47results about How to "Enhanced Feature Representation" patented technology

The invention relates to an Internet of Vehicles multi-agent cooperative sensing method and system. The method comprises the following steps: constructing a collaborative sensing network, wherein the collaborative sensing network comprises a self-agent and a plurality of collaborative agents; acquiring and processing sensing data through the collaborative sensing network; performing feature extraction to obtain intermediate features; self-adaptive sparsification is carried out to obtain sparse features, and the sparse features are compressed and transmitted to a self-agent; performing time sequence feature enhancement on the features of all the agents at the self-agent end; fusing the features to obtain fused features; and constructing an Internet of Vehicles perception model, and realizing perception by the detection model according to the fused features. According to the method, the calculation complexity of traditional global attention is reduced from the square level to the linear level through an adaptive sparsification mechanism, the calculation overhead is remarkably reduced while the multi-agent feature interaction precision is kept, and the method is more suitable for real-time operation on the vehicle-mounted edge equipment with limited resources.

The invention discloses a molten steel element content on-line detection system based on a laser-induced breakdown spectroscopy technology, and relates to the technical field of molten steel element content on-line detection, the molten steel element content on-line detection system comprises an excitation system, an acquisition system, a data preprocessing module and a component detection module; the excitation system is used for emitting laser to the molten steel surface to generate plasma; the acquisition system is used for collecting signal light radiated by the plasma; the data preprocessing module is used for carrying out baseline correction and noise reduction processing on the collected original spectral signals; and the component detection module comprises an element content prediction model based on a gating specific expert attention network LSEA-Net and is used for carrying out quantitative analysis on the preprocessed spectral data to obtain the element content of the molten steel. The technology has the advantages that sample preparation is not needed, online rapid analysis and multi-element synchronous detection are achieved, and the technology is suitable for the high-temperature environment and the like, and the requirement of a steelmaking site for real-time component detection is met very well.

The invention discloses a remote sensing image segmentation method based on multi-scale wavelet transform and Mama. The method comprises the following four steps: firstly, carrying out preprocessing and division on an ISPRS Potsdam data set and a Vaihingen data set; then constructing a segmentation network, wherein the network comprises a local detail extraction branch, a spatial semantic extraction branch, a cross-branch feature fusion part and a decoder; training and optimizing the network by using the training set; and finally, performing segmentation reasoning on a test image by using the trained model. According to the method, the high-frequency detail extraction capability of the image is enhanced through multi-scale wavelet transform, the long-distance dependency relationship is modeled by using the visual state space block, and effective fusion of the features is realized through the double-branch fusion module, so that the feature extraction capability and the semantic segmentation precision are improved, and meanwhile, the training efficiency and the stability are optimized.

This disclosure provides a cross-channel distributed video encoding and decoding method and system based on multidimensional attention, which can be applied to the field of video encoding and decoding technology. The method includes: acquiring a group of video images to be transmitted, the group of video images including a first key reference frame, a second key reference frame, and at least one intermediate video frame; encoding the first key reference frame and the second key reference frame into first encoded data and second encoded data, respectively; encoding each intermediate video frame into third encoded data based on a frame encoder; extracting multi-scale features from each third encoded data based on a multidimensional attention mechanism; processing the multi-scale features into fourth encoded data based on a multi-channel feature extraction mechanism; processing at least one fourth encoded data into a first bitstream; and converting the first encoded data and the second encoded data into a second bitstream, so as to transmit the video image group to the decoder via the first bitstream and the second bitstream.

The application relates to the field of sanitation operation scheduling, in particular to a sanitation operation scheduling method based on the Internet of Things, which comprises the following steps: determining an iterative stop condition of a load CEEMDAN model based on time local weighted variance of load data, so as to generate vehicle load characteristics; generating vehicle oil consumption characteristics through an oil consumption CEEMDAN model based on statistical values of load data residuals and oil consumption data residuals, so as to calculate oil consumption channel residual signals; generating comprehensive vehicle carrying characteristics through a feature extraction model based on a multilayer perception machine, a gating mechanism and a cross-modal attention mechanism; and generating corrected vehicle garbage load through a long short-term memory network and a fully connected layer architecture. The application realizes efficient noise reduction and accurate decomposition of non-stationary time series data of oil consumption and load, realizes accurate processing of Internet of Things data of garbage transfer vehicles, improves the operation efficiency of the garbage transfer vehicles and reduces the operation cost.

