Patents
Literature
Patsnap Eureka AI that helps you search prior art, draft patents, and assess FTO risks, powered by patent and scientific literature data.

30results about How to "Easy to converge" patented technology

V2X cooperative sensing method and system based on lightweight domain adapters

The invention discloses a V2X cooperative sensing method and system based on a lightweight domain adapter. According to the invention, a lightweight domain adapter network is placed in a feature map input branch of each cooperative vehicle of a first V2X cooperative sensing network carried by an own vehicle to form a second V2X cooperative sensing network, and cooperative sensing of the own vehicle and the cooperative vehicles is realized by using the trained second V2X cooperative sensing network. According to the invention, by introducing the domain adapter network, collaborative 3D target detection under the condition of heterogeneous feature maps of the own vehicle and the collaborative vehicle is successfully realized. According to the method, under the condition that the whole basic cooperative detection network does not need to be retrained, the domain adapters corresponding to different feature maps are inserted, so that the domain difference between heterogeneous feature maps is effectively reduced, the detection performance and scene adaptability of cooperative 3D target detection based on V2X are greatly improved, and the detection efficiency is improved. The method has a high practical value for improving the collaborative perception performance of automatic driving.
Owner:ZHEJIANG UNIV

End-to-end traffic light multi-task sensing method

The invention relates to an end-to-end traffic light multi-task sensing method, and the method comprises the steps: enabling a plurality of different tasks to share a model backbone network, and enabling the model parameters to be highly reused; performing weighted summation on the loss of each task to obtain final multi-task loss, and realizing mutual supervision among multiple tasks; in the training stage, a topk10 and topk1 double-branch positive sample distribution strategy is used, and topk10 provides more positive samples for the model, so that training is easy to converge; only a topk1 branch export model is used, so that a model output result is reduced to a lower order of magnitude, complexity caused by using an NMS when multiple targets are used is avoided, and traffic light perception post-processing is accelerated. Three different tasks of traffic light target detection, attribute identification and countdown key point positioning in traffic light perception can be unified into one network model. Traffic image input and single-model reasoning output of a multi-task result are realized. And compared with a multi-model serial mode, the model parameter quantity and the calculation quantity are greatly reduced, and the method is more friendly to embedded equipment.
Owner:SHANGHAI GEOMETRICAL PERCEPTION & LEARNING CO LTD

Humidity control optimization method for dry-type ripening cabinet

The invention discloses a humidity control optimization method for a dry-type ripening cabinet, and belongs to the technical field of PID (Proportion Integration Differentiation) control optimization. A PID controller is optimized by improving a fishing optimization algorithm (ICFOA), accurate control over the humidity of the dry-type ripening cabinet is achieved, and the improvement point comprises the steps that S1, a self-adaptive parameter beta and a disturbance factor delta are introduced to improve a capture rate parameter of the algorithm; s2, introducing a leader guidance strategy to improve a group capture mathematical model; according to the algorithm, various behaviors of human in the fishing process are simulated, an individual optimal solution is searched in a search space of the algorithm, values of the individual optimal solution on each latitude are decoded into Kp, Ki and Kd parameters, the parameters are transmitted into a PID controller, accurate control over the humidity of the dry-type ripening cabinet is achieved, a stable environment needed in the meat ripening process is maintained, and the meat quality is improved. And the quality and stability of the ripening process are ensured.
Owner:SHENZHEN SICAO ELECTRIC APPLIANCES

Regional building group source network load storage and utilization system energy management method based on GRU and A3C

The invention relates to the field of electric power, in particular to a regional building group source network load storage and utilization system energy management method based on GRU and A3C, and the method comprises the following steps: S1, obtaining the electric power information data of a regional building group, a plurality of variables influencing the nitrogen electric power information data, and parameter values of the variables; s2, performing preprocessing according to a plurality of parameter values of the electric power information data to obtain preprocessed data; s3, constructing a GRU neural network according to the preprocessed data; s4, an Actor-Critic algorithm is introduced into the GRU neural network, and a GRU-A3C neural network is obtained; s5, constructing a training set according to the preprocessed data, and training the GRU-A3C neural network; and S6, parameters of the GRU-A3C neural network are optimized according to an optimization algorithm, and a maximum expected cumulative reward value is obtained. The method has the advantages that an effective energy optimization management scheme is designed and realized based on the GRU and the deep reinforcement learning algorithm A3C, convergence can be faster, reward value accumulation is carried out, and cost reduction of energy optimization management is realized.
Owner:STATE GRID ZHEJIANG ELECTRIC POWER CO MARKETING SERVICE CENT +1

