84results about How to "Increase dependence" patented technology

The invention discloses a high-resolution remote sensing image change detection method based on multi-scale convolution decoding, electronic equipment and a storage medium, and belongs to the technical field of remote sensing image change detection. In order to meet the high-resolution remote sensing image change detection requirement, the method comprises the following steps: constructing a remote sensing image data set; a first stage and a second stage are formed based on a CNN module, a third stage and a fourth stage are formed based on a MambaVision module and a Transform module, and four dual-time-phase multi-scale feature maps are respectively obtained through processing of the four stages; constructing a time feature interaction module TFIM; a multi-scale convolution decoder EMCAD is constructed; and constructing a loss function, carrying out training, testing and evaluation by using the constructed remote sensing image data set, and applying the obtained trained optimal model to high-resolution remote sensing change detection. According to the invention, the detection capability of tiny changes in a high-resolution image is improved.

The invention discloses a multi-mode adaptive fusion navigation positioning method and system in a satellite navigation denial environment, the system adopts GNSS / IMU / visual sensors, and the method comprises the following steps: receiving data of the three sensors and completing modeling; initializing the three sensors; judging the strength of the GNSS signal by utilizing chi-square test; optimization variables are set, a residual matrix is constructed, and nonlinear optimization is carried out; setting a GNSS trust factor when the GNSS signal is weak or denial, and calculating a weight; and weighting and fusing sensor data, outputting optimal state estimation of the system, and realizing continuous navigation and positioning. According to the method, a self-adaptive weight adjustment mechanism is established, GNSS trust factors are introduced, and the weights of multi-sensor data are adjusted in a self-adaptive manner, so that the positioning difficulty under the conditions that the GNSS is subjected to deception interference, suppression interference, building blocking and the like can be overcome, and continuous navigation positioning of an unmanned system (such as a robot, an unmanned vehicle, an unmanned aerial vehicle, an unmanned ship, an unmanned underwater vehicle and the like) is realized.

The invention relates to the technical field of video compression, in particular to a video monitoring data compression and transmission method and system, and the method comprises the following steps: analyzing the texture features of a video frame, calculating the gray gradient of each pixel, obtaining the texture change rate, detecting edge information, calculating the distribution density of edge pixels, and extracting a color histogram. And calculating the dispersion degree of color distribution. In the invention, a dynamic coding parameter adjustment mechanism based on content is introduced to analyze the video frame, a compression strategy is adjusted according to the content complexity, and the coding process can accurately meet the requirements of different video contents through comprehensive analysis on the texture change rate, the edge density and the color dispersion degree; according to the scheme, the coding parameters can be dynamically adjusted by monitoring the coded video quality in real time and comparing the coded video quality with the preset quality threshold value, and the storage and transmission of data are further optimized while the video quality is ensured.

The invention discloses a mine slope risk prediction method based on multi-modal deep learning, and relates to the technical field of mine slope risk prediction, and the method comprises the steps: Stepl: inputting multi-modal data, Step2: preprocessing effectiveness data, Step3: constructing a sample set, and Step4: realizing mine slope stability prediction based on a gating cycle unit GRU of a mutual information mechanism. According to the invention, a multi-modal data integration GRU network is adopted to carry out sequence modeling on fused data, more complex sequence features are extracted, slope safety event prediction analysis is completed, high-precision monitoring of a mine slope is realized, and the accuracy and comprehensiveness of a prediction result are improved; according to the method, the mutual information mechanism is applied to the GRU network to maximize the mutual information between the sample and the feature, the dependency between the feature and the label is improved, the prediction capability is guaranteed, and the generalization capability on new data is also guaranteed.