The invention provides a remote sensing lightweight target detection method based on a local state modeling network, which is called as LSM Net. The method comprises the following steps: preprocessing a remote sensing image data set; then constructing an LSM Net which takes the LSM-TD as a backbone network and takes the feature pyramid as a neck network, and taking the preprocessed image as network input; in the training process, a lightweight local state modeling module (LSM Block) is combined to optimize the network, and fine feature extraction and long-range spatial dependence modeling of a tiny target are realized. And finally, inputting a to-be-detected remote sensing image into the trained LSM Net network to obtain a detection frame containing a tiny target category and the confidence coefficient. According to the method, through lightweight network design and multi-dimensional feature optimization, the complex scene adaptability and operation efficiency are remarkably improved while the detection precision is guaranteed, the method is suitable for intelligent monitoring and rapid identification of multi-field tiny targets in remote sensing images, and an efficient and feasible technical scheme is provided for resource-limited edge equipment.

The application discloses a spine-pelvis joint segmentation method based on frequency-space cooperation and adaptive fusion, and belongs to the technical field of medical image processing. A joint segmentation dataset containing spine and pelvis structures is constructed, and standardization preprocessing is completed by supplementing pelvis annotation to public data and combining private clinical data; a multi-scale three-dimensional segmentation network is built, a frequency-space cooperation feature modeling HFSM module is introduced in the coding stage, local details and global semantic features are extracted through joint extraction in the spatial domain and the frequency domain; a selective cross adaptive fusion SCAF module is connected between the encoder and the decoder, boundary enhancement, structure perception and gate weight modulation are performed on different levels of coding features and decoding features, adaptive feature aggregation is realized, the network adopts an end-to-end training mode, a weighted combination loss function is used to optimize model parameters, and the joint segmentation result of the spine and the pelvis is output. The application can effectively improve the segmentation accuracy and integrity of the connection region, complex edge and small structure of the spine-pelvis.

This invention provides a hyperspectral classification method based on band erasure and contrastive learning. First, hyperspectral data is acquired and band erasure is performed, resulting in two band-removed hyperspectral images. These images are then preprocessed to obtain sample patches for two branches of a contrastive learning network. A gradient mask is used to cover all patches in the upper branch, while leaving the patches in the lower branch untouched. Random occlusion is applied to the patches in both branches, followed by data augmentation to obtain positive sample pairs. Hyperparameters are set, and the positive sample pairs are input into the contrastive learning network. After training, the network parameters and features are saved. The extracted features and annotations are used as the training set to train a classifier, achieving hyperspectral classification. This invention combines various data augmentation methods suitable for hyperspectral data with contrastive learning, tapping into the potential of contrastive learning in the hyperspectral field and improving classification accuracy to a new level, effectively achieving the classification of spectral images.

This invention provides an enzyme turnover rate prediction method based on a dual-route hybrid expert mechanism, belonging to the field of bioinformatics technology. It solves the problems of low data utilization, insufficient information mining, and poor robustness in existing prediction methods due to missing values ​​and temperatures. The technical solution includes the following steps: S1: Constructing a unified standard enzyme turnover rate dataset; S2: Extracting multimodal embedding features; S3: Constructing intra- / inter-modal encoders; S4: Combining modalities into an expert hybrid module; S5: Designing an attention fusion mechanism. This invention can achieve high-precision prediction under complex in vitro environmental conditions.

The application discloses a kind of light multi-scale image defogging methods based on attention mechanism, the method is first to the feature extraction of fog image;Then the feature map obtained is input MDSCA- NET network, and the fog-free image is reconstructed;The MDSCA- NET network includes several image defogging models, model fusion module;The image defogging model includes feature attention module, fog-free map generation module;The feature attention module is used to extract the pixel feature in feature map;The fog-free map generation module is used to generate initial defogging image for pixel feature map;The model fusion module is used to use alternating direction multiplier optimization algorithm to carry out model fusion to several image defogging models, and output final defogging image.Experiments show that the MDSCA- NET network proposed in the application obtains significant results under the comprehensive measurement of model performance and parameter quantity.

