399 results about "Scale space" patented technology

The invention provides a dense overlapping target detection method based on a wavelet enhancement sparse hybrid expert model. The method comprises the following steps: firstly, extracting multi-layer features through a backbone network to capture multi-scale spatial representation; secondly, discrete wavelet transform is introduced to each level of features, spatial features are decomposed into a frequency domain, collaborative modeling of frequency domain and spatial domain features is realized, the reservation capability of detail and texture information is improved, a lightweight dynamic hypergraph aggregation module is introduced into the deepest layer of features, a hyperedge structure is adaptively learned, and the feature fusion is realized; modeling a high-order incidence relation in a local area in an explicit manner; and thirdly, in the decoding process, candidate queries are screened and reweighted through an IoU perception query selection mechanism, and a dynamic routing mechanism of sparse hybrid experts is introduced, so that query self-adaptive specialized representation learning is realized, and the target detection precision and reliability in a complex scene are effectively improved.

The invention provides a digital twin three-dimensional model construction method based on a generative large model, and the method comprises the steps: obtaining multi-source monitoring data of a power distribution network, and processing the multi-source monitoring data into a training data set; the method comprises the following steps: mapping multi-source monitoring data into a multi-scale tensor subspace through tensor wavelet structured transformation, adaptively extracting spatial features through a learnable wavelet kernel, and keeping the structural continuity of a physical field in combination with a geometric prior regular term; constructing and training a generative adversarial network through a training data set; inputting and analyzing the physical parameter vector of the target scene, and if the topological similarity score is lower than a preset threshold value, adjusting noise vector regeneration; and if yes, outputting a three-dimensional model tensor and importing the three-dimensional model tensor into a digital twin platform, and driving real-time physical field visualization. According to the method, characteristics of a multi-scale space structure and a nonlinear physical field can be reserved, the physical rationality and generalization ability of the generated model are remarkably improved, and depth identification of topological attributes (such as hole connectivity and surface defects) and local geometric defects of the three-dimensional model is realized.

Some embodiments of a method may include: obtaining a reference 3D Gaussian frame, a camera position C, and a time t; extracting a multi-scale feature for each 3D Gaussian of one or more 3D Gaussians using a neural network block, wherein the multi-scale feature represents multi-scale spatial information about a dynamic object or scene; predicting 3D motion based on the multi-scale features and the time t; predicting a 3D Gaussian frame for time t by manipulating the one or more 3D Gaussians in a spatial domain based on the predicted 3D motion; and outputting the 3D Gaussian frame for time t.

The invention belongs to the field of artificial intelligence and electroencephalogram emotion recognition, and provides a double-branch electroencephalogram emotion recognition method and system based on brain region topology and time-space, and the method comprises the steps: preprocessing a to-be-recognized electroencephalogram signal to obtain a plurality of electroencephalogram fragments, and extracting a difference entropy sequence of each electroencephalogram fragment and a Spearman correlation coefficient matrix between channels; based on the Spearman correlation coefficient matrix, utilizing a bridging dynamic graph attention network module to extract topological features of a brain region; processing the differential entropy sequence by using a multi-scale space-time mixed attention module to obtain multi-scale space-time features; carrying out residual mutual cross attention fusion on the topological features of the brain region and the multi-scale spatial-temporal features to obtain fusion features; and performing classification based on the fusion features, and determining an emotion recognition result corresponding to the electroencephalogram signal. According to the method, the accuracy and robustness of emotion recognition are improved by utilizing the spatial topology characteristics and the multi-topology time dynamic characteristics of the electroencephalogram signals, and the defects of modeling spatial dependence and time dynamic are overcome.

