239 results about "Multi targeting" patented technology

The invention relates to the technical field of intelligent vending, solves the problem that in the prior art, commodity identification cannot be accurately carried out in a multi-target scene, and provides a multi-target commodity identification method, device and system based on multi-modal data processing. The method comprises the following steps: acquiring multiple frames of real-time images in a commodity transaction scene; performing preprocessing and label information extraction on the real-time image, and determining character information corresponding to the target image and the commodity label; performing instance segmentation on the target image, and determining commodity position information; performing feature extraction on the target image, and determining commodity image feature information; according to pre-collected multi-source privatized data in an intelligent vending scene, performing fine adjustment and optimization processing on the open-source multi-modal visual language model to obtain a multi-modal large model; and inputting the commodity image feature information and the text information into the multi-modal large model for information fusion, and determining a commodity target identification result. According to the invention, commodity identification can be accurately carried out in a multi-target scene.

The invention relates to the field of commodity visualization, in particular to a commodity display interaction visualization method and device. The method comprises the following steps: collecting a multi-azimuth image of a commodity, carrying out three-dimensional texture modeling, and constructing a three-dimensional texture mapping model; performing material light rendering on the three-dimensional texture mapping model to generate a material rendering result; collecting an environment detection image of a commodity display environment, and performing environment illumination adaptation compensation on a material rendering result to obtain an illumination compensation rendering commodity; carrying out attribute information visual layout on the illumination compensation rendering commodity to obtain a commodity visual space; and carrying out interaction response animation analysis according to the commodity visualization space, carrying out multi-target parallel rendering, and executing commodity interaction visualization operation. The form and surface details of the commodity in the real world are accurately restored, the visual reality sense is improved, and the interactive experience feeling of browsing the commodity by a user is enhanced.

The invention discloses a video multi-target intelligent detection method and system based on mercuric chloride hardware. A hardware decoding module decodes an input video stream in real time, generates video frames and stores the video frames in a shared memory queue. And inputting the video frame into the YOLOv5 model converted by the mercuric chloride OMG tool, pre-loading a plurality of model instances into a memory by using an ACL interface of the mercuric chloride NPU, calling different model instances through a polling scheduling mechanism, and outputting structured data comprising a multi-target detection frame, a category label and confidence. And carrying out non-maximum suppression processing on the reasoning result, and judging whether the target is in an alarm monitoring area or not by adopting a central point detection method. According to the invention, real-time multi-target intelligent detection of the input video stream can be realized. And the decoded video frames are stored in a shared memory queue, so that efficient data transmission and processing are realized. The multi-model parallel reasoning improves the detection precision, and is suitable for the application scene of real-time video multi-target intelligent detection.

The invention discloses a multi-target and long-distance field domain temperature measurement method based on dual-spectrum imaging. A visible light camera module and an infrared camera module are included. Obtaining internal parameters and external parameters in the visible light camera module and the infrared camera module, and optimizing the internal parameters and the external parameters; performing quick image feature search and image feature stereo matching on the acquired visible light image and infrared image to realize identification and positioning during image measurement; registering and fusing the visible light image and the infrared image by adopting a double-spectrum image fusion algorithm to obtain a fused image; carrying out image intersection measurement positioning and temperature measurement area stereo matching on the image; performing systematic edge calculation on the image by adopting a data-based intelligent analysis algorithm, and analyzing a temperature change trend according to the obtained image data to obtain the temperature of the measured object; the current running state of the detected power equipment is evaluated, the equipment fault is diagnosed, and the application end is used for storing, displaying and processing the detection result; and the temperature detection accuracy of the power equipment is effectively improved.

The invention relates to the field of digital twinning, in particular to an intelligent decision analysis method and system for a digital twinning-based manufacturing system, and aims to improve the generalization ability and adaptability of a digital twinning model in an emerging market small sample scene through deep fusion of meta-learning and adversarial training. According to the method, the problem of prediction distortion caused by data scarcity is relieved, high-quality training samples are expanded in combination with a generative enhancement technology, a multi-target elastic scheduling mechanism is synchronously constructed, dynamic optimization and feasibility verification of a production strategy are realized in a digital twin environment, and finally a closed-loop optimized intelligent decision system is formed. According to the method, through deep fusion of meta-learning and adversarial training, by combining a generative enhancement technology to expand high-quality training samples and a synchronously constructed multi-target elastic scheduling mechanism, dynamic optimization and feasibility verification of a production strategy are realized in a digital twin environment, and finally a closed-loop optimized intelligent decision system is formed.