The invention provides an unplanned return ICU risk prediction method, system and device, and the method comprises the steps: fusing an MIMIC-IV public data set and local private data, and carrying out the standardization preprocessing of multi-source medical data through a rule module and a clinical knowledge base; constructing a partitioned feature system comprising basic features, laboratory indexes, hemodynamic parameters and respiratory metabolism dimensions, and optimizing feature expression by adopting a sliding window and a feature interaction method; performing medical field fine tuning based on a large language model, and converting the structured features into natural language description in combination with cue words; carrying out model training by adopting optimization strategies such as label smoothing and gradient cutting, and evaluating model performance in a mode of combining internal verification and external verification; and realizing interpretability analysis of model prediction by using an SHAP framework. According to the scheme, the accuracy and generalization ability of ICU return risk prediction and the applicability of an actual scene are remarkably improved.

The application provides a PCB segmentation method and system based on X-ray and three-dimensional point cloud, the application extracts two-dimensional perspective features of X-ray by using a residual network combined with a feature pyramid network, and extracts three-dimensional space geometric features of three-dimensional point cloud by using RandLA-Net. Two-dimensional projection of three-dimensional point cloud is completed through a projection matrix, and parameters such as ray propagation direction, path space density and material absorption intensity are calculated to construct a physical information field; two-dimensional local features are obtained by adaptive deformation sampling according to three-dimensional geometric feature prediction projection offset; dynamic perspective confidence weight is constructed by combining occlusion rate, penetration length and the like, and multi-perspective fusion two-dimensional features are obtained by weighted aggregation. Cross-modal adaptive fusion of two-dimensional and three-dimensional features is realized through a gating mechanism, feature restoration and resolution reconstruction are completed through a point cloud decoding network, and a PCB point-level structure segmentation result is output. The application can accurately segment structures such as copper wires, copper columns and dielectric layers in the PCB.

The invention discloses a frequency domain enhanced single-pixel imaging method and a frequency domain enhanced single-pixel imaging system, which are used for acquiring a single-pixel measurement value corresponding to a target object and performing image reconstruction based on the single-pixel measurement value to obtain a target imaging result. The method comprises the following steps that a single-pixel imaging network is constructed, the single-pixel imaging network comprises an input layer, a plurality of coding layers, a plurality of decoding layers and an output layer which are connected in sequence, the number of the coding layers is the same as that of the decoding layers, and each coding layer comprises a basic convolution layer, a Swindow-Transform module and a frequency domain MLP module which are connected in sequence; based on the single-pixel measurement value, performing iterative training on the single-pixel imaging network according to a double-domain self-supervision method, and obtaining a first reconstructed image and a second reconstructed image in each iteration; and taking the optimal first reconstructed image as a target imaging result. The method can give consideration to both imaging quality and reconstruction speed.

This invention discloses a data compliance dynamic detection method and system based on deep learning, belonging to the field of data security processing technology. It addresses the problem that existing methods, which focus primarily on text data processing when performing compliance detection based on extracted comprehensive semantic words, lack the ability to comprehensively process multimodal data and comprehensively evaluate complex compliance issues, leading to inaccurate compliance assessments. The method includes preprocessing the original data stream, extracting features from the preprocessed data stream based on data stream type, and using a compliance analysis model to identify and analyze multimodal feature sets based on a standard rule base. In this invention, preprocessing the original data stream filters and reduces noise in the multimodal data, ensuring the authenticity and validity of the data. Furthermore, the compliance analysis model's identification and analysis of multimodal feature sets ensures comprehensive processing capabilities for multimodal data and comprehensive evaluation capabilities for complex compliance issues.

