25results about How to "Accurate expression" patented technology

The invention discloses a power cable heating state monitoring and fault prediction method based on artificial intelligence, and relates to the technical field of cables, and the method comprises the steps: collecting power cable data in an operation state, and marking the collected power cable data; a training framework of the state monitoring and fault prediction model is a meta learning and transfer learning fusion framework; a neural network based on feature dynamic influence is adopted; and an auto-encoder based on feature path decoupling is adopted. According to the method, the feature independence is optimized through a feature path decoupling mechanism, the independence among the power cable data features is evaluated and optimized through a regularization loss function, in addition, linear dependence among the power cable data features is calculated by adopting covariance, the expression ability of the independent power cable data features is enhanced through a penalty mechanism, and the power cable data features are more accurately extracted. The processing capability of the model to the power cable data is improved, and the model can still keep accurate expression of the power cable data structure even under the condition that the power cable data features are missing.

The invention discloses an AFDX (Avionics Full Duplex Switched Ethernet) data processing method, and relates to the technical field of data processing. Comprising the following steps: step 1, data acquisition; step 2, data storage; 3, data preprocessing; and 4, data processing. According to the AFDX data processing method, an initial distillation sample is obtained by sampling an initial data set, and a first expression loss value of each initial sample is calculated according to the similarity between the initial distillation sample and each initial sample in the initial data set; therefore, whether the initial distillation sample can completely express each initial sample in the initial data set or not can be determined, so that the sample data can express each initial sample in the initial data set more completely and more accurately.

The invention discloses an operation area electronic fence automatic generation and intrusion alarm method and a medium in the technical field of electronic fences. The method comprises the steps of obtaining a full-scene standard three-dimensional model based on three-dimensional data of a transformer substation building and primary equipment and shooting data of secondary equipment; a digital model is constructed based on key information of typical work tickets and secondary safety measure tickets, correlation mapping of the digital model and a full-scene standard three-dimensional model is established through a global unique identifier to generate digital expression, and intervals and screen cabinets of operation and inspection work are divided into a plurality of work areas. Automatically generating a virtual electronic fence based on PRM sampling path planning and collision early warning; different monitoring precision virtual electronic fences are set for indoor and outdoor environments according to the type of a working area, the position of an irregular obstacle is obtained in real time, and an alarm is triggered immediately when an intrusion behavior is found. According to the invention, automatic generation of the electronic fence and intrusion alarm are realized, the operation safety risk is reduced, and the human resource configuration and process management and control efficiency are optimized.

The invention discloses a non-pairing cross-modal retrieval method based on entity semantic guidance and relation fusion. The method comprises the following steps: extracting entity image features and entity text features from an entity data set; obtaining an entity public semantic space through negative sample selection and intra-modal neighbor relation constraint based on the entity image features and the entity text features; based on the entity public semantic space, a high-confidence image-text pseudo-pairing set is obtained through a pseudo-pairing generation and screening method; on the basis of the high-confidence image-text pseudo-pairing set, relation information is captured and aligned through a relation fusion module, and a fused public semantic space is obtained; and calculating the similarity between the images and texts based on the fused public semantic space, sorting according to the similarity, and outputting a cross-modal retrieval result. Compared with other unpaired cross-modal retrieval methods, the method has the advantages that the performance is obviously improved, a space obtained by learning can accurately express data from a deeper level, and the whole method is better in interpretability.

The invention provides an image semantic communication method and system based on a multi-modal large language model (MLLMs), and aims to improve the semantic accuracy and system robustness of image transmission. The method comprises the following steps: extracting semantic features from an image source, detecting through an expert AI model, processing an OOD problem in combination with a general AI model, optimizing recognition precision by using a Bayesian method, encoding the features and transmitting the encoded features through a wireless channel. And after decoding by a receiving end, reconstructing the image by using a generation-evaluation framework based on the MLLMs and performing iterative optimization so as to ensure accurate reconstruction of semantic information of the image. According to the method, the OOD data processing capability is particularly enhanced, the image reconstruction quality is improved through the generation-evaluation framework, the limitation of a traditional model on new data distribution is solved, and the reliability of the system and the accuracy of image reconstruction are enhanced.