The invention provides a text code generation method based on multi-modal semantic embedding and a dynamic knowledge graph, and belongs to the technical field of artificial intelligence. According to the automatic code generation method based on multi-modal semantic embedding, the dynamic knowledge graph, constraint-driven code generation and context-aware repair, the semantic understanding precision and the code generation quality are remarkably improved by integrating the technologies of multi-head Transform, the graph neural network, reinforcement learning optimization, genetic algorithm sequence adjustment, the generative adversarial network and the like. According to the method, the knowledge graph can be dynamically constructed to enhance structured semantic modeling, high-quality codes conforming to specific industry standards are generated, functionality, efficiency and conciseness are considered, meanwhile, the iteration cost is reduced through an efficient error positioning and repairing mechanism, and the method is suitable for large-scale popularization and application. And the robustness and the flexibility of the system in diversified scenes are improved by utilizing adaptive weight optimization and multi-target balance.

The invention provides a multi-target fire positioning method and device based on multispectral dynamic fusion, and relates to the technical field of unmanned aerial vehicle inspection, and the method comprises the steps: obtaining a multispectral image sequence collected by an unmanned aerial vehicle in an inspection task execution process; according to the visible light image and the infrared image contained in the image frame group, smoke identification processing, fire source identification processing and result dynamic fusion processing based on confidence are carried out, and a target pixel position identified as a target area in the image frame group is obtained; and performing ray inversion by taking the pose data of the unmanned aerial vehicle as positioning compensation based on the target pixel position identified as the target area in the image frame group, and determining the target longitude and latitude position identified as the target area in the image frame group. According to the invention, the method can achieve the recognition and positioning of a plurality of target regions during the cruise task execution process of the unmanned aerial vehicle, facilitates the improvement of the positioning efficiency of the target regions, and also can remarkably improve the positioning precision of the target regions.

In various embodiments, systems and methods for generating interfaces including interface elements representative of next-best actions are disclosed. A request for an interface including a set of features representative of a user associated with the request is received and a user state representation including an implicit user state representation and an explicit user state representation is generated based on the set of features and session data for at least one session associated with the user. An action reward value for each of a plurality of candidate actions is generated based on the user state representation and an interface including at least one interface element representative of a candidate action having a highest action reward value is generated.

The invention discloses a visual tracking method for a target aircraft target to identify a high-dynamic target, and relates to the technical field of dynamic visual tracking. According to the method, multi-source information of a visual frame stream and an event stream is fused, and the definition and stability of an initial envelope are ensured through the steps of infrared enhanced filtering, local contrast reserved filtering, multi-scale ROI screening, unified perspective correction and the like; in the trajectory generation and paragraph division process, two-norm fusion analysis is performed according to high-dynamic target maneuvering characteristics, trajectory mutation can be caught acutely, and the rationality of trajectory paragraph division is improved; a cross-level backtracking compensation mechanism further corrects envelope deviation caused by visual delay or jitter in high-speed motion, breakage and drifting of a target trajectory are effectively reduced, the accuracy of trajectory attribution can be kept under the condition of multi-target overlapping, target confusion is avoided, and layered early warning of target aircraft off-target behaviors is achieved.

The invention relates to an airport intelligent bird repelling system based on multi-target detection optimization, and belongs to the technical field of image recognition. The system comprises a detection unit, an identification unit, a repelling unit and an intelligent optimization unit, the detection unit obtains the three-dimensional coordinate information of a whole airspace target, regularly scans the airspace around an airport, continuously tracks and updates the position information of the airport in real time, and then transmits the information to the visual confirmation module. And the visual confirmation module performs visual verification on the bird target found by the radar detection module. And the identification unit performs dynamic identification by using the target tracking model, and performs classification and threat level evaluation on the identified birds according to a preset bird threat level. According to the evaluation result, the repelling unit takes corresponding bird repelling measures; and the intelligent optimization unit optimizes a bird repelling strategy according to the historical bird repelling data and the real-time environment information. The system improves the accuracy and real-time performance of bird detection through a multi-target detection technology.