The application relates to the technical field of intelligent traffic monitoring, and discloses a road surface water detection and identification method based on multi-view feature fusion. The method comprises the following steps: collecting image data, depth information and environmental parameters by sensor nodes arranged at multiple positions on a road surface to form an original monitoring data set; performing multi-source feature extraction and fusion processing on the data set to obtain a waterlogging feature graph set; performing spatial domain analysis on the graph set to generate a road surface partition consistency graph; extracting time dimension data based on the graph to perform dynamic change evaluation and output a waterlogging evolution report; detecting abnormal patterns from the report and identifying abnormal waterlogging areas; and calculating risk indexes according to the abnormal areas to generate a final road surface water risk graph. Through multi-source data fusion and spatio-temporal joint analysis, the method realizes accurate detection, dynamic evolution tracking and risk evaluation of road surface waterlogging, and significantly improves the accuracy and early warning capability of urban road waterlogging monitoring.

The application relates to the technical field of computer vision and deep learning, in particular to a multi-scale edge information multi-target detection method fusing text, which comprises the following steps: constructing a basic YOLO-World model comprising a backbone network, a neck network and a head network, and optimizing the structure of the basic YOLO-World model, including: replacing an original C2f module with an MSES module in the backbone network; introducing an AFCAtion attention module at the end of the backbone network; improving the feature fusion mode of an original feature pyramid network by adopting an SSFF module in the neck network; training the optimized model to obtain a target detection model; using the target detection model to perform target detection on an input image and outputting a target detection result; and solving the problems that an existing text fusion method is difficult to efficiently process the semantic difference between an image and text when facing a complex background, and the combination of an image and text is still insufficient for understanding of part features.

The application discloses a kind of power equipment defect identification and warning method and system based on deep learning.The method comprises: synchronously collecting and registering the visible light and infrared thermal imaging image on the surface of power equipment, constructs the instance segmentation network including light weight feature extraction network, multiscale feature fusion network and frequency domain mask prediction branch;Adopt the generative adversarial strategy to enhance the diversity of training sample;Based on graph neural network, analyze the association between defect and equipment topology, historical record, infer the cause-effect relationship of defect and risk level;Generate the interpretable warning information including heat map, natural language report and repair suggestion;Real-time detection and deep analysis are realized using end-edge-cloud collaborative architecture;Through closed-loop optimization mechanism, continuously improve system performance.The application realizes high-precision defect detection under multi-modal data fusion, has strong robustness and interpretability, significantly improves the intelligent level of power equipment operation and maintenance.

The invention discloses a point cloud individual tree instance segmentation method and a point cloud individual tree instance segmentation system based on soft clustering and Transform. According to the method, the problem that the calculation amount of large-scale forest point cloud data is too large is effectively solved; a soft clustering module is adopted to calculate a soft attribution weight through a probabilistic attribution mechanism, so that the feature characterization capability of a boundary region is enhanced, and the defect that instance boundary distortion is easily caused in boundary fuzzy regions such as branch intersection and shielding in a traditional hard clustering method is overcome; according to the method, the problems of excessive combination of instances, fuzzy boundary, high calculation cost and the like of a traditional method are avoided, the precision, robustness and multi-scene adaptation capability of individual tree instance segmentation are remarkably improved, and the actual requirements of precise forestry and ecological monitoring in a complex forest scene can be met.

This invention discloses a land use classification method based on intelligent fusion of multi-source data. It acquires multi-source geospatial data of the study area, including remote sensing imagery, street view imagery, building vector data, road network vector data, points of interest (POIs), mobile phone signaling data, and Weibo check-in data, forming a land use classification experimental dataset. It constructs remote sensing imagery features, street view imagery features, built environment features, POIs, and population distribution features for each study unit, forming a feature set for the land use classification task. The constructed features are paired to generate feature pairs composed of different modalities. A contrastive learning method is used to align the features of different modalities, and the aligned features are added to the feature set for the land use classification task. Land use classification is performed based on an XGBoost ensemble learning model, and five-fold cross-validation is used to verify the model's effectiveness. This invention improves the accuracy of land use classification.