The invention discloses a three-dimensional blood vessel image segmentation method and system. Belongs to the technical field of medical image processing and particularly relates to the technical field of three-dimensional blood vessel image segmentation. The method solves the following problems existing in a blood vessel segmentation task in an existing method: small blood vessel features are difficult to accurately extract in a CTA image which is low in contrast and contains noise and artifacts; global context modeling is difficult to consider and local correlation is difficult to guarantee, so that long-distance dependent modeling is insufficient or a local structure is fractured; limited by a fixed geometrical shape of a traditional convolution kernel, the traditional convolution kernel is difficult to adapt to deformation characteristics of a complex topological structure of a blood vessel, resulting in discontinuous segmentation or fuzzy boundary of a branch region. Channel dynamic grouping and energy-driven attention generation are achieved through a grouping self-adaptive attention module, and self-adaptive modeling of a blood vessel complex branch structure and a geometrical shape is achieved through a multi-scale space structure aggregation module in combination with a strip-shaped deformable convolution and cross-scale guiding mechanism.

The invention discloses a data container storage and management method based on a multi-scale space subdivision grid, and the method comprises the steps: constructing a multi-scale space subdivision grid coding system covering the earth surface and the height dimension based on an earth space subdivision theory, and endowing each grid with a unique code to form a unified space-time framework; multi-source data of images, vectors, non-space and the like are collected and transmitted to a processing center, and after consistency detection and normalization processing are carried out, unified grid codes are generated for various kinds of data based on the coding system, and mapping is established; and finally, storing the coded data into a database, updating an index, realizing dynamic scheduling, logic aggregation and visual display of multi-source data by means of unified coding, realizing efficient management and control and cross-domain collaboration of massive heterogeneous data, and improving data storage consistency and retrieval and display efficiency. According to the method, the problems of fragmentation of spatial big data, slow retrieval response, difficulty in cross-domain fusion and the like are effectively solved, and the management efficiency and the sharing depth of the spatial data are remarkably improved.

The invention relates to the technical field of computer vision, and discloses an anomaly detection method and system for unmanned aerial vehicle fence routing inspection, a terminal and a storage medium, and the method comprises the steps: obtaining routing inspection video data collected by an unmanned aerial vehicle, and carrying out the preprocessing of the routing inspection video data, and obtaining an image sequence; inputting the image sequence into a pre-constructed enhanced YOLO target detection model, and capturing texture details and structure information in a multi-scale space according to the image sequence to obtain a single-frame detection result; the single-frame detection result is input to a track-level time sequence verification module, a detection target is identified according to the single-frame detection result, and a target track sequence is generated; and analyzing the duration, the movement direction and the speed change of the target track sequence, and triggering an alarm if the detection target is judged to be a real fence abnormal event. According to the invention, high-precision identification and stable detection of abnormal events such as fence damage, deformation, personnel crossing and the like are realized.

The invention discloses a self-supervised traffic flow prediction method based on multi-scale space-time-frequency fusion. The method comprises the following steps: acquiring enhanced data; performing multi-scale spatial-temporal feature coding; performing frequency domain residual filtering; generating a traffic flow prediction result; and carrying out joint target optimization. According to the invention, the multi-scale space-time frequency encoder is designed, local and global time features are captured at the same time through the mixed time encoding module in the time dimension, the multi-scale space encoding module aggregates space features under different distances in the space dimension, and the prediction precision is significantly improved. A frequency domain residual filtering module is embedded in the encoder to adaptively purify frequency domain features in an end-to-end mode, enhance key periodic features and suppress irrelevant noise, space-time features are fused through residual connection, space-time-frequency three-dimension collaborative modeling is achieved, meanwhile, frequency domain consistency loss is introduced in the optimization stage, and time-frequency three-dimension collaborative modeling is achieved. And the method is more robust when facing real traffic data containing noise.