The invention provides an advertisement putting dynamic game decision-making method and system based on multi-objective optimization, and the method comprises the steps: obtaining user behavior key data which comprises user visual focus position information; transmitting the user visual focus position information to a multi-target optimization model in a preset decision server, and generating multi-target optimization model input data; analyzing the input data of the multi-target optimization model and a preset game constraint condition by using the multi-target optimization model to generate a multi-target game decision parameter; and carrying out dynamic game deduction calculation based on the multi-target game decision parameters, and generating a dynamic advertisement position adjustment scheme so as to carry out advertisement putting dynamic game decision. According to the method, millisecond data is transmitted to the multi-target optimization model, the problem of strategy oscillation caused by data delay and target conflict in traditional advertisement decision making in a real-time attention scene is solved, and accurate dynamic matching of user focus drift and advertisement display is realized.

An additive manufacturing real-time process parameter optimization method based on reinforcement learning driving comprises the steps that firstly, a multi-mode online monitoring hardware platform is constructed, and a visible light camera, a thermal imaging camera and an acoustic sensor are integrated to sense the working condition of the additive manufacturing process in real time in an omnibearing mode; secondly, extracting key features, including visible light images, thermal imaging and acoustic signal branches, of modal data on line through a lightweight CNN-Transform hybrid network, and generating a unified low-dimensional state vector through convolution feature extraction and fusion of a Transform encoder; then, a reinforcement learning model is established, a multi-target reward function is designed in combination with the extracted feature data, a process jitter penalty term, an overheating penalty term and the like are covered, and dynamic mapping of parameters and performance is achieved. And finally, an online learning and real-time decision-making system is deployed, the trained strategy network is embedded into manufacturing equipment, self-adaptive adjustment and closed-loop control of technological parameters are achieved, and the stability and product performance of the additive manufacturing process are effectively improved.

A method and an apparatus for multi-target domain adaptation are disclosed. According to one aspect of the present disclosure, the provided computer-implemented method for multi-target domain adaptation of a model that performs an image-based task comprises the steps of: integrating a plurality of different target domains into a target superdomain; masking at least some images in the target superdomain so as to generate one or more masked target images; and updating parameters of the model on the basis of the prediction of the model for the target superdomain and the masked target images.

The invention discloses a code generation method and system based on multi-objective reinforcement learning, and belongs to the technical field of digital integrated circuit design, and the method comprises the following steps: S1, generating an SFT training data set; s2, generating an MORL data set; s3, performing supervision and fine tuning on the model by taking the SFT training data set as input, so that the model masters the basic RTL code generation capability; s4, taking the MORL data set as input to train the model after supervision and fine tuning in the S3, and realizing multi-objective optimization; and S5, according to a user demand, describing a multi-target constraint through vector rewards, screening non-dominated candidate codes by using Pareto optimization to obtain an optimal RTL code, and outputting the optimal RTL code, vector reward details and a PPA equilibrium curve. According to the method and the system, basic constraints are met, and meanwhile generation of RTL codes with PPA indexes capable of being customized according to requirements is achieved.

The invention discloses a multi-modal multi-target tracking method and a related device, and relates to the technical field of multi-target tracking, and the method comprises the steps: carrying out the fusion processing of a multi-target-containing image detection result and a multi-target-containing point cloud detection result corresponding to a detection region at a current moment, and obtaining a first-class target set and a second-class target set, according to the track corresponding to the detection area at the previous moment, predicting to obtain a three-dimensional track state estimation value and a two-dimensional track state estimation value corresponding to the detection area at the current moment, and matching and associating the targets in the first type of target set with the three-dimensional track state estimation value, and matching and associating the targets in the second type of target set with the two-dimensional trajectory state estimation value, and finally determining the trajectory corresponding to the detection area at the current moment. According to the invention, multi-target tracking based on multi-modal data fusion can be realized, and the tracking effect is improved.

