93 results about "Spatiotemporal encoding" patented technology

The invention provides a cellular network flow space-time prediction method and system based on a hybrid neural network, and belongs to the technical field of intelligent communication. The method adopts a layered deep neural network architecture, and comprises a data embedding layer, a space-time coding layer, a feature fusion layer and an output layer. The data embedding layer maps a historical traffic sequence, cross-domain external data and metadata into high-dimensional features; the space-time coding layer is used for respectively fusing one-dimensional causal convolution and a Mama neural network to extract multi-scale time features and densely connecting convolution and a multi-head attention mechanism to capture multi-scale space features through time and space modeling branches; the feature fusion layer realizes adaptive weighted fusion of spatial-temporal features, cross-domain features and metadata features by using a gating fusion mechanism; and the output layer performs linear transformation on the fusion features to generate a final prediction result. According to the method, the spatial-temporal dynamic capture of the service traffic is accurate, the prediction curve is highly fit with the true value, and the accurate prediction of the multi-service traffic of the cellular network is realized.

The invention discloses a cross-modal large model construction method and system based on track spatio-temporal characteristics, and belongs to the crossing field of artificial intelligence and dynamic spatio-temporal data processing, and the method comprises the steps: carrying out the sliding sampling and spatial distribution difference judgment through the semantic dynamic segmentation of multi-scale track spatio-temporal data, and generating spatio-temporal data blocks with consistent semantics; designing a space-time encoder of a hybrid architecture, extracting track time sequence association and spatial features, and unifying dimensions; constructing a text space-time fusion mechanism, and dynamically adapting cross-modal features by means of an anchor interface and gating fusion; a staged instruction fine tuning strategy is adopted, semantic alignment of space-time and text features is optimized firstly, then model top-layer parameters are trained cooperatively, complex scene adaptation is enhanced in combination with instruction difficulty progression and hard sample mining, and space-time constraint regular terms are introduced to guarantee output rationality. According to the method, high-precision cross-modal reasoning capability is provided for scenes such as track analysis and track prediction.

The embodiment of the invention provides an image processing method and device based on computer vision and artificial intelligence, and the method and device achieve the privacy protection of data through innovatively designing a multi-dimensional data safety collection mechanism and through feature branch isolation and space-time coding. A hierarchical feature fusion network is constructed, and a safe and reliable target tracking system is established in combination with environmental constraints and physical laws. A dynamic weight adaptive mechanism is introduced, and the accuracy of a processing result and user privacy are ensured through multi-scale feature extraction and prediction correction. According to the method, the data security is protected, and meanwhile, the defects of the traditional technology in the aspects of data fusion, feature extraction, prediction constraint and the like are effectively overcome.

The invention discloses a track user association method based on semantic perception and space-time coding, relates to the technical field of location service and user behavior analysis, and aims to solve the problems of excessive dependence of POI identifiers, limited space-time representation capability, insufficient cross-city generalization capability and the like of the existing TUL method in practical application. By introducing a pre-trained large language model to carry out POI category semantic coding, multi-frequency sine space-time coding and a double-flow transfer learning mechanism, the method can effectively improve the prediction precision and the model generalization ability.

The invention provides a leaky-wave antenna and a frequency-modulated continuous wave generation method based on the leaky-wave antenna, and relates to the field of millimeter-wave radar, and the method comprises the steps: constructing a space-time coding matrix, configuring a radiation unit state regulation and control beam direction in a space domain, modulating a radiation unit state switching rule in a time domain to generate a frequency-modulated continuous wave, and achieving space-time dual-domain regulation and control. And frequency-modulated continuous waves with controllable bandwidth are generated. The space-time coding leaky-wave antenna is formed by alternately arranging 1-bit radiation units in an upper row and a lower row, switching of radiation / non-radiation states is achieved by loading a synchronous bias PIN diode across an annular gap, a space-time coding matrix is constructed, the states of the radiation units are configured in a space domain to regulate and control the wave beam direction, and the space-time coding leaky-wave antenna is formed. Modulating the switching rule of the radiation unit in a time domain, and generating a frequency-modulated continuous wave; the radiation efficiency is improved by the alternative arrangement of the radiation units, the control difficulty and the hardware cost are reduced, multiple functions are integrated into a single device, and the problem of low integration level of the traditional separation architecture is solved.