The embodiment of the application provides a video generation method, system, electronic equipment, storage medium and program product, relates to the computer technical field, and the method comprises the following steps: identifying the skeleton structure information of a target role from a reference image containing the target role; acquiring a reference posture sequence used for representing reference posture changes; adjusting the skeleton structure information in the reference posture sequence by using the skeleton structure information of the target role, and generating a target posture sequence, wherein the skeleton structure information in the target posture sequence matches the skeleton structure information of the target role, and the posture change in the target posture sequence matches the reference posture change; and generating a target video according to the target posture sequence and the reference image, wherein the target video expresses the reference posture changes based on the target role. The technical scheme of the embodiment of the application can avoid the problem that the deviation of the target video and the target role in the reference image is large, and accurately expresses the reference posture changes in the reference posture sequence.

This invention discloses a robot interaction method, device, electronic device, and storage medium. The method includes: detecting first pose parameters of the first facial target in response to a first condition being triggered by a first facial target; detecting first feature information of the first facial target in response to a angular difference between the facial orientation indicated by the first pose parameters and the optical axis of the target robot's camera being less than a preset angular difference, wherein the first feature information is used to indicate facial feature information and gesture feature information; determining the task intent associated with the first gesture feature information from multiple sets of task intents associated with the first gesture feature information based on the first gesture feature information in response to a matching degree greater than a preset matching degree of the first feature information and a third feature information among multiple sets of second feature information; and controlling the target robot to execute a corresponding feedback action based on the task intent associated with the first gesture feature information. This invention can improve the environmental anti-interference capability and intent recognition accuracy of robot interaction.

The invention discloses an intelligent question and answer method and system for enterprise safety and environmental protection management, and belongs to the technical field of intelligent question and answer. The method comprises the following steps: acquiring a security and environmental protection problem asked by a user, analyzing the security and environmental protection problem asked by the user by using a security and environmental protection semantic analysis model to obtain an analysis result, and judging whether the security and environmental protection problem asked by the current user belongs to the field of security and environmental protection systems or not according to the analysis result; if the method belongs to the field of security and environmental protection systems, rewriting a security and environmental protection problem proposed by a user by using a statement rewriting model to obtain an instruction statement; searching a rule unit in a security and environmental protection knowledge base based on the instruction statement to obtain a search result; and the recall summary model extracts content directly related to the user intention from the search result, and generates a safety and environmental protection answer. According to the method, the corresponding safety and environmental protection result can be generated in time according to the questions of the employees, the operation risk increase is reduced, and the safety and environmental protection system awareness rate of the employees is improved.

The present application provides a method and system for remote sensing assessment of surface anomaly intensity based on multi-scale features, which belongs to the field of remote sensing data processing technology. A conceptual model of surface anomaly intensity is constructed from four scales: pixel scale, structure scale, object scale, and scene scale; a mathematical model of surface anomaly intensity is constructed based on the conceptual model of surface anomaly intensity; a remote sensing assessment model of surface anomaly intensity is constructed based on the conceptual model and the mathematical model; remote sensing image data is used as basic data to measure each remote sensing feature in the remote sensing feature system of surface anomaly intensity, and the weight of each remote sensing feature is calculated using the entropy weight method, and then the value of the comprehensive surface anomaly intensity of the study area is calculated through the mathematical model of surface anomaly intensity according to the value of each remote sensing feature and its corresponding weight. This scheme can adaptively adjust the feature weights at each scale, improving its ability to be applicable to different types of surface anomaly intensity assessments.