The invention relates to a multi-modal perception fusion target following method and system. The method comprises the following steps: adopting an improved KCF algorithm to realize a closed-loop process of target tracking, multi-scale space construction, feature fusion, response calculation, optimal scale decision and model updating; designing a four-level shielding processing mechanism fusing motion prediction and depth verification, and realizing tracking recovery in a shielding scene through shielding judgment, motion prediction, fine search and template protection; constructing a target distance mapping function fusing geometric distortion correction and attitude compensation, and realizing high-precision distance estimation based on monocular vision; laser radar point cloud information is integrated, an obstacle threat degree model is constructed, and cooperative path planning of following and obstacle avoidance is realized in combination with an improved TEB algorithm; and designing a linear velocity control law and an angular velocity control law based on the distance deviation and the azimuth angle deviation, and driving the robot to complete target following motion. According to the invention, high-precision and robust following of the robot to the target can be realized.

The invention belongs to the field of structural modal identification, and particularly relates to an automatic modal parameter identification method based on a multi-scale space, which comprises the following steps: collecting a vibration data sample from a target structure, the vibration data sample comprising an acceleration signal of the structure; selecting power spectrum calculation parameters including a window function and a fast Fourier transform (FFT) length; performing frequency spectrum estimation on the acceleration signal by using interpolation power spectrum estimation to obtain frequency spectrum data; analyzing the spectrum data by using a multi-scale space algorithm, and detecting a peak value in the spectrum data to obtain a peak value frequency; through a frequency domain decomposition method, according to the peak frequency, modal parameters of the structure are calculated, the modal parameters comprise the modal frequency and the damping ratio, interpolation power spectrum estimation and a multi-scale space algorithm are innovatively combined, automatic and accurate identification of the modal parameters is achieved, and the identification precision and efficiency are remarkably improved.

The invention discloses an epilepsy abnormal brain network identification method based on a multi-scale static-dynamic fusion network, and belongs to the field of brain image analysis. The method comprises the following steps: firstly, constructing a static function connection weighted graph and a dynamic function connection graph; and fusing the static and dynamic representations by adopting a cross attention module. In order to describe a multi-scale spatial relationship, performing lexical meta-processing on brain connection according to anatomical partition and a functional network; and the local-global fusion module is used for integrating the fine granularity and the macroscopic relationship, so that the brain region with diagnostic significance is highlighted. In the training stage, cross entropy, reverse contrast loss and sparse regularization based on contrast graph adjacency matrix entropy are jointly used. The method is verified on multi-center functional magnetic resonance data, compared with other mainstream depth models, the classification accuracy, generalization and interpretability are remarkably improved, an abnormal brain region consistent with an epilepsy network can be positioned, and brain image markers with biological significance can be connected and recognized.

The invention relates to the technical field of unmanned aerial vehicle remote sensing detection, and discloses a change detection method and system for an unmanned aerial vehicle remote sensing image. The method comprises the following steps: acquiring a multi-temporal unmanned aerial vehicle remote sensing image data set containing high-resolution image data acquired at different times in a target area; carrying out geometric correction and radiation normalization processing on the data set to generate a standardized remote sensing image data set, and eliminating geometric distortion and illumination difference between images; then extracting multi-scale spatial features in the standardized remote sensing image data set, generating a spatial feature matrix containing texture, spectrum and structure information, and comprehensively capturing image details; inputting the spatial feature matrix into a time sequence feature fusion network, calculating feature differences among different time phase images, and generating a change feature vector; and finally, constructing a dynamic change detection model based on the change feature vector, and generating a target area earth surface change detection result.

The invention discloses an attention anti-mask double-branch distillation method for three-dimensional point cloud completion. The method comprises the following steps: firstly, extracting a multi-scale space attention map from a teacher model, and generating an anti-mask point cloud based on the multi-scale space attention map; then constructing a double-branch student network, and respectively processing the intermediate feature point cloud and the anti-mask point cloud of the student model; then, a joint recursive distillation module is adopted to carry out recursive upsampling and feature aggregation under geometric consistency constraint on point cloud features of the two student branches and the teacher model; and finally, training a student model by utilizing a triple supervision optimization module and jointly optimizing CD distance loss, characteristic distillation loss and cross-branch contrast loss. According to the invention, the point cloud completion precision and robustness of a student model in a complex automatic driving scene are effectively improved by constructing an anti-mask mechanism guided by teacher attention, introducing a double-input branch architecture, designing a combined recursive distillation mechanism and constructing a triple distillation loss function at the same time.