The invention discloses a small sample image classification method and system based on multi-scale cross-modal prompt enhancement, and relates to the technical field of small sample image classification, and the method comprises the steps: obtaining a query image and a support image small sample data set with a labeled category; inputting the acquired data into a trained small sample image classification model, extracting visual features of a query image and enhanced visual features of each support image through a feature extractor based on hierarchical multi-modal prompt enhancement in the model, and performing global average pooling on all the extracted enhanced visual features to obtain an enhanced visual feature of each support image; obtaining an enhanced visual feature prototype of each category, calculating similarity scores of the categories according to the visual features of the query image and the enhanced visual feature prototype of each category, determining the category with the highest score as a classification result of the query image, and outputting the classification result; wherein the small sample image classification model is obtained by training a global-local self-guiding supervision mechanism. According to the invention, the precision and robustness of small sample image classification can be significantly improved.

The invention discloses a facial expression recognition method based on an intelligent accompanying robot, relates to the technical field of intelligent recognition, and aims to recognize mask-type expressions (namely expressions inconsistent with real emotions) by analyzing emotional contradiction degree Qmd, effectively avoid the problem of emotion misjudgment, and improve the recognition efficiency especially when facing complex or camouflage emotions. The robot can distinguish the real state of the emotion and adjust the feedback intensity according to the information, so that the emotion demand of the accompanying object can be accurately perceived and properly responded. According to the dynamic feedback adjustment mechanism, a feedback response intensity value Fqdz is constructed, the feedback intensity is dynamically adjusted according to the emotional contradiction degree, and the interaction intensity of the robot can be accurately adjusted according to emotional states in different situations. By combining initial setting and historical data, the intelligent accompanying robot can optimize the behavior mode of the intelligent accompanying robot according to long-term interaction feedback.

The invention relates to an image classification method and system based on a hierarchical competitive spiking neural network, belongs to the field of image classification, solves the problem of low recognition accuracy during image classification, and comprises the following steps: inputting preprocessed image data into a trained SNN network to obtain a prediction classification label of an image; wherein a competition strategy is introduced into the SNN, and the competition strategy comprises the following steps: sorting feature maps of different positions of input image data, and realizing excitation or suppression of the feature maps; and sorting different spiking neurons based on the sensitivity to realize excitation or suppression of the spiking neurons. The training of the SNN network comprises the following steps of: allocating a weight to each convolution kernel in the network, and allocating a weight to each neuron; calculating to obtain a training loss value; and carrying out back propagation, and calculating to obtain the weight of each convolution kernel, the weight of each neuron, an excitation coefficient and a suppression coefficient. The feature expression ability of the pulse neural network is enhanced, and the accuracy of image recognition is improved.

The invention discloses a medical risk assessment system and method based on multi-source data analysis, and relates to the technical field of intelligent medical analysis. The system acquires physiological data and psychological data of a patient, and performs correlation analysis on the physiological data and the psychological data to obtain an influence coefficient YX of a psychological state of the patient on physiological health; a patient comprehensive medical risk score CMRI (t) is obtained, a medical risk prediction model is constructed by using a random forest algorithm, the patient comprehensive medical risk score at a future time point is predicted in combination with historical medical data, and an early warning mechanism is implemented in combination with a set medical risk change threshold value JKFY, so that the health condition of the patient can be prevented from further deteriorating. According to the method, the hospitalization rate and the occurrence frequency of first-aid events are reduced, so that the medical cost and resource waste are reduced, the treatment compliance of the patient is improved, the specific health requirement of the patient can be effectively met, and the overall health result of the patient is improved.

The invention belongs to the technical field of information pushing, and discloses an information pushing method based on AIGC. The method comprises the following steps: constructing a knowledge graph generation model, a user portrait generation model, an information push content generation model and an information push strategy generation model; generating a knowledge graph by using a knowledge graph generation model according to the plurality of pieces of real-time knowledge data; according to the real-time user behavior data, user portrait generation and knowledge mapping are carried out by using a user portrait generation model; generating an information push strategy by using an information push strategy generation model according to the real-time user portrait after knowledge mapping; and according to the real-time information push strategy and the real-time user portrait after knowledge mapping, using an information push content generation model to generate information push content. According to the method and the device, the problems of insufficient individuation degree, single content generation, low pushing timeliness and poor user experience in the prior art are solved.