The invention relates to the technical field of stereoscopic vision, and discloses a synchronous speed visual matching method and system, electronic equipment and a storage medium, and the method comprises the steps: synchronously collecting a stereoscopic video sequence with a predefined frame rate; executing multi-target hybrid tracking and motion induction detection, and outputting target motion information including position and velocity vectors; extracting hierarchical motion features of the target from continuous multiple frames of the stereoscopic video sequence, and performing unified space-time coding; under geometric constraints of stereoscopic vision, scale cosine similarity, direction similarity and trajectory consistency measurement are calculated and serve as observation evidences to be input into the probabilistic reasoning model for fusion, and a posterior probability representing matching reliability is output; the weight distribution of the speed similarity and the direction similarity is adjusted according to the motion characteristics of the targets in the scene, and the stable corresponding matching relation between the left view target and the right view target is established. According to the method, high-time-resolution information can be utilized, motion features and geometric constraints can be effectively fused, and the method has self-adaptive capacity.

The invention provides a space-time fusion coding method and a data relation generation method and system, and is applied to the technical field of physical world-oriented multi-source heterogeneous data organization management. Space information of a target data object is mapped to a geographic space subdivision grid coding system to generate a space code; the method comprises the following steps: standardizing time information according to preset time granularity (such as year, month, day and hour), mapping the time information to a layered time coding system to generate a time code, and fusing a space code, the time code and object identification information according to a preset rule to generate a unique space-time fusion code; and finally, the space-time fusion codes and attribute information are bound to form standardized data records, data objects of different sources and different formats are mapped to a unified space-time coding system, data islands are broken, and a data foundation is laid for constructing a unified space-time relation network.

The invention relates to the technical field of digital content production, and discloses an interactive digital content production system based on a Transform architecture. According to the system, text, image and audio data of original digital content are acquired through a content feature extraction module, cross-modal feature alignment is performed by using a multi-head attention mechanism, and content feature tensors with space-time relevance are generated; the dynamic weight distribution module calculates relative importance scores of different modal features based on the tensor, and adopts a gating mechanism to perform dynamic weight fusion to form content semantic enhancement representation; the interaction intention analysis module performs space-time coding matching on the enhanced representation and the user operation instruction stream, and analyzes an intention distribution matrix of user operation on the content dimension; a hierarchical decoding generation module constructs a multi-scale content generation path in a Transform decoder according to the intention distribution matrix; and the real-time rendering engine module loads implicit representation output by the path, so that efficient and intelligent digital content creation is realized.

The application discloses a gesture recognition method, device, equipment, medium and program product based on multiple sensors, which comprises the following steps: decomposing multiple sensor data into multiple action frames in time sequence based on sensor timestamps, and generating an original sequence data set based on the multiple action frames; preprocessing the original sequence data set to generate a hand action time sequence, wherein the preprocessing comprises compensating the original sequence data set based on environmental data; distributing data weights of each sensor based on the contribution degree of each sensor, performing feature fusion on multi-modal hand action data at each time point in the hand action time sequence based on the data weights, performing time-space coding on the fused features, inputting the time-space coding vector into a gesture recognition model for gesture recognition, effectively fusing multi-modal sensor data, and effectively reducing the noise influence of the environment on the sensor by combining environmental data for data compensation, thereby greatly improving the gesture recognition accuracy.

The invention discloses an intensive care unit patient hospitalization time prediction method based on time-space historical features, and the method comprises the steps: screening patients from a public database MIMIC-IV, respectively extracting the static features of the patients and the dynamic features divided according to the hour granularity, and eliminating the dynamic features with a high missing rate. According to the method, features per hour of hospitalization records of a patient are input into a space-time encoder, features with time and space dependence are extracted, then original features and space-time historical features of historical time points are fused and input into a decoder, global and local space-time dependence is fully extracted, and the decoder outputs predicted remaining hospitalization time. In this way, the global and local dependency relationships among the dynamic data at different time points in the hospitalization records of the patient can be better captured. And finally, resource allocation is optimized and medical cost is reduced for the intensive care unit according to a prediction result, and meanwhile, scientific auxiliary decision support is provided for clinicians.