This invention relates to the field of remote sensing image processing technology, and particularly to a hyperspectral band selection method based on multi-feature enhancement. The method involves spatial max pooling of the original hyperspectral data to obtain information representation data for the corresponding bands; filtering the original hyperspectral data using a mask constructed from the information representation data set to obtain a preliminary hyperspectral data set; dividing the preliminary hyperspectral data set and extracting multi-dimensional features from the preliminary hyperspectral data of each band; calculating the relative proximity of the corresponding bands based on the multi-dimensional features of each preliminary hyperspectral data set, and constructing an initial candidate band set by selecting the bands with the highest relative proximity; filtering the bands in the initial candidate band set; and finding the optimal band set from the resulting simplified band set. This invention significantly improves the feature representation capability of band subsets, achieving a significant improvement in computational efficiency while maintaining high classification accuracy.

The application discloses a self-supervised depth estimation method in mine safety hidden danger scene three-dimensional reconstruction, first, the depth estimation network and the attitude estimation network model of normal illumination image and low illumination image are constructed respectively, the position perception module of self-attention mechanism is used in the middle of the encoder decoder, to obtain the context information of the scene structure and better feature representation; in the process of network training, normal illumination image and low illumination image obtained through CycleGAN are used for training, and then the image output by CycleGAN is processed by using a mapping image enhancement (MIE) algorithm, so that the need of keeping the brightness consistency is met, and the influence caused by low illumination and uneven illumination is solved. The feature representation at the detail is enhanced, and the depth estimation effect on the complex background is strengthened. The added mapping image enhancement module makes the brightness and contrast of the low illumination image obviously improved, so that higher visibility is brought to the low illumination image, and more details are reserved.

The application discloses a medical image segmentation method combining selective edge aggregation and deep neural networks, first constructs a Transformer-based encoder, and replaces MSA and MLP in a standard Transformer block with a selective edge aggregation module and a densely connected feedforward network to realize feature fusion and complementation; then constructs an encoder and a decoder based on densely connected CNN, connects the two encoders in parallel, enables the network to interact information at multiple levels, and fuses multi-scale features from the double encoders and the low-to-high up-sampling path based on the decoder of the densely connected CNN to restore the spatial resolution of the feature map in a fine-grained and deep-level manner; finally, a loss function combining target edges and regions is designed to simultaneously optimize the encoder and the decoder with a multi-level optimization strategy, so that the network further learns more semantic information and boundary details to refine the segmentation result. The application can solve the medical image segmentation problem in a real scene.

The invention discloses an electric vehicle cluster flexibility resource probabilistic prediction method and system fusing multi-source spatio-temporal data, and relates to the technical field of intelligent power grid dispatching and artificial intelligence. Comprising the steps of obtaining multi-source heterogeneous spatio-temporal data of an electric vehicle cluster and performing preprocessing; dividing the predicted target area into space-time grids and generating a space-time grid index; a unified space-time-feature tensor is constructed; constructing an electric vehicle cluster flexibility resource probabilistic prediction model based on the probabilistic deep space-time network; training the model by using historical data of an electric vehicle cluster; inputting the space-time-feature tensor into the trained model, and generating and outputting a prediction result; and continuously performing prediction, and when the data is abnormal, triggering lightweight increment updating of the model parameters. According to the method, multi-source data can be comprehensively utilized, the uncertainty of probabilistic prediction of the flexibility resources of the electric vehicle cluster is quantified, the spatial-temporal dynamic characteristics are accurately described, and a reliable basis is provided for accurate dispatching of a power grid.

The application discloses a speech emotion recognition method based on a multi-scale feature extraction of an attention mechanism, and comprises the following steps: constructing a training data set; constructing a speech emotion recognition network model, wherein the speech emotion recognition network model comprises a multi-scale feature extractor module, a multi-scale feature encoder module, a feature fusion module and a speech emotion recognition classifier, the multi-scale feature extractor module is used to obtain multiple speech features of different scales, the multi-scale feature encoder module is used to encode the speech features to obtain speech features of different scales after encoding, the feature fusion module is used to obtain multi-scale speech fusion features, and the speech emotion recognition classifier is used to obtain a final classification result by using the multi-scale speech fusion features; training the speech emotion recognition network model; and obtaining an emotion recognition result of a to-be-recognized speech. The multi-scale feature extractor is used to learn features of speech data under different receptive fields as much as possible, so that the feature diversity is ensured.