The invention discloses a driving strategy dynamic adjustment method and system based on driving style recognition. The method comprises the following steps: establishing a driving style classification model; monitoring the behavior data of the driver in real time, completing the confirmation of the driving style of the driver by using a data-driven adaptive algorithm, and dynamically updating the driving style classification model; on the basis of the driving style of the driver, specific behaviors of the driver in different situations are considered, and a scene-driven self-adaptive control strategy is applied to output a current driving strategy of the vehicle; and optimizing the current driving strategy of the vehicle through the interactive feedback of the driver in the driving process. The method can assist the automatic driving system in providing a driving experience mode more conforming to the personal preference of the driver, and improves the comfort and satisfaction of the driver in combination with an intelligent interaction and strategy closed-loop mechanism, and meets the individuation and dynamic adaptability of the automatic driving system.

The invention relates to the technical field of artificial intelligence, and discloses an intelligent interaction system and method based on legal consultation, and the system comprises a front-end user interface which supports text input, voice input and image input and has a multi-language switching function; the middle layer processing module comprises an emotion analysis module, an information integration module, a context management module and a response generation module and can dynamically adjust a processing strategy; and the rear-end legal knowledge base is used for storing laws and regulations, case analysis and common legal question answers and supporting real-time updating, version management and efficient retrieval. Through combination of a deep learning model and fine tuning of a special data set in the legal field, the system can accurately identify emotion features in user input, the accuracy not only improves the ability to understand legal terms, but also enhances the ability to process complex emotion expression, so that legal consultation is more humanized, and the efficiency is improved. The emotion change of the user can be responded in time, so that the satisfaction degree and the trust feeling of the user are improved.

The invention relates to the technical field of automatic driving testing, in particular to a generative adversarial network and method for generalization generation of an automatic driving testing scene, and the method comprises the steps: carrying out the standardization processing of actual driving data, embedding multi-source information, and combining with an improved condition generation and multi-generator architecture time sequence generative adversarial network (CMG-TimeGAN). According to the method, the system and the device, namely, the Conditional Multi-Generator Time GAN technology is adopted, a TCN module is used for replacing an RNN network in the whole Time GAN architecture, the defects of an existing Time GAN model in the aspects of time dependence capture, high-dimensional data processing, data quality sensitivity, mode collapse and the like are overcome, and multiple dangerous driving scenes are automatically generated.

The invention discloses an online learning system and method based on a self-adaptive engine, and belongs to the technical field of online learning. In order to solve the problems of insufficient learning demand identification and insufficient personalized recommendation, learning demands and weak links of a user can be accurately identified by monitoring the learning progress and behavior of the user in real time, a basis is provided for personalized recommendation, a personalized learning path is generated for the user according to a learning model and progress of the user, and the user experience is improved. The most relevant resources are obtained from the learning resource management module based on the learning features and preferences of the user, optimal distribution and utilization of the learning resources are achieved, personalized learning content and teaching modes are pushed in real time according to the learning requirements of the user, it is ensured that the user can obtain the needed learning resources in time, the learning efficiency is improved, and the user experience is improved. By collecting and analyzing the feedback data of the user to the pushed content, the pushed content and the teaching mode can be dynamically adjusted, and it is ensured that the pushed content is matched with the learning progress and interest of the user.

The invention is suitable for the technical field of machine learning, and provides a human physiology and motion state anomaly detection method and device. The human physiology and motion state anomaly detection method comprises the steps of obtaining real-time sign data collected by the sign collector and standard sign data which are obtained through pre-training and respectively correspond to multiple preset states; performing semantic classification on the real-time physical sign data to obtain real-time semantic information; each piece of real-time semantic information corresponds to one or more preset states; and according to standard physical sign data corresponding to the real-time semantic information, determining whether the real-time physical sign data is in an abnormal state or not. According to the technical scheme, by introducing the anomaly detection method based on semantic classification, the accuracy of physical sign data analysis in the wearable device is remarkably improved, and the real-time performance and the intelligent level of the device are enhanced.