The invention belongs to the technical field of a method for automatically extracting a specific target in a complex landscape, and particularly relates to a method for automatically extracting a specific target in a complex landscape based on multi-source remote sensing time series data, which comprises the following steps of: acquiring multi-source remote sensing time series data, preprocessing the multi-source remote sensing time series data, and constructing a Pheno-ViT model; the Pheno-ViT model comprises a backbone network, a phenology sensing module and a feature fusion module, the backbone network adopts a SwinTransform architecture, the SwinTransform architecture comprises four feature extraction stages, multi-scale spatial feature extraction is sequentially executed in each stage, the phenology sensing module adopts a bidirectional GRU network and is used for dynamically generating attention weights of all time steps to model crop growth time sequence features, and the feature fusion module adopts a cross-modal attention mechanism and is used for carrying out feature extraction on the crop growth time sequence features. The image fusion module is used for fusing the Sentinel-2 optical image features and the Sentinel-1 SAR image features, and the decoder adopts four up-sampling stages and is used for realizing spatial resolution reconstruction and outputting a non-grain crop feature map.

The embodiment of the invention discloses an architectural drawing signature character recognition and control method and device, equipment and a medium. A specific embodiment of the method comprises the steps of obtaining an original architectural drawing image; performing full-graph character perception detection on the original building drawing image to obtain coordinate information of a full-graph detection textbox; generating an image label ROI image based on the coordinate information of the total image detection textbox; self-adaptive direction correction is carried out on the picture label ROI image, and a picture label ROI image after direction correction is obtained; performing scale space transformation enhancement on the ROI image of the picture label after the direction correction to obtain an enhanced ROI image of the picture label; performing structured semantic analysis extraction on the enhanced drawing tag ROI image to obtain structured drawing metadata; and based on the structured drawing metadata, downstream physical equipment is controlled to execute associated automation operation. The implementation mode provides key technical support for digital management of the constructional engineering drawings.

The invention provides an emotion recognition method based on electroencephalogram feature fusion and a double-stage attention mechanism, and the method comprises the following steps: A, electroencephalogram signal processing: carrying out the preprocessing of an electroencephalogram signal; and B, double-stage attention feature fusion: in each selected frequency band, adopting a double-stage attention mechanism to fuse the electroencephalogram features, and generating fusion features for emotion classification. And C, double-branch feature extraction: performing double-branch 3D convolution processing on the fused features, extracting multi-scale space-spectral time features, and splicing the multi-scale space-spectral time features along a channel dimension to form uniform features. And D, classification and output: inputting the unified features into a classifier, and generating an emotion category prediction result through a flattening layer and a full connection layer. According to the method, the difference entropy, the power spectrum density and the difference entropy asymmetry feature are fused through unified three-dimensional feature representation, a double-stage attention mechanism is introduced, and high-accuracy emotion recognition is achieved.

The invention provides a continuous action recognition method based on a spatial-temporal characteristic dynamic attention fusion network. The method comprises the following steps: constructing an action recognition model; the method comprises the following steps: acquiring an action signal of a to-be-measured target through a millimeter-wave radar, and preprocessing the action signal; constructing a continuous point cloud sequence according to the preprocessed action signal; merging the continuous point cloud sequences; the merged point cloud is input into the trained action recognition model, and a time sequence feature extraction module is adopted to extract a point cloud sequence to carry out adaptive multi-scale time sequence features; performing multi-scale neighborhood aggregation on each frame of point cloud by adopting a spatial feature extraction module; performing complementary fusion on the self-adaptive multi-scale time sequence features and the multi-scale spatial features by adopting a spatio-temporal feature interactive attention fusion module to obtain fusion features; inputting the fusion features into a classifier to obtain a continuous action recognition result; the multi-scale graph convolutional network structure with the adaptive domain weight distribution strategy designed by the invention dynamically adjusts the contribution weight of neighborhood points to feature extraction according to the local density of the point clouds and the spatial geometrical relationship aiming at the characteristics that the millimeter wave radar point clouds are irregularly distributed and the spatiality is difficult to excavate; the problems of coarse granularity and poor robustness of spatial feature extraction of a traditional method are solved.