The embodiment of the invention discloses a power system information acquisition and control method and terminal equipment. The method comprises the following steps: collecting multi-modal data; the multi-modal data comprises system-level data and region-level data; extracting input features from the multi-modal data; performing space-time coding on the input features by using an artificial intelligence AI model to obtain time-dependent features and spatial features; processing the time-dependent features and the spatial features by using a double-layer attention mechanism to obtain at least two levels of weights; according to real-time power grid data in the multi-modal data, two-stage weights are dynamically adjusted, and transient event information is output; obtaining a first configuration instruction according to the adjusted two-stage weight and the transient event information; optimizing the first configuration instruction based on the region-level data to obtain a second configuration instruction; the second configuration instruction comprises an acquisition instruction and / or a resource scheduling instruction; the resource scheduling instruction is used for scheduling software and hardware resources of the terminal equipment.

The application discloses a multi-view video synchronization and target association method based on space-time coding, comprising the following steps: step one, multi-view video acquisition; step two, space-time coding generation; step three, inter-frame synchronization based on space-time coding; step four, cross-view target detection; step five, construction of a space-time association cost matrix; step six, global target association; and step seven, output of an association track. The application uses space-time coding to replace absolute time stamps, effectively resists differences in frame rates of different cameras and time drift problems, and reduces the requirement of the system on hardware synchronization; the space coordinate information is integrated into the coding, so that the target association no longer simply depends on the appearance, and the association robustness in complex scenes such as light change and dramatic view change is greatly improved; through coding matching and global optimization, the video synchronization and target association tasks can be completed at the same time, and the calculation efficiency and system integration are improved.

The present application relates to a kind of construction safety risk early warning method and system based on multi-source data fusion, first acquisition including positioning trajectory, sensor signal, inspection text and video features and so on multiple types of real-time data, in combination with the construction field knowledge graph of semantic consistency verification, realize unified space-time datum and semantic association mapping;Through multimodal feature enhancement, space-time coding and cross-modal fusion, high-dimensional hidden state sequence is established;Causal structure model is constructed using knowledge graph weak causal relationship edge, and counterfactual disturbance response modeling is carried out on high-dimensional data, and the association of context and risk event is extracted;Dynamic adjustment risk assessment result, in combination with risk threshold realizes closed loop early warning response, and through feedback self-optimizing model parameter.The present application improves the precision and self-adaptive ability of construction site safety risk identification and response.

The invention discloses an internet resource long-term storage management and storage system, which comprises a space-time coding layer which is based on an FPGA (Field Programmable Gate Array) main control chip, comprises a parallelized tensor processing unit and is used for converting digital resources and metadata and context relationships thereof into a space-time coding body with a time self-description capability; the storage unit is used for receiving the space-time coding body from the space-time coding layer and permanently solidifying the space-time coding body in a tamper-resistant physical form; and the four-dimensional continuation control layer is used for dynamically managing the integrity, readability and authenticity of resources in the whole life cycle through an autonomous evolution algorithm and a distributed verification network. Through collaborative innovation of a space-time coding body (an information layer), a holographic cold memory (a physical layer) and a continuation control network (a management and control layer), the basic challenges of long-term storage of Internet resources in the aspects of authenticity, integrity, availability, safety, cost, technical dependence and the like are systematically handled.

The invention relates to the technical field of target detection, and provides a video target detection method and system based on hybrid Transform-Mama. The method comprises the following steps: based on all frame images in a video to be detected, generating a Token feature sequence by adopting a shared feature extractor, fusing the Token feature sequence and a position code, and inputting the fused Token feature sequence and the position code into a spatial adaptive deformable Transform encoder to obtain spatial encoder features of all frames; splicing the space encoder features of all the frames and then inputting the spliced space encoder features into a time sequence cascade bidirectional Mama encoder to generate space-time encoder features of all the frames; inputting the space-time encoder features of all frames and the target query into an entangled Mama-Transform decoder, and enriching instance-level context information of the target query through query-feature interaction and fine granularity alignment to obtain space-time decoder features; and inputting the space-time decoder features into a shared feed-forward network for classification and bounding box regression to obtain a target detection result of each frame.