The invention relates to the technical field of medical services, in particular to an integrated intelligent medical service system, method and equipment. The system comprises a doctor-seeing appointment service module, a hospital guide service module, a post-diagnosis service module and a service rating module. The doctor-seeing appointment service module is used for generating doctor-seeing appointment suggestions according to doctor-seeing demands input by the patient and receiving doctor-seeing appointment information selected by the patient from the doctor-seeing appointment suggestions; the hospital guide service module is used for carrying out hospital guide on the patient based on the treatment appointment information in the treatment process of the patient and recording treatment data in the treatment process of the patient; the after-diagnosis service module is used for generating health suggestions based on the treatment data and recording feedback information of the patient for the health suggestions; and the service rating module is used for generating a rating report of the doctor seeing service for the patient based on the doctor seeing reservation information, the doctor seeing data and the feedback information, and generating an improvement strategy based on the rating report. The service quality and efficiency can be improved.

The invention relates to the technical field of crop yield prediction, and discloses a crop yield prediction method based on LSTM and GRU, and the method comprises the steps: obtaining crop planting data; the crop planting data comprises climate parameters and soil parameters; constructing a crop yield prediction model based on LSTM and GRU; extracting a first time sequence feature from the crop planting data by using an LSTM network; extracting a second time sequence feature from the crop planting data by using the first GRU network; performing feature fusion on the first time sequence feature and the second time sequence feature by using a feature fusion layer in a weighted average mode; using a second GRU network to extract tense features from the fusion features; and performing linear transformation on the tense characteristics by using a full connection layer to obtain a final crop yield prediction result. According to the method, the multi-stage time features can be extracted and processed more accurately by adopting a double-circulation structure, so that the accuracy of crop yield prediction is improved.

The invention relates to the technical field of seismic data processing, in particular to a seismic waveform inversion method and system based on deep learning image registration, and the method comprises the steps: carrying out the least square migration based on small offset observation data and a background velocity field, and constructing a reflection coefficient model; performing Born forward modeling according to the reflection coefficient and the background velocity field to obtain a simulated seismic record image; acquiring a seismic record deformation field by adopting deep learning image registration based on the simulated seismic record image and the seismic observation image; calculating a normalized deformation field, and judging whether the normalized deformation field meets a preset condition or not; if yes, terminating the iteration process; otherwise, executing the next step; and according to the seismic record deformation field and the accompanying seismic source, updating the background velocity field through a conjugate gradient algorithm, and repeating the steps. According to the method, high efficiency and high precision of deep learning are utilized, and the problem that a cycle jump problem which is prone to occurring in waveform matching falls into a local minimum value is avoided.

The invention belongs to the technical field of remote sensing image processing, and particularly relates to an open vocabulary and remote sensing image interaction-oriented multi-modal semantic segmentation method. According to the method, a double-branch text encoder is introduced, and scene-level semantics are extracted from various text data to realize dynamic cross-modal fusion. The text derivative features dynamically interact with visual embedding through a proposed text guide visual semantic fusion module, and domain perception feature refinement and human interpretable decisions are achieved. Besides, text integration opens up a way for opening word semantic segmentation, so that the system can recognize and classify invisible categories through natural language description, and strict constraints of a predefined category classification method are overcome.

The invention belongs to the technical field of image target recognition, and particularly relates to an anti-interference target recognition method combining a pulse neural network and a capsule network, a program, equipment and a storage medium. A capsule structure is introduced into a pulse brain-like model, a self-attention mechanism is introduced, the protocol degree between capsules can be learned in single forward propagation, a capsule attention map is generated through a self-attention layer, and then the capsule attention map is mapped to a classification layer through a one-dimensional convolution layer, so that the model structure and the calculation complexity are simplified, and the accuracy of the model is improved. And irrelevant capsule voting can be selected and inhibited by utilizing a self-attention mechanism, so that information useful for entity recognition is reserved, and the capabilities of long-distance dependence and feature selection are improved. The method has good robustness under the conditions of different noises and different intensities, has good generalization and invariance for affine transformation, and improves the anti-interference capability of target identification.