The invention belongs to the field of hyperspectral image classification, and discloses a hyperspectral image classification method and system based on multi-scale spatial-spectral joint representation and dynamic context modeling, and the method comprises the steps: carrying out the feature dimension reduction processing of a hyperspectral image through principal component analysis; spatial spectrum collaborative information of hyperspectral data is deeply mined through a multi-scale spatial spectrum joint characterization module, and adaptive fusion and enhancement of spatial spectrum characteristics under different scales are realized; a dynamic context modeling strategy is introduced, and the perception ability of the model to context information is optimized by establishing a long-range dependency relationship between features; advanced feature integration and nonlinear transformation are carried out through a multi-layer perceptron, and precise classification of hyperspectral image ground objects is completed. According to the method, the performance superior to that of a current mainstream method is obtained on three public data sets, and the effectiveness and generalization ability of the method are verified.

The invention relates to the technical field of image processing, in particular to an aluminum alloy die casting surface defect detection method based on machine vision. The method comprises the steps of obtaining a grayscale image of the surface of a die casting; constructing a Gaussian scale space of the grayscale image, and obtaining scale images under a plurality of scales; calculating a defect saliency index corresponding to each pixel point in the grayscale image based on the multi-scale image, and generating a defect saliency map; and performing threshold segmentation on the defect saliency map to obtain a segmented image, performing morphological processing identification on the segmented image, and determining the position of a surface defect area. According to the invention, missed detection and false detection in the surface detection process of the aluminum alloy die casting can be reduced.

The invention provides a monocular depth estimation method and product based on convolution compensation dual-channel self-attention, and relates to the field of computer vision. Comprising the following steps: converting an event flow of a target scene into three-dimensional tensor representation; obtaining event image fusion multi-scale spatial features based on the image of the target scene and the three-dimensional tensor representation; modeling spatial context correlation in a spatial dimension by utilizing event image fusion multi-scale spatial features through a context modeling self-attention branch to obtain a context modeling self-attention result; through a modal fusion self-attention branch, using the event image to fuse the modal correlation of the multi-scale spatial features in the channel dimension modeling image and the event, and obtaining a modal fusion self-attention result; and pixel-level depth prediction is carried out by using a context modeling self-attention result and a modal fusion self-attention result to obtain a depth map, so that complementary characteristics between an event and an image are fully mined, fine-grained depth fusion expression is realized, and depth estimation precision and generalization ability are effectively improved.

The invention relates to the technical field of image processing, and discloses a printing defect detection method based on image processing, and the method comprises the steps: generating an illumination distribution diagram according to the regional illumination distribution characteristics of a printed matter image, and carrying out the self-adaptive illumination correction of the printed matter image, and obtaining a corrected image; performing multi-scale space decomposition on the corrected image to obtain a first scale texture feature and a second scale texture feature; based on a preset space reference, performing granularity alignment on the first scale texture feature and the second scale texture feature to obtain a fused texture feature; performing multi-modal feature coupling on the contour feature of the corrected image and the fused texture feature to obtain a comprehensive feature, and identifying a potential defect region according to the difference between the comprehensive feature and a reference template; based on the reference template, performing multi-dimensional feature recognition on the potential defect area to obtain a defect area; according to the invention, the printing defect detection efficiency of image processing can be improved.