The invention discloses a multi-source meteorological refined prediction method based on a physical migration model, and the method comprises the steps: introducing dual-channel space-time coding, a physical prior guided attention mechanism, and a multi-scale sliding window input and gating feature fusion strategy into a model architecture; and in combination with technologies such as Bayesian optimization driven hyper-parameter automatic search, a self-adaptive mixed loss function, uncertainty quantification and physical consistency post-processing, the modeling capability and prediction robustness of the model for a complex weather process (especially an extreme event) are remarkably improved. No matter based on a simulation test of reanalysis data or in verification of a real meteorological observation station, the method shows prediction precision, stability and interpretability superior to those of a traditional numerical mode and a single deep learning model. The method provides reliable technical support for refined meteorological service, disastrous weather early warning and energy scheduling decision, and has important application value and popularization prospect.

The invention discloses a multi-dimensional first-aid resource dynamic scheduling system and method based on space-time coding, and particularly relates to the technical field of first-aid resource dynamic scheduling. A four-dimensional space-time resource state matrix capable of simultaneously expressing spatial distribution, time evolution and medical capability characteristics is formed by multi-source heterogeneous first-aid data, a dynamic monitoring mechanism of a scheduling direction high-frequency fluctuation coefficient and a multi-target optimization weight dynamic drift coefficient is introduced, and a resource scheduling decision jitter model is constructed. Carrying out conjoint analysis on the two types of fluctuation characteristics to generate a resource scheduling decision jitter index, and comparing the resource scheduling decision jitter index with a preset threshold value in real time; and a differentiated stabilization strategy is adopted for different instability levels, so that the problems of repeated adjustment of a driving path, interruption of hospital reception preparation, disorder of a traffic signal strategy and the like caused by frequent scheduling switching are effectively reduced.

The application discloses a rPPG signal extraction network based on a CBAM attention mechanism and a bidirectional Mamba, and a non-contact physiological monitoring method and device, wherein spatial noise is filtered by introducing a CBAM attention mechanism, long time sequence dependence is modeled at low cost by using a Mamba architecture, and signal purity is improved by combining a frequency domain and a confidence gate. The rPPG signal extraction network effectively extracts weak skin color change features through an adaptive differential fusion mechanism of the fusion backbone network, realizes long time sequence global modeling through a Mamba architecture (with linear calculation complexity) in the space-time encoder, performs double cleaning on the features in the spatial domain and the frequency domain through a confidence perception gate module and a frequency domain gate module, and finally reconstructs a high-fidelity rPPG waveform through the bidirectional Mamba regression head.

The invention particularly relates to a method for realizing space-time similar trajectory search under non-uniform data based on domain invariant learning. The method can realize generalization trajectory similar search across multiple fields. Specifically, in order to solve cross-domain distribution differences in trajectory representation, a dynamic confrontation game is established between an adaptive domain discriminator and a space-time encoder, and the space-time encoder is guided to eliminate domain specific information in a feature space, so that shared and domain-independent representation is extracted while trajectory semantics are reserved; a bidirectional gating circulation unit is adopted to map tracks of different lengths to a feature space of a fixed dimension, and feature mismatching caused by inconsistent sequence lengths in a traditional method is avoided; a bi-directional gating circulation unit and learnable Fourier features are adopted to encode a track, so that a space-time encoder supports a track sequence with irregular time intervals.

The application discloses a robot imitation learning method and device based on a hybrid perception nonlinear model predictive control, and the method comprises the following steps: collecting an external perception picture sequence and an ontology perception information sequence to obtain a first data set; sampling the first data set according to a preset field of view length; performing static kinematics coding and dynamic space-time coding processing on the ontology perception information sequence in the sampled data set to obtain a second data set; constructing a composite loss function; inputting the second data set into a prediction model, outputting a prediction sequence through a nonlinear model predictive control algorithm; according to the prediction sequence, a robot performs an action to complete an imitation learning task; and according to the composite loss function, the prediction model is updated, and the step of collecting the external perception picture sequence and the ontology perception information sequence to obtain the first data set is returned. The application can improve the reasoning speed and the imitation learning task accuracy, and can be widely applied to the technical field of robot imitation learning.