RISC-V-based AI processor design method for small convolution kernel acceleration

The invention relates to the technical field of RISC-V instruction set processors, in particular to an RISC-V-based AI processor design method for small convolution kernel acceleration, which comprises the following steps: determining the acceleration demand of a 3 * 3 convolution kernel in a convolutional neural network (CNN), selecting an open architecture RV32I of RISC-V as a basis, defining a core architecture of an AI processor, comprising a calculation unit, a register file and a control unit, the processing of the small convolution kernel is optimized; designing and realizing a vector arithmetic logic unit (VALU) and a convolution operation unit (CUU); the VALU has parallel computing capability and can execute addition, subtraction and multiplication vector operations to support instruction set extension; a convolution arithmetic unit CUU processes 3 * 3 convolution operations, including dedicated hardware logic to perform element-by-element multiplication and accumulation operations, thereby accelerating the convolution calculation process.
Owner:CIVIL AVIATION UNIV OF CHINA

Delay search method and system of delay phase-locked loop, computer and medium

The invention discloses a delay search method and system for a delay phase-locked loop, a computer and a medium, and the method comprises the following steps: enabling a signal to activate a finite-state machine, and jumping to a search state; based on the preset configuration of the finite-state machine, linear step length search is carried out in the guard band range to locate a target phase, if the target phase is not found in the guard band range, linear step length search is carried out again after the guard band range is expanded, and when the real-time phase is equal to the target phase, a symbol state is entered; in the symbol state, the slope of real-time phase change is calculated, and symbol matching is verified; and if the symbols are matched, entering a phase-locked state, fixing a real-time delay change value in the phase-locked state, and pulling up a phase-locked loop activation signal. According to the invention, the digital code which is output by the phase discriminator and represents the phase relationship between the reference clock and the input clock can be received, the delay is searched in a linear search mode, the control signal is sent to the delay line, and the clock delay is changed until the clock meeting the requirement is generated.
Owner:NO 24 RES INST OF CETC

A recognition method for aircraft skin defect detection

The invention discloses an identification method for detecting aircraft skin defects. It comprises the following steps: step A: image acquisition, using a CCD camera to take an aircraft skin picture and preprocessing the original aircraft skin image of the taken aircraft skin picture; step B: using a grayscale co-occurrence matrix to extract the characteristic parameters of the skin surface under different defect types, thereby establishing an aircraft skin characteristic parameter set; step C: using a fuzzy control system to fuzzify the characteristic parameters in step B, and obtain a characteristic parameter set represented by fuzzy language; step D: according to the different types of aircraft skin surface defects, the aircraft skin defect real-time detection is realized through BP neural network classification and recognition based on fuzzy control. The advantages are: realizing the rapid and accurate recognition and judgment of aircraft skin surface defects, effectively verifying the advantages of combining fuzzy control theory with BP neural network technology, and realizing the real-time classification and recognition of aircraft skin surface defects quickly and accurately.
Owner:JIANGSU AVIATION VOCATIONAL & TECH COLLEGE

Infrared and visible light image fusion system and method

ActiveCN114187214Bclear detailsExcellent evaluation indexImage enhancementImage analysisFeature extractionImaging Feature
The present invention discloses an infrared and visible light image fusion system and method, wherein the image fusion system comprises a multi-scale feature extraction unit for respectively extracting features of infrared images and visible light images, a feature screening unit and an image reconstruction unit located after the multi-scale feature extraction unit. Firstly, the multi-scale features of the infrared image and the visible light image are respectively extracted by the multi-scale feature extraction unit, and then the multi-scale features are feature screened by a dual attention mechanism, and then the image reconstruction unit completes the image fusion, thereby obtaining an infrared and visible light fusion image with clear details.
Owner:STATE GRID LIAONING ELECTRIC POWER CO LTD +2

Model equivalent processing method in energy storage container system simulation calculation