The invention provides a connector pin defect intelligent detection method and system based on machine vision, and relates to the technical field of machine vision, and the method comprises the steps: obtaining a multi-view image under a multi-light-source illumination condition, calculating three-dimensional shape data, extracting a pin region, building a correlation edge, and combining the characteristics of a surface normal vector, a boundary contour and the like to obtain a pin defect detection result; and optimizing node features based on a graph neural network, carrying out pin instance extraction and defect classification, and iteratively updating a defect label through a defect propagation probability. According to the invention, various defects of the connector pins can be accurately identified, the detection precision and efficiency are improved, and the omission ratio and the false detection ratio are reduced.

The invention relates to the technical field of intelligent medical treatment, and discloses a lumbar endoscopic postoperative chronic neuropathic pain early warning deep learning large model system. According to the system, a postoperative multi-modal data initial pool of a patient is constructed, and original data streams are collected; performing space-time alignment and missing value interpolation on the original data stream by using the dynamic feature extraction network to form a standardized time sequence feature matrix; inputting the feature matrix into a hierarchical attention feature fusion network, calculating interaction weights of different modal features on a plurality of time scales, and generating enhanced feature representation fused with multi-scale context information; deducing an evolution path of a neural function state under different physiological parameter combinations through a risk trajectory simulation engine based on enhanced feature representation, and constructing an individualized risk evolution trajectory map; and according to the map, adopting a self-adaptive threshold decision model to identify an abnormal fluctuation interval deviating from a normal recovery mode, and outputting a chronic neuropathic pain risk early warning signal.

The application discloses a workshop production quality online detection system based on machine vision and AI, and particularly relates to the field of machine vision detection, and comprises the following steps: through a multi-modal data acquisition and preprocessing module, physical signals, process parameters and actual quality values of products are synchronously acquired, and after data cleaning, semantic alignment and standardization processing, structured data blocks of associated product information are generated; then, through a feature extraction and mixed vector generation module, physical information features and multi-modal data driven features are extracted and fused into a mixed vector, a network is trained to output a comprehensive quality index and a corresponding time sequence; finally, through a quality field energy coupling decision module, a quality toughness coefficient and a process fluctuation conduction coefficient are calculated, three types of energy fields, i.e., compliance, consistency and risk, are defined, a comprehensive decision value is obtained through coupling, four-level quality decisions, i.e., high-quality, qualified, to-be-checked and rejected, are output according to the value and field characteristics, production circulation is guided, and the accuracy and efficiency of quality detection are effectively improved.

The invention discloses a lake water quality parameter inversion method based on a remote sensing image, a medium, equipment and a product, and relates to the field of water quality parameter inversion, and the method comprises the steps: obtaining a Sentinel-2 remote sensing image and water quality sampling data, re-sampling a wave band in the remote sensing image to a uniform resolution, and carrying out the time-space matching of the wave band and the water quality sampling data, selecting a dual-band combination, a three-band combination and a four-band combination related to the water quality sampling data from the matched resampling bands, and constructing a training set; a deep learning inversion model is constructed, the deep learning inversion model comprises three feature extraction sub-networks, an attention mechanism and a fusion main network, wave band combinations are input into the three feature extraction sub-networks respectively, obtained three feature vectors are subjected to weighted fusion through the attention mechanism and then input into the fusion main network, and predicted water quality parameter concentration is output; and training the model by using the training set, and performing lake water quality parameter inversion by using the trained model. According to the invention, the inversion precision of water quality parameters is improved.

The application provides a multi-modal feature extraction network training method and a three-dimensional feature representation method, including: obtaining a multi-modal training data set; wherein the multi-modal training data set includes a three-dimensional model, a multi-angle rendering image set corresponding to the three-dimensional model, and a category text tree; extracting a first point cloud feature vector of the three-dimensional model, a first image feature vector of the multi-angle rendering image set, and a first text feature vector of the category text tree through an initial feature extraction network; determining a joint modal feature vector based on the first image feature vector and the first text feature vector through a cross-modal joint condition modeling network; and training the initial feature extraction network based on the first point cloud feature vector, the first text feature vector, and the joint modal feature vector to obtain a target feature extraction network. The application can improve the problems of information degradation and insufficient collaboration in the prior art.