This application discloses a multimodal large-scale robot control method and system based on eye-tracking feature enhancement. The method includes: multimodal teaching data acquisition; spatiotemporal encoding of gaze features; construction of teleoperation datasets; construction of a multimodal large-scale model; model training; and intent recognition and real-time inference. The method provided in this application, based on the OpenVLA large-scale model, achieves a natural interaction mode of "one-time command issuance and long-term intent guidance" by introducing visual gaze features. Furthermore, through a "Gaze-Guided Loss" function, it achieves strong supervised guidance of the large-scale model's attention mechanism based on human visual priors at the model training level. While retaining the powerful cross-task, zero-shot generalization capabilities of the OpenVLA large-scale model, the method provided in this application significantly enhances the system's adaptability to complex and unstructured environments.

The application discloses a kind of multi-person vital sign monitoring method and system based on space-time coding hypersurface, utilize the harmonic wave beam scanning area of attention generated by space-time coding hypersurface to carry out human detection, and allocate the beam of frequency orthogonal for each detected human target, to accurately estimate their respiratory and heartbeat rate.The system can monitor the vital sign of multiple people simultaneously, and accurately estimate the respiratory and heartbeat frequency of human target.The control ability of space-time coding hypersurface used in the application can reduce the noise reflected by non-related objects, thereby improving the signal-to-noise ratio of human chest echo.The application uses wireless signals, has unique advantages such as not easily disturbed by environmental factors and protecting visual privacy.Compared with traditional vital sign monitoring instruments, the non-contact characteristics of the application can reduce the stress and interference on the user, and can be used for long-term, remote health monitoring.

This invention discloses a time series anomaly detection method and system based on multi-scale spatiotemporal modeling, comprising: extracting multivariate time series from business data; generating multiple sub-time series at different scales using one-dimensional convolution; independently encoding each scale sub-time series using a scale-independent spatiotemporal encoder to obtain spatiotemporal enhanced features of the multi-scale time series; employing a cross-scale hybrid expert mechanism to achieve information exchange between scales, obtaining a multi-scale sequence representation after scale interaction; integrating the interaction representations of each scale and performing decoding and reconstruction; generating anomaly scores by calculating the reconstruction error between the original input and the reconstructed sequence, and obtaining anomaly detection results. This method decomposes complex time series into multiple scales, with each scale collaboratively modeling spatiotemporal dependencies, which helps to more accurately discover and verify anomalous signals, improving the accuracy and reliability of detection.

The invention relates to a confocal laser endoscope video redundancy removing and filtering method, device and system, and relates to the technical field of computers, and the method comprises the following steps: receiving an input confocal laser endoscope CLE video clip; a video clip is processed through a parallel multi-path feature coding architecture, wherein the architecture comprises a microscopic feature coding path, a macroscopic space-time coding path and a time sequence stability coding path; dynamically integrating the feature vectors output by the three paths to generate a fused feature vector; on the basis of the fusion feature vector, a weighted ordinal regression classifier is used for generating classification output, the classifier considers the sequence relation between diagnostic value levels and allocates weights for different categories to deal with the problem of data imbalance, and the classification result accurately corresponds to a predefined operator cognitive intention stage. According to the method, the technical problem that key diagnosis fragments in confocal laser endoscope videos cannot be accurately and efficiently screened due to the design of'literal blindness' and'single integrality 'in the prior art is solved.

The invention relates to the technical field of intelligent logistics scheduling, and discloses a cargo transportation scheduling method and system based on artificial intelligence. The method comprises the following steps: performing standardized formatting preprocessing on original transportation task data to obtain standard task description data; task space-time coding is executed, and task space-time coding data containing time sensitivity weights and path conflict marks are generated; inputting the coded data into a trained deep neural network scheduling optimization model, and outputting a group of preliminary scheduling schemes; inputting the initial scheme into a multi-scheme game decision module, and selecting an optimal scheduling scheme based on a preset game rule and a real-time road condition information flow; and performing dynamic environment deduction on the optimal scheme, and generating an execution deduction path with a time label and a resource label. According to the method, constraint perception is enhanced through space-time coding, dynamic preferential selection is realized through game decision, and the accuracy and adaptability of cargo transportation scheduling are improved.