The invention discloses a model equivalent processing method in simulation calculation of an energy storage container system, the energy storage container system comprises a battery bin, a water machine bin and a hydrogen pumping chamber, and the model equivalent processing method comprises equivalent simplification processing of the battery bin, the water machine bin and the hydrogen pumping chamber; when equivalent simplification processing is carried out on the battery compartment, the energy source type of a battery pack and a high-voltage box in the battery compartment is selected as a total heat source, a heat power value is input, and heat generated by solar radiation is given to a container shell as heating power; carrying out equivalent simplification treatment on the water machine cabin, namely carrying out equivalent simplification treatment on a fan, filter cotton and a shutter; carrying out equivalent simplification treatment on the hydrogen pumping chamber, namely carrying out equivalent simplification treatment on a fan, filter cotton and a metal mesh pore plate; the complex model is equivalently processed by applying a theoretical modeling mode, so that the calculation precision is not influenced, the number of equivalent model grids is relatively small, the model grids are easy to converge, and the accuracy of a simulation calculation result is greatly ensured.
Owner:SUZHOU RCT POWER ENERGY TECH CO LTD

Water level monitoring system based on OCR technology

The invention relates to the technical field of water level monitoring, in particular to a water level monitoring system based on an OCR (Optical Character Recognition) technology, which comprises an embedded edge Internet of Things proxy server, and the embedded edge Internet of Things proxy server comprises the following modules: an image acquisition module, which is used for acquiring video image data of a water gauge in real time through a camera; the image recognition module is used for analyzing the video image data in real time according to a water level recognition algorithm based on an artificial intelligence neural network model to obtain water level detection analysis data; the water level recognition algorithm is based on a numeric symbol detection algorithm and an OCR algorithm, and numeric symbols and scale lines on the water gauge are detected and recognized. The network communication module is used for being in communication connection with a cloud information management platform and returning the water level detection analysis data. According to the method, old water level monitoring system equipment does not need to be transformed, the input cost is reduced, and the water level recognition precision of a traditional visual image method is improved.
Owner:重庆市科源能源技术发展有限公司

Behavior recognition method, device, storage medium and electronic device

The present disclosure relates to the field of computer technology, and specifically to a behavior recognition method, a behavior recognition device, a storage medium, and an electronic device. The behavior recognition method includes: obtaining multiple skeletal data features of a target object; using a branch displacement convolution subnetwork to perform branch processing on the multiple skeletal data features to obtain multiple feature maps; using a fusion displacement convolution subnetwork to perform mainstream fusion on the multiple feature maps to obtain a feature vector; and using a fully connected layer to perform recognition based on the feature vector to obtain the behavior information of the target object. The behavior recognition method provided by the present disclosure can solve the problem of large amount of calculation in the prior art when performing behavior recognition based on bones.
Owner:泰康保险集团股份有限公司

Model training method, 3D face reconstruction method, device and storage medium

ActiveCN114972603Beasy to convergeImage enhancementImage analysis
The present application relates to a model training method, a 3D face reconstruction method, a device and a storage medium. Since the method introduces compensation parameters in the process of training the model, the training loss that can be obtained is more accurate than that without compensation parameters, and it is easier for the neural network model to converge. And since the PCA basis is not introduced in the process of training the model, when the trained model is used for 3D face reconstruction, there is no problem of low reconstruction accuracy due to the influence of the accuracy and dimension of the PCA basis.
Owner:BEIJING QIYI CENTURY SCI & TECH CO LTD

Training method for spiking neural networks, spiking neural networks and devices

This application provides a training method, a spiking neural network, and an apparatus for a spiking neural network. The method includes: iteratively training the spiking neural network using training data of the object to be processed, wherein the spiking neural network contains at least one neuron model; calculating a task loss function based on the training results of the spiking neural network and updating the weight parameters; updating the current discharge threshold parameter of the neuron model based on the historical discharge frequency of the neuron model, wherein the value of the current discharge threshold parameter increases with the increase of the historical discharge frequency; and obtaining the trained spiking neural network model using the updated weight parameters and the updated discharge threshold parameter after multiple iterations of training. This method facilitates the training of a sparsely discharging spiking neural network, reduces the energy consumption during spiking neural network applications, and improves model energy efficiency by using the trained spiking neural network model to process the data of the object to be processed.
Owner:PEKING UNIV