This invention discloses a structure-aware Gaussian localization and mapping system based on multi-Gaussian refinement, belonging to the fields of computer vision and robot navigation technology. It aims to solve the problems of existing Gaussian localization and mapping systems, such as underutilization of scene geometry, easy Gaussian map degradation, unstable tracking, and poor map reconstruction quality. Taking RGB and depth images as input, it includes three core modules: structure-aware tracking, structure-enhanced mapping, and structure-driven keyframe management. It achieves color-independent stable tracking through joint modeling of global geometry and Gaussian covariance, employs a two-layer Gaussian map to balance real-time performance and reconstruction accuracy, introduces transparency regularization to reduce Gaussian redundancy, and combines structure-driven keyframe selection and dynamic weight allocation to optimize map performance. This invention improves the tracking robustness, map representation capability, and real-time performance of the system in complex environments, and can be widely applied in fields such as robot navigation and augmented reality.

The invention discloses a prediction time coding method based on large language model prior, and belongs to the technical field of marine meteorology. According to the method, periodic and nonlinear change rules of ocean state variables on different time scales are systematically coded into time embedding, and the expression ability of time features is improved. According to the method, the prior knowledge of a large language model is utilized, the extraction and mapping process of time features is optimized, the time embedding quality and efficiency are improved, time information is converted into context semantic embedding, and accurate expression and efficient utilization of the time features are achieved. The forecasting precision can be improved, and error accumulation is reduced; the model training and maintenance cost is reduced; the time feature expression ability is enhanced, and the model generalization ability is improved; the operation convenience and practicability of the forecasting system are improved; resource utilization is optimized, and energy consumption is reduced; and the reliability and stability of the forecasting system are improved.

The present application discloses a video feature extraction method, device, equipment and computer storage medium, belonging to the field of video processing technology. In the training of the video feature extraction model, the method uses the topic tags added when the sample video is released as the real video tags, so there is no need for manual labeling, avoiding the high error rate caused by manual labeling, improving the accuracy of the trained model, and the model training efficiency is higher. In addition, the topic tags added when the sample video is released are usually closely related to the content of the video, so they can more accurately express the semantics of the video and are more in line with the actual video application scenario, so that not only can the trained model accurately express the semantic content of the video, but also can be more in line with the actual video application scenario.

The present application provides a method and system for supporting online editing and automatic generation of user manuals, which relates to the technical field of software development. The method includes: obtaining user input data content; converting the data content into multiple data items, the data items including type values ​​and value values; judging whether the type type of the data item is a content type according to the type value; if the type type is determined to be a content type, converting the value value into a DOM tree; according to preset rules, converting the DOM tree into a Markdown document. If the type type is determined to be a title type, converting the value value into a Markdown document according to the title level of the data item. The present application can automatically convert documents into computer syntax to improve the efficiency of user manual generation.

The present disclosure provides an image generation method, device, computer equipment and storage medium. The method comprises: obtaining an initial image and an initial text; determining a segmentation scheme according to the size of the initial image, segmenting the initial image according to the segmentation scheme to generate at least one segmented image; using an image encoder to extract features of the initial image and the at least one segmented image respectively to obtain global features and local features, and fusing the global features and the local features to obtain an image encoding result; and inputting the image encoding result and the initial text into a preset image generation model to generate a target image.

The present application discloses a method, system, device and readable storage medium for generating a three-dimensional reconstruction model, the method comprising: calculating the current signed distance function value from each spatial unit to the surface of the first three-dimensional reconstruction model; decomposing and weightedly fusion each current signed distance function value with the initial signed distance function value of the spatial unit in different directions, and updating the initial signed distance function value to the obtained fusion result; determining whether the current data frame is the last frame; if so, establishing a second three-dimensional reconstruction model according to each initial signed distance function value. The present application decomposes and weightedly fuses the current signed distance function value with the initial signed distance function value, realizes a more accurate recording and expression of the reconstruction process data, solves the problem of the accuracy of the reconstruction process data recording, solves the problem of the smoothing of the reconstruction result details caused by the data fusion process, better retains the reconstructed geometric details, and improves the accuracy of the three-dimensional reconstruction model.

