle of the robot, carrying out the preprocessing, and obtaining the incomplete data of the point cloud of the object; a point cloud classification encoder is adopted to encode the point cloud classification task, a point cloud classification decoder is adopted to decode the point cloud classification task, and a classification result of the object is obtained; based on the object point cloud incomplete data, a point cloud data encoder is adopted to carry out position embedded coding, and point cloud features containing position codes are obtained; acquiring tactile data of an object at a single-view blind area position of the robot, and encoding the tactile data by a tactile space encoder to obtain tactile features; a visual touch cross-modal fusion module is adopted to fuse the point cloud features and the touch features, and a classification result is utilized to guide in the fusion process; and decoding the fused features by adopting a pyramid decoder to obtain a complete point cloud of the object. Compared with the prior art, the method has the advantages of high precision and the like.