The invention discloses a color fixative for colorful potatoes and a color protection method for the colorful potatoes, and belongs to the technical field of deep processing of agricultural products. The color fixative used in the invention is composed of citric acid, ascorbic acid, L-cysteine, gallic acid and water, and the mass ratio of the citric acid to the ascorbic acid to the L-cysteine to the gallic acid to the water is (0.4-0.6): (0.4-0.6): (0.2-0.3): (0.6-1): 100. The preparation method comprises the following steps: peeling fresh flower-color potatoes, dicing, soaking in a color fixative for 40 minutes, pre-drying diced potatoes, freezing and thawing, drying with hot air, crushing and sieving to obtain raw whole flour; through the color protection treatment, enzymatic browning of the flower-color potatoes can be reduced, meanwhile, the color of the potatoes is reserved, and the stability of anthocyanin is improved.

The invention relates to a turnout detection method based on computer vision and an intelligent monitoring pile. In order to improve the detection precision, an improved YOLOv8 model is provided, an original path aggregation feature pyramid network structure is jointly improved by using a space-depth convolution module (SDC) and a feature enhancement-fusion module (FEF) in a check layer of the network, and a small target detection enhancement pyramid structure is provided to enhance the small target detection capability. In addition, an improved loss function with a scaling factor and an auxiliary bounding box is used to further enhance the detectability of the model. Compared with the prior art, the method has higher accuracy and robustness.

The invention discloses a degradable polycarbonate cationic carrier and application thereof, and belongs to the technical field of drugs, the degradable polycarbonate cationic carrier is a nano-micelle formed by self-assembly of diblock polymers, the diblock polymers are represented as A-B, the A block is mPEG, and the B block is PAMA, PDMA or PDEA; the molecular weight of the block A is 2000 to 5000, and the polymerization degree of the block B is 50 to 150. The degradable polycarbonate cation carrier has an mRNA delivery function, and is combined with mRNA through a positive and negative electricity interaction principle, so that the problem of gene waste caused by mRNA loading through a microfluidic method can be solved, and efficient delivery of antigen presenting cells (APC) is realized.

The invention discloses an ultra-black coating with wide-angle absorption as well as a preparation method and application of the ultra-black coating, and belongs to the technical field of ultra-black coatings and preparation of the ultra-black coatings. The problem that an existing ultra-black coating cannot present extremely low scattering and high absorptivity under large-angle incident light is solved. According to the method, carbon black is used as an absorbent, a high-temperature-resistant silica sol solution is used as a binder, a stable structure is formed by adopting a one-step sintering method, and meanwhile, an ultra-black coating with a forest-shaped light trap array structure is constructed. The refractive index of the surface of the coating is obviously changed by utilizing the forest-shaped array structure on the surface of the ultra-black coating, so that the optical reflection of the surface is reduced; the surface forest-like light trap structures are closely arranged, so that light incident from each angle can be reflected back and forth in the structure, the incident light is absorbed to the greatest extent, stable high light absorption at a wide angle is realized, the reflectivity of the ultra-black coating does not change obviously along with the angle of the incident light, and the deep space detection precision of a space optical system is improved.

The invention belongs to the technical field of recommendation systems, and discloses a user cold start recommendation method based on meta-learning, and the recommendation technology comprises the detailed steps: step 1, user behavior log collection: collecting behavior data of a user on each platform, including behaviors of preference, resource browsing, clicking, searching and sharing; step 2, user feature preprocessing: key user features are selected based on feature distribution and significance indexes; and step 3, constructing a double-tower multi-target recall model: constructing the double-tower multi-target recall model based on the characteristics of the users and the articles. Through deep learning of consumption interest distribution of old users and correlation of articles, a more detailed and accurate new user portrait can be constructed, so that interests and preferences of the users can be matched more accurately; on the basis of accurate construction of the new user portrait, the recommendation system can formulate a personalized recommendation strategy, a one-by-one recommendation mode is avoided, and recommendation content better meeting user requirements is provided.