The invention discloses a cross-platform performance transplantable congestion degree calculation method and system, and relates to the technical field of multi-objective evolutionary computation.The method comprises the steps that a population individual structure body is designed, the number of individuals is verified, filling is conducted according to needs, a global ordered individual array is obtained by designing and merging topological recursive sorting, and then the congestion degree of a non-dominated sorting layer is calculated. According to the method, the problems of high vectorization and parallelization difficulty and low calculation efficiency in the existing congestion degree calculation process are solved, the cost of a large amount of redevelopment and manual adjustment and optimization during performance optimization on different hardware platforms is reduced, the solving precision and Pareto frontier coverage capability of a multi-objective evolutionary algorithm can be improved, and the calculation efficiency of the multi-objective evolutionary algorithm is improved. The algorithm is prevented from falling into a local optimal solution, and practical application requirements are met.

The invention relates to a microgrid operation method and system based on carbon flow tracking and multi-target collaborative optimization, and the method comprises the following steps: S1, carrying out the high-precision, real-time and synchronous data collection and processing, and providing complete and reliable data input for carbon flow tracking and optimal scheduling; s2, according to the data acquired in the S1, realizing carbon flow dynamic tracking through a directed graph, quantifying the carbon emission intensity of each node in the microgrid, introducing a carbon label technology, tracking the carbon emission source of each kilowatt-hour, and providing a data basis for optimization; s3, by taking minimization of carbon emission and operation cost as targets, a long-term scheduling plan is generated for the equipment, and multi-target collaborative optimization is realized; and S4, further optimizing the energy storage system on the basis of realizing the multi-target collaborative optimization in the step S3 so as to improve the demand response speed. According to the invention, the carbon emission reduction rate and the demand response speed during the operation of the industrial micro-grid can be effectively improved.

The invention discloses a flood storage and detention area change pattern spot identification label generation technology based on style decoupling. The technology comprises the following steps: S1, constructing and preprocessing a change detection data set; s2, constructing a conditional diffusion generation network based on a decoupling encoder; s3, decoupling extraction and orthogonalization representation of content-style features are carried out; s4, constructing a multi-target training strategy and two-stage model training; s5, injecting and fusing style features based on cross attention; s6, cross-domain style migration and diversified label generation; and S7, based on label quality screening of physical and semantic double constraints, outputting a high-quality change detection expansion data set. Compared with the prior art, the method has the advantages that by introducing a content-style decoupling mechanism, the style and the content of the generated sample are independently and accurately controlled, and the change detection label which is consistent in ground feature layout, diversified in imaging style and accurately labeled at a pixel level is generated.

The invention discloses a multi-target adaptive scheduling method based on a light storage integrated system, belongs to the technical field of power distribution network optimization, and aims to improve the electric energy utilization efficiency of a charging station. Firstly, a multi-objective optimization model is established by taking maximization of the electric energy utilization rate, the charging pile satisfaction rate and the photovoltaic consumption degree as objectives; secondly, in order to avoid the premature problem of the traditional NSGA-III algorithm, a self-adaptive binary mutation operator is constructed to improve the population diversity; and finally, aiming at the problem that a Pareto optimal solution set obtained by multi-objective optimization solution is difficult to screen, a unique optimal solution is selected from the Pareto optimal solution set by adopting a fuzzy analytic hierarchy process, and the strategy can better realize optimal utilization of electric energy.

The invention discloses a multi-target tracking method for an offshore rain, fog and jitter environment, and the method comprises the steps: obtaining multi-target image data in the offshore rain, fog and jitter environment, and carrying out the target cutting and fusion of the multi-target image data based on a target enhancement segmentation strategy, and obtaining an enhanced sample image; mapping the enhanced sample image and target features in the multi-target image data into a candidate frame set of a pixel scale, and obtaining a sample data set containing adaptive Anchors according to the candidate frame set based on a clustering algorithm; performing model training on the multi-target detection network through the sample data set to obtain an optimal detection model so as to realize target detection; and defining a target state vector and a target observation vector according to a detection result, and realizing multi-target tracking under the marine rain and fog and jitter environment based on an improved Kalman filtering algorithm. The problems that in the prior art, the precision of multi-target detection under the marine rain and fog and jittering environment is insufficient, and a systematic solution for multi-target tracking under the marine rain and fog environment and the jittering scene is lacked are solved.