This invention relates to a multi-vehicle trajectory prediction method for complex dynamic traffic scenarios. It belongs to the field of intelligent transportation and autonomous driving technology, specifically focusing on a method for multi-vehicle trajectory prediction in complex dynamic traffic scenarios. The purpose of this invention is to address the problem of low accuracy in multi-vehicle trajectory prediction of existing models in complex dynamic traffic scenarios. The process is as follows: extracting the state information of all vehicles in a traffic scenario over a continuous historical time period; constructing a multi-vehicle trajectory prediction model for complex dynamic traffic scenarios; the multi-vehicle trajectory prediction model for complex dynamic traffic scenarios includes a multi-scale interactive spatiotemporal coding module and a target conditional flow matching trajectory generation and decoding module; obtaining a trained multi-vehicle trajectory prediction model based on a total loss function; acquiring the state information of vehicles in a traffic scenario over a continuous historical time period at the current moment; and obtaining the vehicle trajectory for the predicted time period based on the trained multi-vehicle trajectory prediction model.

A space-time coding array differential eddy current probe and a detection system and method thereof belong to the technical field of nondestructive testing, and are characterized in that a differential detection module is provided with a flexible printed circuit (FPC), M miniature planar coils with independent channels are integrated on the surface of the FPC, the M miniature planar coils are distributed according to a regular triangle array, and M is an integer greater than or equal to 2. The coil surface of each miniature planar coil is attached to the surface of the flexible printed circuit board FPC; coil centers of any three adjacent miniature planar coils are used as three vertexes of a regular triangle to form a group of differential detection units, the M miniature planar coils form M-2 groups of differential detection units, and each group of differential detection units comprises an excitation coil and two receiving coils; and the micro planar coil is alternately used as an exciting coil or a receiving coil according to the conversion of the currently excited differential detection unit. The method has the effects that the detection efficiency, the resolution and the signal-to-noise ratio of the dense defects of the metal matrix can be improved.

A video action recognition method based on multi-view spatiotemporal feature collaborative modeling relates to the technical field of video recognition, and introduces a guide refining network composed of a guide token generation module and a query guide view refining module on the basis of multi-view spatiotemporal features extracted by a spatiotemporal encoder sharing parameters. And in combination with cross-view-angle consistency constraint, multi-view-angle attention fusion and a category prototype alignment mechanism, collaborative modeling and selective enhancement, alignment and aggregation of multi-view-angle video action spatial-temporal features are realized, and the accuracy and robustness of multi-view-angle video action recognition are improved.

The invention relates to the technical field of radar signal processing and automatic tracking, in particular to an unmanned aerial vehicle tracking and trajectory prediction method based on radar plot information. The method comprises the following steps: acquiring and preprocessing a radar plot sequence; carrying out adaptive space-time coding on the sequence and extracting a physical constraint feature sequence; inputting the features into a motion mode analysis network to obtain a motion mode label and a mode enhancement feature sequence; inputting the space-time coding feature sequence, the physical constraint feature sequence and the mode enhancement feature sequence into a space-time hybrid module based on a Transform architecture, and performing acceleration increment prediction; correcting the prediction state by using adaptive Kalman filtering; and performing multi-step prediction based on the corrected state to generate a future trajectory. According to the method, the physically guided deep learning and the classical filtering technology are fused, so that the precision, the physical credibility and the system robustness of state estimation and trajectory prediction of the low, slow and small target in a complex environment are improved.

The invention discloses a video anomaly detection method based on multi-task learning, a model comprises a space-time encoder, a prototype memory network, a double-decoder architecture and a reconstruction decoder, and the method comprises the following steps: carrying out hierarchical feature extraction on a training video frame sequence by using the space-time encoder; performing memory enhancement processing on the spatial-temporal characteristics through a prototype memory network; respectively carrying out future frame prediction and current frame reconstruction through a double-decoder architecture; calculating a multi-task loss function, and updating parameters of the space-time encoder, the prototype memory network and the double decoders through back propagation according to the multi-task loss function; according to the method, a more comprehensive self-supervised learning framework is constructed, and the generalization ability of the model is enhanced.