The invention discloses a fine-grained urban functional area identification method based on multi-source data, and belongs to the technical field of intelligent identification. The method comprises the following steps: firstly, performing spatial registration and preprocessing on remote sensing images and POI data; then generating a self-adaptive multi-scale space unit in a data driving mode by mining a spatial co-occurrence relation and semantic embedding of POI categories; on the basis, a multi-scale attention module for globally guiding the local part is designed, and context visual features are dynamically aggregated to enhance the discrimination ability of the local area; heterogeneous feature fusion is carried out based on the attention weight, and deep complementation of remote sensing image visual features and POI semantic features is achieved; and finally, performing end-to-end training through a multi-layer joint loss function, and outputting an urban functional area identification result. According to the method, adaptive generation of the analysis unit and deep collaborative fusion of heterogeneous modal features can be realized, and the precision, robustness and scene generalization ability of urban functional area recognition can be remarkably improved.

The invention belongs to the field of computer vision, particularly relates to an air-ground cross-view-angle pedestrian re-identification method based on multi-scale and view angle perception double flow, aims to match the same pedestrian individual under different view angles, and is applied to the industries of intelligent monitoring, public safety and the like. The system comprises two core branches, the multi-scale flow captures feature representations of different granularities, multi-scale semantic features are grouped and aggregated, and a feature representation which keeps consistency in a multi-scale space is learned, so that the model has robustness. According to the visual angle perception flow, a dynamic training strategy is adopted for different visual angle pairs, difficult samples are mined, the model can process visual angle changes more intelligently, and the cross-visual-angle matching capacity is remarkably improved. According to the method, feature representation with higher robustness and discrimination capability can be learned, and the Re-ID performance in a complex air-ground scene is remarkably improved.

The invention discloses a time-space multi-scale feature fusion deep learning landslide susceptibility evaluation method based on number object dual drive, and the method comprises the steps: constructing a time sequence branch + Swindow-Transformer (upper branch) + CNN (lower branch) time-space fusion DL-LSA model, combining an AFF multi-scale space fusion module and a TSF time-space fusion module, and carrying out the deep learning of landslide susceptibility. A complete technical scheme of high-quality sample screening, deep spatial-temporal feature extraction, multi-scale adaptive fusion and dynamic probability prediction is formed, and normal form upgrading of geological disaster risk management from single-drive vector object double-drive and from static evaluation to dynamic early warning is promoted. Comprising the following steps: 1, acquiring influence factors (X); 2, constructing an initial sample set (X-Y pairs); 3, preprocessing the influence factor (X); and 4, influence factor coding (generating a standardized feature X '). And 5, providing physical constraints (optimizing a sample set X '-Y) based on the P-LSA model. And 6, constructing a space-time fusion DL-LSA model. And 7, designing a feature fusion module. And 8, outputting a landslide susceptibility evaluation result.

The invention discloses a multi-scale convolutional layer structure and application thereof in target detection, and belongs to the technical field of computer vision. The structure comprises a multi-scale feature extraction part and a feature aggregation extraction part, multi-scale feature synchronous extraction of an input feature map is realized by arranging a plurality of different-scale convolution kernels and expansion convolution modules in parallel, and the perception ability of a model to target multi-scale spatial features is enhanced. Pixel-by-pixel addition and convolution operation is adopted in the feature aggregation part, efficient fusion and refining of multi-scale features are achieved, and the richness and distinction degree of feature expression are improved. The structure can directly replace a common convolutional layer in an existing target detection network, significantly improves the feature extraction efficiency and scale adaptability of the model on the premise of keeping the number of input and output channels consistent, is suitable for detection scenes with complex backgrounds and target scale dynamic changes, and has good robustness and practicability.

The application discloses a SIFT image splicing method and system based on multi-feature fusion and adaptive optimization, and comprises the following steps: performing data preprocessing on acquired image data to generate a gray image; constructing a scale space by using a SIFT feature extraction method, detecting local extreme points in the gray image to generate a descriptor, and performing SIFT feature matching optimization; performing false matching elimination by using a nearest neighbor ratio criterion and a matching threshold value, and maintaining the distribution balance of feature points in the gray image by using a uniform sampling method; performing random sample consensus estimation by using an adaptive threshold value in a feature matching point set, eliminating outliers, and calculating a homography matrix to realize image registration; and realizing image splicing in an overlapping area based on a weighted fusion strategy, avoiding the abruptness of a joint transition by combining a smoothing weight, and performing black edge trimming to optimize a splicing result. The application can optimize a traditional SIFT image splicing algorithm, and thus the splicing quality of an image can be improved.