The invention discloses a long tail recommendation method and device based on heterogeneous-homogeneous view comparative learning, and the method comprises the following steps: heterogeneous-homogeneous composition joint modeling: constructing a graph structure containing three perspectives, and respectively modeling user-article interaction, a high-order cooperation relationship and implicit semantic association; cross-view contrast learning optimization: heterogeneous-isomorphic information fusion is realized through a residual aggregation graph encoder RAGE and dual-granularity contrast loss; and carrying out multi-target joint training. Semantic distortion caused by random disturbance can be avoided, the influence of hot and unpopular nodes is balanced, and long-tail distribution deviation is relieved.

The invention provides a multi-target seamless tracking method and system based on active prediction and feature pre-fusion, and the method comprises the steps: obtaining real-time video streams of a plurality of intelligent cameras for registration, and obtaining a unified fusion video stream under a global geographic space coordinate system; a plurality of dynamic targets are recognized according to the fused video stream, a globally unique identity label is distributed to each dynamic target, a three-dimensional space prediction track and a virtual relay area of each dynamic target are determined by combining the fused video stream, and a relay intelligent camera is determined according to a shooting condition index; the relay intelligent camera is aligned with the corresponding virtual relay area, when the relay intelligent camera recognizes the dynamic target in the virtual relay area, the master control authority is switched to the relay intelligent camera for seamless tracking, and the virtual relay area where the dynamic target appears is actively predicted in a prospective mode for alignment of the relay intelligent camera; shooting view is ensured, and target losing is avoided.

The invention relates to a medical video classification method based on multi-target segmentation prior driving, and belongs to the technical field of video recognition. The method comprises the following steps: collecting and preprocessing a medical video frame sequence to obtain a video original frame sequence; constructing a multi-target segmentation-classification data set, and dividing the multi-target segmentation-classification data set into a training set and a test set; constructing a medical video classification model based on multi-target segmentation prior driving, wherein the medical video classification model comprises a feature extraction-time sequence coding-fusion network and a classifier; training the model by using data in the training set; in the training process, the model is optimized through a loss function, and a trained model is obtained; and inputting data in a test set into the trained model to obtain a medical video classification result. According to the invention, the accuracy and generalization ability of medical video classification can be enhanced.

The invention discloses an industrial internet multi-service real-time management method for industrial intelligence, which relates to the technical field of artificial intelligence, and comprises the following steps: establishing income models and multi-target benefit models of content providers, infrastructure providers and internet service providers in a mobile edge computing scene; a multi-agent Markov decision model and a Stackelberg game model are adopted to establish an optimal solution model of each object under the condition of ensuring own interests; a multi-agent layered reinforcement learning architecture is constructed, a centralized leader strategy is formulated by using global information, a follower network is trained according to the leader strategy, the decision of each follower is adjusted, strategy optimization is realized, and a network transmission mechanism which preferentially guarantees key services is adopted. Therefore, by the adoption of the industrial internet multi-service real-time management method for industrial intelligence, benefit maximization of the system can be efficiently coordinated in the complex environment of the mixed action space, and the overall performance and the resource utilization rate of the system are remarkably improved.

The invention relates to a digital twinning-based resource allocation method for balancing the scene quality and cost of the Internet of Vehicles, and belongs to the field of mobile communication. The method comprises the following steps: constructing a digital twin-driven vehicle edge computing system architecture, and carrying out data sensing and uploading modeling on a real-time state of a vehicle physical environment; designing a new index for quantitatively evaluating system quality and cost to represent a tradeoff relationship between the digital twin quality and the system cost; establishing a multi-objective optimization problem to simultaneously realize system quality maximization and cost minimization; a multi-agent multi-target deep reinforcement learning algorithm is adopted, replay experience is stored through a distributed actor network, a learner with a lesion evaluation network is adopted to evaluate agent actions, and an optimal resource allocation strategy is solved. According to the method, the contradictory problem between the digital twin quality and the digital twin cost is solved by jointly optimizing the digital twin quality and the digital twin cost, and the resource overhead is remarkably reduced while the precision of the twin model is ensured.