This invention relates to the field of multimodal sentiment analysis technology, and in particular to a method and system for multimodal sentiment analysis based on sequential translation under uncertain and missing modalities. The method includes: acquiring multimodal sentiment data, including visual, textual, and auditory modalities; constructing a multimodal sentiment analysis model based on sequential translation, firstly by extracting self-attention encoding features of each modality using a Transformer encoder; and then designing a bidirectional sequential translation architecture centered on the textual modality. This invention improves the utilization of textual modal features in the feature fusion process, thereby enhancing the quality of multimodal fusion features and providing strong support for subsequent complex sentiment analysis.

The embodiments of the present application disclose a model training and image processing method, apparatus, device, storage medium and computer program product, wherein the method comprises: determining a first scene image and a second scene image based on a first image sample; there is an overlapping area between the first scene image and the second scene image; using a first model to be trained, respectively extracting features from the first scene image and the second scene image to obtain a first scene feature of the first scene image and a second scene feature of the second scene image, and using a second model, respectively extracting features from the first scene image and the second scene image to obtain a third scene feature of the first scene image and a fourth scene feature of the second scene image; determining a target loss value based on the first scene feature and the second scene feature, the third scene feature and the fourth scene feature; based on the target loss value, updating the model parameters of the first model at least once to obtain the trained first model.

The present invention discloses a facial expression recognition method based on an intelligent companion robot, which relates to the field of intelligent recognition technology. The present invention can identify masked expressions (i.e., expressions inconsistent with real emotions) by analyzing the degree of emotional contradiction Qmd, effectively avoiding the problem of emotional misjudgment. In particular, when facing complex or disguised emotions, the robot can identify the true state of the emotion and adjust its feedback intensity based on this information to ensure that the emotional needs of the companion object can be accurately perceived and appropriately responded to. Dynamic feedback adjustment mechanism: By constructing a feedback response intensity value Fqdz and dynamically adjusting the feedback intensity according to the degree of emotional contradiction, the present invention can accurately adjust the robot's interaction intensity according to the emotional state in different situations. By combining initial settings and historical data, the present invention enables the intelligent companion robot to optimize its own behavior pattern based on long-term interactive feedback.

The utility model relates to clothing machinery, in particular to a collar rib folding, sewing, stacking and fixing all-in-one machine which comprises a machine table. The machine table is provided with a material folding unit capable of folding a neckline rib fabric piece on the table top in the length direction and a pressing and conveying unit capable of pressing and holding the folded double-layer neckline rib fabric piece and conveying the double-layer neckline rib fabric piece to a first sewing machine to conduct splicing sewing on the open end of the double-layer neckline rib fabric piece. The seam allowance turning unit is used for turning over an upper layer seam allowance of the spliced and sewn ring-shaped collar rib fabric piece by 180 degrees; the folded end of the seam allowance of the ring-shaped collar rib fabric piece is folded in the width direction, the folded ring-shaped collar rib fabric piece is pressed and held and conveyed to a second sewing machine, and the folded ring-shaped collar rib fabric piece is subjected to reinforcing sewing; and the laminated pressure feeding unit is used for laminating. According to the utility model, after the splicing process is completed, folding and reinforcing sewing in the width direction of the neckline rib can be automatically carried out, so that the automation degree of the whole processing process is improved.

This invention belongs to the field of human-computer interaction technology, specifically relating to a microphone-based human-computer interaction method, device, medium, and system. The method includes: acquiring audio signals collected by at least one microphone on an electronic device; performing transient pulse detection on the audio signals, identifying short-time sound pulse signals, and calculating the pulse signal-to-noise ratio (SNR); performing spectral feature analysis on the audio signals, detecting the spectral changes of ambient sound signals before and after pressing; dynamically adjusting the comprehensive confidence weighted fusion weights based on the pulse SNR to determine whether a valid event has occurred; and generating corresponding interactive control commands if a valid event has occurred. Compared with existing technologies, this invention solves the problems of increased production costs due to added physical structures and incompatibility with existing production lines, as well as insufficient detection accuracy and susceptibility to false recognition when relying solely on microphones. This solution achieves microphone-press interaction without relying on additional hardware and exhibits high recognition rate and low false alarm rate under various environmental noise conditions.