A simulation method for the fuel-saving orbit change mission of "one-to-many" approach to space targets

ActiveCN114818474BFast convergenceeasy to convergeFlight vehicleClassical mechanics
The invention relates to a simulation method for a fuel-saving orbit change task for "one-to-many" approach to space targets. The objective function is designed with the minimum velocity increment vector modulus required in the orbit change process as the goal; the optimization levels are set, the first level: setting approach sequence optimization variables, and determining the approach sequence of a service aircraft arriving at a target satellite through the variable size; the second and third levels: determining the value range of the approach orbit change end time optimization variable for each target satellite that satisfies both a measurement and control window and an illumination window; determining the value range of the approach orbit change start time optimization variable for each target satellite; the fourth level: under a given approach sequence, optimizing the given approach orbit change end time and approach orbit change start time according to the objective function to obtain an optimization result; according to the set optimization level order, starting the optimization by level, selecting the best result from the optimization results to obtain a fuel-saving orbit change strategy.
Owner:CHINA ACADEMY OF SPACE TECHNOLOGY

Electricity consumption prediction method, device and equipment for charging service industry and medium

The application relates to a charging and battery swapping service electricity consumption prediction method and device, equipment and medium, and relates to the technical field of power load prediction. The application comprises the following steps: receiving charging and battery swapping service electricity consumption and external variable data to construct features; performing multi-seasonal decomposition on endogenous data to obtain trend, daily season, weekly season and residual components and input the components into a feature matrix; extracting electricity consumption at the same time of the previous week, converting the prediction target into a residual sequence of the current electricity consumption and the electricity consumption at the same time of the previous week; constructing a sliding supervision sample, inputting a Seq2Seq model containing a bidirectional LSTM encoder, an additive attention and an LSTM decoder to generate a residual prediction sequence; and adding the residual prediction sequence and the electricity consumption sequence at the same time of the previous week point by point in time to obtain a charging and battery swapping service electricity consumption prediction value. Through explicit period decomposition, target residualization and attention enhancement, the application significantly improves the prediction accuracy and robustness under complex weather conditions and load fluctuation scenarios.
Owner:POWER SUPPLY SERVICE & MANAGEMENT CENT STATE GRID JIANGXI ELECTRIC POWER CO LTD

Magnetic anisotropy energy research method, device, equipment, medium and program product

ActiveCN119517260Aeasy to convergeSpeed ​​up the calculation processComputational materials scienceComplex mathematical operationsElectronic structureDensity functional theory
The invention discloses a magnetic anisotropy energy research method, device and equipment, a medium and a program product, and relates to the technical field of magnetic materials, and the method comprises the steps: obtaining a to-be-researched target magnetic material, and configuring the initial magnetic state of the target magnetic material; performing density functional theory calculation of primary spin-orbit coupling on the target magnetic material in the initial magnetic state to obtain an electronic structure physical quantity of the target magnetic material; and according to the physical quantity of the electronic structure in combination with a Green function method, performing one-time calculation to obtain a magnetic anisotropy energy result pointed by each magnetic moment of the target magnetic material, and performing magnetic anisotropy energy analysis based on the magnetic anisotropy energy result pointed by each magnetic moment of the target magnetic material. According to the method, the complex calculation process of the density functional theory of adding spin-orbit coupling for multiple times can be avoided, the problem that the calculation process is difficult to converge is avoided, and the efficiency of calculating and analyzing the magnetic anisotropy energy is improved.
Owner:ARTIFICIAL INTELLIGENCE RES INST OF HEFEI COMPREHENSIVE NAT SCI CENT (ANHUI ARTIFICIAL INTELLIGENCE LAB)

A method and device for simulation analysis of strength of screw connection

The present invention provides a strength simulation analysis method and device for screw connection, the method comprising: obtaining a screw connection structure; segmenting an annular contact surface and a screw connection area; setting a tetrahedron unit to replace the screw, so that the tetrahedron unit connects the annular contact surface and the screw connection area to form a new screw connection structure; applying surface pressure on the tetrahedron unit, setting boundary conditions according to calculation conditions and surface pressure, and performing strength simulation analysis on the screw connection structure, and outputting strength simulation analysis results. Compared with the prior art, the present invention replaces a screw model with a large amount of data with a tetrahedron unit with a small amount of data, simplifies the screw connection structure, reduces the amount of calculation, ensures the accuracy of the results, and makes the results easy to converge. In addition, by applying surface pressure on the tetrahedron unit, setting boundary conditions according to calculation conditions and surface pressure, the strength simulation analysis results can be closer to the actual working conditions.
Owner:GUANGZHOU GRG METROLOGY & TEST CO LTD