The invention relates to a disaster situation video description generation method for an emergency disaster scene. The method comprises the following steps: carrying out frame-level sampling processing and multi-class target detection on a disaster video stream; calculating the closeness degree of the target in the space to construct an apparent visual image; constructing a hidden layer visual map; fusing the apparent spatial features and the latent spatial features to obtain fused features; splicing and mapping the alignment features and the fusion features to obtain cross-modal splicing features; and generating a word sequence from the cross-modal splicing features, and performing optimization training on the word sequence. According to the invention, based on an apparent space and latent space double-level visual correlation modeling mechanism, the multi-target structure relation modeling capability in a disaster video and the semantic accuracy of description generation are improved. According to the method, a visual association graph is constructed from two dimensions of dominant structure association and implicit semantic association between targets, and a graph convolutional network is introduced to perform high-level semantic feature extraction on a graph structure, so that the problem of description information missing caused by scene chaos and association complexity is effectively relieved.

The invention provides an endoscopic surgery video multi-target segmentation method based on a basic segmentation large model and mixed expert fine tuning. The method comprises the following steps: preprocessing a data set, establishing an SAM2 baseline segmentation network, constructing a hierarchical hybrid expert module, constructing a stage gating network, establishing an endoscopic surgery video segmentation network, training the endoscopic surgery video segmentation network, and automatically segmenting an endoscopic surgery video by the segmentation network. According to the invention, accurate segmentation of different operation scenes can be realized. The hierarchical hybrid expert module and the stage gating network are introduced, and the problems of scene difference and multi-organization segmentation commonly existing in endoscopic surgery video segmentation tasks can be well solved.

The invention discloses a multi-target interactive behavior retention monitoring video concentration method based on semantic enhancement and space-time constraint, which belongs to the technical field of video concentration and mainly comprises the following steps: acquiring a video background through a Gaussian mixture model MOG2 in combination with random frame sampling and a median fusion strategy; performing multi-target tracking and behavior identification on the video by using a DeepSORT algorithm and a deep learning model based on CNN + GRU architecture, and extracting a target trajectory, a spatial position, a moving direction, a moving speed and behavior label features; according to the spatio-temporal characteristics and the semantic information, interaction behaviors between the targets are judged, and the targets with interaction are divided into the same target tube group; the collision constraint and the time domain constraint are integrated, a new time label is distributed to the target tube group through a greedy strategy and a space hash algorithm, and meanwhile sequential logic is optimized by means of a sequential loss function; and fusing the tube group with the background to generate a concentrated video. The video concentration method provided by the invention can accurately retain the interaction behavior between the targets.

The invention discloses a multi-target real-time monitoring system and method for rearview mirror lens production and processing, relates to the technical field of automobile part manufacturing, and aims to solve the problem that an existing monitoring system cannot effectively distinguish false abnormity caused by dynamic environment interference in a rearview mirror lens crimping process and the real defect of the crimping process. The system comprises a data acquisition module used for acquiring first monitoring data, second monitoring data and third monitoring data of a crimping process in real time, the first monitoring data being visual image data of the crimping state of a lens and a supporting plate, the second monitoring data being visual image data of the crimping state of the lens and the supporting plate, and the third monitoring data being visual image data of the crimping state of the lens and the supporting plate; the second monitoring data is dynamic environment data of the crimping station, and the third monitoring data is crimping process operation data. The method has the advantages that the environmental interference and the coupling type of the rearview mirror lens crimping process are accurately identified, false abnormity and real defects are distinguished, and the monitoring misjudgment rate and the crimping reject ratio are reduced.

The invention relates to the technical field of advertisement putting, in particular to an automatic advertisement bidding method and system based on predictive putting planning, and the method comprises the steps: constructing a strategic putting picture composed of multi-dimensional advertisement activity measurement vectors based on historical advertisement putting data, and extracting a high-quality strategic putting picture; the advertising demand of the advertiser is coded into a standardized commercial intention code; a forward disturbance and conditional reverse recovery process training conditional picture synthesizer is constructed by using a high-quality strategic release picture and commercial intention coding, and in an actual bidding stage, starting from an initial strategic release picture, an optimal strategic release picture meeting a release requirement is gradually recovered and generated through the conditional picture synthesizer; and constructing and training a non-backtracking bid command derivation device based on the optimal strategic release picture, and generating a bid command in real time. According to the method, the multi-target and multi-constraint long-term delivery requirement proposed by an advertiser can be met, so that a stable, flexible and optimal-effect advertisement bidding strategy is difficult to realize on the whole.