The invention discloses a land utilization change comprehensive feature analysis method based on multi-source heterogeneous data fusion, and relates to the technical field of remote sensing and geographic information processing. Building volume, building density, night light, surface temperature and population density are aligned under unified coordinates and resolution; constructing an average building height according to volume / density and coding and recording quality; uniform noctilucence is realized through partition three-order mapping; interpolation is carried out by adopting fractional learning under the missing measurement mask, and pixel-level uncertainty is output; uniformizing the annual temperature; extrapolation is carried out through multi-scale space-time small block coding-decoding containing flux conservation and Laplace constraint, and uncertainty is given; implementing uncertainty weighted principal components according to the five indexes to obtain a comprehensive strength index; concentric buffering, set splitting regression and partial response confidence band recognition threshold values are combined. The method has the advantages of cross-period comparability, physical consistency, traceable uncertainty and the like.

The invention discloses a self-supervised monocular depth estimation method based on an enhanced multi-scale pose network, and relates to the field of computer vision. The problems that in an existing self-supervision monocular depth estimation method, the pose network structure is simple, the time sequence modeling capacity is insufficient, and geometric constraints are missing are solved. According to the method, a self-supervision joint training framework composed of a depth estimation sub-network and an enhanced pose estimation sub-network is constructed; wherein the pose estimation sub-network extracts multi-scale spatial structure features through a layer-by-layer feature fusion encoder, and adopts a context fusion decoder based on time sequence attention to model a motion dependency relationship between continuous frames; meanwhile, a self-supervised pose consistency loss function is introduced, geometric continuity of a camera track is enhanced through forward and reverse transformation consistency constraint and closed-loop geometric constraint, and collaborative optimization of depth prediction and pose estimation is realized. The method is also suitable for the application fields of automatic driving, robot perception, augmented reality and the like.

The invention discloses an action sports scoring method and device based on big data analysis, and relates to the technical field of computer vision, and the method comprises the following steps: S1, constructing a human body posture tensor manifold; s2, generating aligned action track characteristics; s3, generating symmetrical positive definite manifold features; s4, generating deep manifold distribution parameters; s5, quantifying global distribution deviation characteristics; s6, calculating a residual vector based on the deep manifold distribution parameters and the standard action distribution parameters, inputting the residual vector into the improved AGCN model for processing, constructing an adaptive topology based on the residual vector, and performing aggregation analysis to obtain a joint-level physical angle error after multi-scale space-time convolution and manifold enhancement; and S7, outputting a comprehensive score vector. According to the method, the limitation that a traditional method only depends on local geometric features and ignores global topological structure constraints and statistical distribution priori is overcome, and an efficient solution is provided for intelligent scoring of sports actions.

The invention relates to a power distribution network tower defect automatic identification and classification method and system based on deep learning. According to the method, firstly, a tower main body area is positioned and extracted through a convolutional neural network, and background interference is eliminated; and then the visual saliency of the defect area is improved by adopting a self-adaptive contrast enhancement algorithm based on local statistical characteristics. In the feature extraction stage, a pyramid distraction attention module is introduced to fuse multi-scale space information and channel attention, and a two-dimensional selective state space module is used for modeling a long-range dependency relationship. And multi-resolution features are further aggregated through a layered feature fusion architecture and a self-adaptive anchor frame mechanism, and targets of different sizes are matched. And finally, a self-adaptive edge enhancement module is adopted to enhance the edge of the defect, and a multi-branch detection head is adopted to realize category judgment, position regression and confidence evaluation of the defect in parallel. The method effectively improves the detection precision and robustness of tower defects under a complex background, and is especially suitable for the automatic recognition of micro-scale defects.