Dynamic mapping method for multi-bank access, electronic device and storage medium

The present invention relates to the field of chip design, and in particular to a dynamic mapping method, electronic device and storage medium for multi-bank access, which obtains the address of a target memory bank bank0 by parsing an access request command cmd0; if the number of state machines currently allocated to a bank is greater than a preset allocation threshold, a mapping relationship between bank0 and a target state machine FSM0 is established through dynamic mapping, and the steps of dynamic mapping include: if there is a state machine with an unallocated bank, a state machine FSM0 of an unallocated bank is allocated to bank0, and a mapping relationship between FSM0 and bank0 is established; and at the same time, the state machine FSM0 is released. i With bank i The mapping relationship between them is used to obtain the state machine FSM of the unassigned bank. i It reduces area overhead and also improves the timing of logic design.
Owner:NANJING QIJIAN SEMICON TECH CO LTD +1

Coupling methods of aerodynamics and vehicle dynamics applicable to high-speed trains

This invention discloses a coupling method for aerodynamics and vehicle dynamics applicable to high-speed trains, relating to the interdisciplinary field of aerodynamics and vehicle dynamics. The method includes: establishing separate aerodynamic and vehicle dynamic models for the high-speed train; exchanging data between the two models using a dual-time-step time-progression method, improving computational time accuracy, step size, and stability. While comprehensively considering the interaction between the high-speed train and its surrounding flow field, the influence of track irregularities, and tunnel effects, a multi-region mesh generation method is adopted, improving the quality and efficiency of mesh generation. Furthermore, the overlapping mesh method avoids mesh reconstruction problems. Simultaneously, the use of six-degree-of-freedom motion velocity to reflect the attitude changes of the high-speed train simplifies the coupling steps. The entire process saves time on mesh generation, calculating the displacement of mesh nodes around the train, mesh reconstruction, and computational iterations, enabling the coupling method to be implemented quickly and on a large scale.
Owner:INST OF MECHANICS CHINESE ACAD OF SCI

Adversarial knowledge distillation method and system based on feature semantic guidance and feature orthogonalization

PendingCN122693766Aeffective decouplingsuppress confusion
The application discloses an adversarial knowledge distillation method and system based on feature semantic guidance and feature orthogonalization, which firstly acquires an image classification dataset, extracts text features, and constructs a semantic feature space.Secondly, a knowledge distillation framework including a teacher model, a student model and a projection head is constructed, the input of the teacher model and the student model is an image in the dataset, and the output is an image category prediction result. Then, the student model parameters are frozen, the projection head is updated through momentum tracking, in the feature level semantic adversarial sample generation stage, all network parameters are frozen, and normalized disturbance is constructed to obtain adversarial features. Finally, in the student model multi-objective joint optimization stage, the teacher model and the two projection heads are fixed, the student model parameters are updated, the classification loss, the distillation loss, the consistency loss and the orthogonalization loss are introduced, and optimization is completed. The application improves the fine-grained discrimination and generalization ability of the student model image classification, and does not introduce additional reasoning overhead.
Owner:HANGZHOU DIANZI UNIV +1

A method for predicting the aerothermal structure response based on Delaunay triangulation interpolation

This invention discloses an aerodynamic thermal structure response prediction method based on Delaunay triangulation interpolation, relating to the aerospace technology field, to address the computational difficulties of traditional coupling methods and the insufficient accuracy of decoupling methods. The method first obtains aerodynamic heat flux and pressure data of an object surface at different wall temperatures through parameterized calculations; then, using Delaunay triangulation interpolation, it establishes a continuous functional relationship between heat flux and pressure and wall temperature for each surface node of the structural finite element mesh; finally, it uses this functional relationship as precise boundary conditions to drive finite element calculations, thereby efficiently and accurately predicting thermal deformation, thermal stress, and natural vibration frequencies. This invention achieves effective decoupling of the flow field and structural field through an intelligent surrogate model, significantly improving computational efficiency and engineering practicality while achieving near-coupling accuracy.
Owner:CHINA AERODYNAMIC RES & DEV CENT EQUIP DESIGN & TESTING TECH INST +1