The embodiment of the invention provides a state publishing method and device, equipment and a storage medium, and relates to the technical field of computers, and the method comprises the steps: responding to a scene viewing operation triggered by a target object, and presenting a public scene page, wherein the virtual character information at least comprises respective published virtual characters and state description information of each contact object having an association relationship with the target object; presenting an editing page in response to the triggered switching operation, wherein the editing page comprises a plurality of candidate virtual roles associated with the target object; in response to a triggered editing related operation, presenting a target virtual character selected from the target pair in the editing page, and presenting target state description information set by the target object for the target virtual character; and in response to the triggered release operation, presenting the target virtual character and the target state description information in the public scene page. The method is used for improving information reading efficiency and information expression accuracy.

This invention provides a video search method, apparatus, device, and storage medium based on cross-modal retrieval, belonging to the field of data processing. The method includes: acquiring retrieval text data and converting it into background and foreground images; searching for each video to be searched based on the background images to obtain a first video search result; searching for each video to be searched based on the foreground images to obtain a second video search result; and determining the video search result corresponding to the retrieval text data based on the first and second video search results. This invention can accurately represent the background and target object of the retrieval text data, shorten retrieval time, and improve retrieval accuracy.

The application discloses an infrared and visible light visual information fusion method based on gradient transformation prior, and the method comprises the following steps: acquiring a registered infrared image and a registered visible light source image, wherein the registered infrared image and the registered visible light source image represent a consistent geometric transformation relationship in space; constructing a gradient transformation prior network; inputting the registered infrared image and the registered visible light source image into the gradient transformation prior network for feature extraction image fusion processing to obtain a reconstructed fusion image. By using the application, the detailed texture information in the visible light and the thermal radiation target information in the infrared image can be accurately extracted, and the fusion quality of the image is improved. The application can be widely applied to the field of image fusion technology as the infrared and visible light visual information fusion method based on the gradient transformation prior.

The present application provides a driving behavior data analysis method and system for premium management. By identifying each driving behavior event included in the user driving behavior data set to be analyzed, the driving behavior event identification result is obtained, and an event feature description array of each driving behavior event is generated. The behavior sequence relationship between each driving behavior event included in the user driving behavior data set to be analyzed is extracted based on the event feature description array of each driving behavior event to obtain a driving behavior event relationship description array. Based on two features, whether the user driving behavior data set matches the reference driving behavior data is identified, thereby obtaining a driving behavior pattern and formulating a premium strategy. The present application expresses the driving behavior data more accurately based on the event feature description array of the driving behavior event and the driving behavior event relationship description array, thereby improving the accuracy of matching and identification, helping to obtain a more accurate driving behavior pattern, and formulating a more accurate premium strategy.

Embodiments of the present application provide an acoustic feature acquisition method and device and a readable storage medium. The method comprises: acquiring a target phoneme sequence; performing encoding processing on the target phoneme sequence based on a split self-attention mechanism to obtain a first feature matrix; and performing decoding processing on the first feature matrix based on the split self-attention mechanism to obtain an acoustic feature of the target phoneme sequence. The split self-attention mechanism converts the unified processing multi-head attention in the multi-head self-attention mechanism into split multi-head attention by performing scale transformation on the key-value matrix and the value matrix in a preprocessing manner, realizes modeling of information at different levels and different scales at the same time, takes into account the overall and detailed grasp, improves the feature analysis capability of the phoneme feature, better meets the needs of emotional speech synthesis, and makes the expression of the finally synthesized speech more accurate and vivid.

The present invention discloses a recombinant human type IV collagen protein and a preparation method and application thereof. The amino acid sequence of the recombinant human type IV collagen protein is shown in SEQ ID No. 2. The present invention optimizes the design of the DNA sequence encoding the fusion protein, so that type IV collagen protein is more suitable for expression in Pichia pastoris, and then the mature peptide fragment of human type IV collagen protein is expressed through the Pichia pastoris system; the expressed recombinant protein peptide is endotoxin-free, has a high yield, a simple culture medium, stable expressed protein, and can be secreted to the extracellular space for easy purification. The recombinant humanized type IV collagen protein has the effects of promoting proliferation, promoting migration and anti-aging, and the skin care products prepared therefrom also exert anti-wrinkle and wrinkle-removing effects on human skin. Therefore, the recombinant human type IV collagen protein of the present invention has broad application prospects.