Non-invasive myocardial transmembrane potential reconstruction method based on global feature-based fast iterative soft threshold contraction algorithm network

The present invention discloses a non-invasive myocardial transmembrane potential reconstruction method based on a global feature-based fast iterative soft threshold shrinkage algorithm network, which combines the convergence theory of the traditional fast iterative soft threshold shrinkage algorithm (FISTA) and the flexibility and generalization advantages of deep learning, and uses an attention mechanism to extract global features, and uses a neural network to replace the soft threshold shrinkage step in the traditional FISTA, thereby replacing the specific regularization form in the traditional method. At the same time, a learnable momentum accumulation step is added to the iterative network to replace the momentum accumulation step in FISTA, thereby achieving better reconstruction accuracy and faster reconstruction speed.
Owner:JIAXING RES INST ZHEJIANG UNIV +1

A communication interference resource allocation method based on multi-agent reinforcement learning

A communication interference resource allocation method based on multi-agent reinforcement learning, which relates to a communication interference resource allocation method. The present invention aims to solve the problem that the existing interference decision algorithm does not take into account that with the increase in the number of multi-agents, the action space and the state space will grow rapidly, and the algorithm may be difficult to converge. On the basis of the multi-agent proximal strategy optimization algorithm, the present invention proposes a multi-agent proximal strategy optimization algorithm with a selective experience storage strategy, selectively stores the experience entering the experience pool, improves the convergence ability of the algorithm, and adds an attention mechanism to the multi-agent proximal strategy optimization algorithm. The present invention belongs to the field of communication technology.
Owner:HARBIN ENG UNIV

A UAV intelligent navigation method based on deep reinforcement learning

A method for intelligent navigation of unmanned aerial vehicles based on deep reinforcement learning, the scheme of which is as follows: (1) establishing a quadcopter unmanned aerial vehicle model in a three-dimensional simulation environment and generating an environment of obstacles and target points; (2) designing a state space and a discrete action space for multi-dimensional feature fusion; (3) designing a reward function based on Euclidean distance; (4) designing a strategy network and a state value network; (5) designing a temporary experience pool to store information feedback when the unmanned aerial vehicle interacts with the environment; (6) training a reinforcement learning network based on a discrete PPO algorithm; (7) realizing intelligent navigation of the unmanned aerial vehicle. The present invention can effectively accelerate the learning and training efficiency of unmanned aerial vehicles and improve the quality of decision-making of unmanned aerial vehicles in different environments.
Owner:XIDIAN UNIV

Advertising copywriting creation method and its device, equipment, medium and product

The present application relates to a method for creating advertising copy and its apparatus, equipment, medium, and product, the method comprising: obtaining a product image of a target product; extracting the image features of the product image through an image encoder, and obtaining an image feature vector that maps the product image to a specific semantic space; providing the image feature vector as an initial input vector to a decoder to gradually autoregressively decode each text vector in a text sequence of the advertising copy, wherein the text vector of the previous step is used to query the key features in the image features of the product image as the input vector corresponding to the text vector of the next step; and restoring the text sequence of the advertising copy according to the text vector sequence predicted by the decoder. The advertising copy obtained in the present application is obtained based on the image features of the product image, can more comprehensively reflect the selling point characteristics in the product image, and has uniform quality without being affected by the creator's style.
Owner:BUSINESS LINE TECHNOLOGY (SHENZHEN) CO LTD

A heart sound segmentation method and system based on synchronous squeezed short-time Fourier transform and BiGRU