The present application relates to the technical field of oral health, and particularly relates to a pregnant woman oral health examination management system and method, the system comprising: a gestational week physiological characteristic analysis module for constructing a corresponding oral sensitivity characteristic vector of each gestational week; a morning sickness characteristic analysis module for constructing a gastric acid erosion risk characteristic vector; a tooth characteristic analysis module for constructing a tooth acid resistance characteristic vector; an erosion risk assessment module for coupling analysis of the gastric acid erosion risk characteristic vector and the tooth acid resistance characteristic vector, and evaluating the tooth surface acid erosion risk caused by vomiting; and an oral health risk assessment module for joint modeling based on the oral sensitivity characteristic vector and the tooth surface acid erosion risk evaluation result, and generating a pregnant woman individualized oral health risk assessment result. The present application quantifies the gastric acid erosion risk characteristic caused by morning sickness, the tooth acid resistance characteristic, and the gestational week oral sensitivity characteristic, and effectively improves the prediction accuracy and reliability of the oral health risk during pregnancy.

The present invention relates to the field of commercial data analysis technology, and in particular to an intelligent demand prediction and analysis method and system for commodity auctions. The method comprises the following steps: acquiring auction item data; analyzing auction item feature data according to the auction item data; performing dynamic association network analysis of the auction item based on the auction item feature data to generate a dynamic auction item association network; detecting user interaction data; performing user interaction behavior feature analysis and processing according to the user interaction data to generate user interaction behavior feature data; performing instant user demand auction item probability prediction based on the user interaction behavior feature data to generate instant user demand auction item probability prediction data; transmitting the instant user demand auction item probability prediction data to the dynamic auction item association network to perform instant user associated demand auction item intelligent analysis to generate instant user associated demand auction item prediction data. The present invention achieves the improvement of the accuracy and efficiency of demand prediction in the commodity auction platform.

This invention discloses a lemon health analysis method, relating to the fields of intelligent agricultural monitoring and plant health diagnosis. It involves acquiring continuous images of lemon plants using a hyperspectral camera, and then using these images to detect changes in a dynamic differential spectral segmentation model. Sequence modeling is performed on the pixel-level multi-band spectral data. An adaptive motion-aware scheduler automatically suppresses loss when the proportion of motion change mask pixels exceeds a preset dizziness threshold to avoid background misjudgment, generating a pixel-level dynamic change map. This is compressed into a pathological dynamic state vector using a variational autoencoder, and end-to-end training is performed with action-conditional video prediction as the training objective. In the extrapolation phase, a recurrent dynamic model is used to perform multi-step imaginative prediction based on candidate action sequences, outputting the future evolution trajectory of the pathological dynamic state vector and generating a quantitative early warning or decision simulation report on the outbreak risk level. The system also includes change type classification and model closed-loop optimization functions.

The invention relates to a driving behavior primitive clustering method considering a data change trend, and the method comprises the following steps: 1, collecting natural driving data, and selecting a driving behavior analysis variable, 2, performing driving behavior primitive extraction on the # imgabs1 # by using a BMASS method to obtain a driving behavior primitive set P = {Pj; j = 1: N; 3, performing preliminary discretization on the driving behavior primitive data; 4, judging the data change trend of the driving behavior primitive according to the derivative and data fluctuation condition of the t point and the neighborhood of the t point in the Pj; 5, integrating the data distribution item and the data change trend item to obtain a discretization result of the driving behavior primitives, and clustering the driving behavior primitives based on an LDA topic model; according to the method, the clustering accuracy of the driving behavior primitives can be improved, accurate expression of semantic information of the driving behavior primitives is realized, and a driver is helped to accurately and intuitively understand the driving behavior of the driver.