ActiveCN118245860BImprove aggregation effectSimple structureStethoscopeBiological models
The present invention discloses a heart sound segmentation method and system based on synchronized squeezed short-time Fourier transform and BiGRU, the method comprising: constructing a heart sound segmentation model based on synchronized squeezed short-time Fourier transform (SSTFT) and BiGRU, the model consisting of SSTFT unit, BiGRU network, Softmax classifier; training the heart sound segmentation model using samples of heart sound data set; inputting the heart sound signal to be segmented into the trained heart sound segmentation model for state classification, and dividing it into four state categories of the first heart sound (S1), the second heart sound (S2), S1-S2 interval, and S2-S1 interval; correcting the wrong state category predicted by the heart sound segmentation model; the system comprises: a model construction module, a model training module, a heart sound state classification module, and a state correction module. The present invention can effectively improve the positioning precision and accuracy of heart sound segmentation.
Owner:NANJING UNIV OF POSTS & TELECOMM

Traffic order prediction method, device and equipment

The invention provides a transportation order prediction method, device and equipment, and the method comprises the steps: obtaining transportation orders of a plurality of regions of a target city in each unit time period in a first time period, and obtaining a plurality of first historical order sequences; extracting spatial feature vectors from the first historical order sequences in the plurality of regions through a self-attention neural network model; extracting a time feature vector from the first historical order sequence in each unit time period through a self-attention neural network model; and through a self-attention neural network model, according to the time feature vector and the space feature vector, predicting a traffic travel order of at least one region in one or more unit time periods in the future. According to the method, the spatial feature vector and the time feature vector are extracted from the historical order sequence, and the future traffic order is predicted in combination with the spatial feature vector and the time feature vector, so that the calculation frequency of the model is changed from full quantity to sparse quantity, and the calculation efficiency of the model is improved.
Owner:CETC NEW SMART CITY RES INST CO LTD

Object automatic labeling method and system suitable for gate channel

The invention discloses an automatic object labeling method and system suitable for a gate channel. The method comprises the following steps: constructing an initial data set based on gate passing characteristics and a priority strategy; constructing an automatic labeling model and performing model training by using the initial data set; and performing target detection on the to-be-labeled image by using the automatic labeling model, and taking pedestrian target information in a target detection result as labeling information of the to-be-labeled image. According to the method, the priority strategy is introduced into the initial data set of the automatic labeling model, so that labeling results have high consistency when shielding and overlapping occur. The model can generate the labeling box according to the preset priority rule regardless of the condition that adults, adults and children or adults carry articles, so that the problem of non-uniform labeling standards caused by human subjective judgment is avoided. And by introducing the priority loss, the automatic labeling model pays more attention to the features of important positions, and the recognition capability in a complex scene is improved.
Owner:IVES (SUZHOU) SPECIAL EQUIP CO LTD

Cell classification method, system, cell graph Transformer model training method and electronic device

The present invention relates to a cell classification method, system, cell graph transformer model training method and electronic equipment, belonging to the field of cell classification technology, including: acquiring a cell image; inputting the cell image into a cell segmentation model to obtain a cell segmentation map or inputting the cell detection model to obtain a cell detection map; inputting the cell image into the feature extractor to obtain a feature map; using the cell graph construction module to generate a cell map according to the feature map and the cell segmentation map or generating a cell map according to the feature map and the cell detection map; using the cell graph symbolization module to generate a symbol vector according to the cell map; using the transformer encoder to obtain a cell classification according to the symbol vector. The present invention makes full use of the powerful classification performance of the transformer encoder and the global topological relationship information of the cell nucleus of the cell image to improve the accuracy of cell classification.
Owner:SHENZHEN RES INST OF BIG DATA

Computational method of UAV swarm intelligence model based on anti-predation characteristics of bird flocks

The present invention provides an intelligent model and calculation method of drone clusters based on the anti-predation characteristics of bird flocks, including the following steps: Step 1, analyzing the source and mechanism of the anti-predation characteristics of bird flocks, and establishing the dynamic action characteristics relationship of multiple types of birds; Step 2, establishing an intelligent model of drone clusters based on the anti-predation characteristics of bird flocks; Step 3, designing an intelligent calculation method of drone clusters based on the anti-predation characteristics of bird flocks, and setting optimization rules and parameter adjustment methods. The present invention provides an effective way to efficiently and dynamically control drone clusters, ensuring that drone clusters can respond quickly to changes in the external environment. The model and calculation method can also be applied to other cluster intelligent control and optimization problems.
Owner:THE 28TH RES INST OF CHINA ELECTRONICS TECH GROUP CORP