3228 results about "Identification technology" patented technology

The invention relates to the technical field of multi-image fusion recognition, in particular to a data fusion method and system for CCD visual detection, and the method comprises the following steps: obtaining a horizontal pixel row calculation gradient construction trend sequence, repairing an edge fracture to generate an integrity index, extracting a gray value to detect feature mutation, and distributing fusion weights to establish a mapping relation. And executing image fusion and balancing the contrast to generate a fusion matrix result. According to the method, the fracture edge region is identified, interpolation compensation is executed, the structural similarity index of the local gray sequence in the image overlapping region and feature direction mutation detection are combined, accurate identification of the edge matching result is guided, and fusion weight factor mapping corresponding to signal-to-noise ratio distribution is introduced; according to the method, the distribution relation between the pixels in the region and the credible weight is effectively established, the edge transition among the multi-source images is more natural through Poisson constraint and contrast balance adjustment of the fusion region, and the structural fidelity and the judgment stability of the fusion image are remarkably enhanced.

The invention relates to the technical field of remote sensing image recognition, in particular to a remote sensing sewage area recognition method and system based on a graph structure and multi-stage enhancement. The method comprises the steps of performing data preprocessing and representation enhancement on an acquired remote sensing image; performing sewage salient region preliminary screening on the enhanced remote sensing image, including abnormal enhancement mapping construction based on local statistical distribution; pollution candidate graph extraction based on spatial structure prior driving; enhancing the response of the stable region based on a structure consistency enhancing mechanism of the polluted region; high-precision segmentation and identification of the sewage area comprises the following steps: constructing a multi-resolution residual pyramid structure; carrying out fine-grained boundary structure modeling and uncertainty suppression; generating a sewage distribution probability graph and optimizing structural consistency; according to the method, the multi-resolution residual pyramid structure is constructed, image context information under different perception scales is fully mined, and the sensitivity and edge integrity of the model to a sewage area under a complex texture background are remarkably enhanced.

The invention relates to the technical field of pet emotion recognition, in particular to a pet dog emotion recognition method and system based on deep learning. Comprising the steps of collecting pet dynamic data, pet physiological data and scene data to obtain a structured data set; labeling the structured data set through a cross validation labeling mechanism to obtain a labeled data set; multi-modal features are extracted based on the labeled data set, and multi-modal feature integration is carried out through a cascade SEblock array to obtain multi-modal fusion features; adversarial sample data generated by a stress scene simulator is injected in a training stage, and deep learning model training is performed based on the adversarial sample data and the multi-modal fusion features to obtain a pet emotion recognition model; and performing emotion recognition on a to-be-recognized pet through the pet emotion recognition model to obtain a pet emotion recognition result. According to the method, the data quality and the model robustness of pet emotion recognition are improved, and then the human-pet interaction quality is improved.

The invention discloses a video analysis-based multi-scene operator violation behavior identification method and system, and belongs to the technical field of intelligent operation safety monitoring and artificial intelligence identification, and the method comprises the steps: collecting a real-time video stream of a multi-scene operation site; recognizing a continuous action time sequence in the real-time video stream by using an action recognition depth model; constructing the continuous action time sequence into an action behavior sequence; the action behavior sequence is constructed into a directed behavior graph with time, space and action labels, the directed behavior graph is compared with a directed behavior graph corresponding to the standard action behavior sequence, and illegal behaviors are recognized; and carrying out multi-mode early warning on the identified illegal behaviors. According to the method, the bottleneck that the traditional image recognition technology is weak in action sequence semantic understanding and poor in environmental adaptability is broken through, and accurate recognition and real-time early warning of illegal behaviors in multi-scene operation are achieved.

The invention relates to an intelligent visual identification technology-based cold solder joint microcrack intelligent discrimination method, which comprises the steps of collecting an initial RGB image of a to-be-detected welding spot, carrying out two-dimensional discrete cosine transform and inverse two-dimensional discrete cosine transform on the initial RGB image to obtain an enhanced image, and fusing the enhanced image with an R channel of the initial RGB image to obtain a fused image; forming a dual-channel feature map; calculating the phase consistency of the dual-channel feature map, and obtaining a suspected candidate region of the pseudo soldering microcrack through an adaptive threshold segmentation method; acquiring an RGB image sequence of a continuous time sequence of the welding spots, and performing anomaly detection to obtain an abnormal region set; and constructing a welding spot thermal diffusion model, and inputting the geometric parameters and the environmental parameters in the abnormal region set into the welding spot thermal diffusion model to obtain a final judgment result of the pseudo soldering microcracks. According to the method, through multi-dimensional feature fusion and continuous time sequence dynamic tracking, the detection precision of the pseudo soldering microcracks is remarkably improved, the false detection rate is reduced, and the final judgment result is more accurate.

The invention discloses a photovoltaic power grid fault identification method and system based on circuit analysis, and relates to the technical field of fault identification, and the method comprises the following steps: obtaining the operation parameters of a photovoltaic power grid, and constructing a circuit analysis model; based on the circuit analysis model, equivalent response curves in different fault scenes are extracted, and the reference operation state is compared to generate a differential residual sequence; performing time-frequency joint decomposition on the differential residual sequence, and stripping photovoltaic output fluctuation from a load disturbance component to obtain a pure circuit characteristic component; based on the pure circuit characteristic component, a multi-dimensional characteristic coordinate space is formed, and the fault type is judged by using the dynamic bending rate of the fault response track; and mapping a fault type discrimination result back to the circuit analysis model, and positioning the position of a fault branch in combination with local disturbance distribution of the node impedance matrix. According to the method, pure circuit characteristic component extraction and multi-dimensional characteristic space dynamic analysis are combined, and accurate judgment of complex fault types and fault branch positioning are achieved.

The invention relates to the technical field of Internet of Things anomaly identification, in particular to an Internet of Things anomaly traffic identification method and system based on a deep neural network. The method comprises the following steps: collecting communication data of each piece of IoT equipment in real time from an edge gateway of the Internet of Things; preprocessing the collected communication data, and constructing a multi-dimensional feature vector; based on a convolutional neural network and a bidirectional long-short-term memory network, performing time sequence feature extraction and anomaly discrimination on the multi-dimensional feature vector to output a traffic anomaly probability; and comparing the abnormal probability output by the depth time sequence modeling neural network with a dynamic threshold value, and if the abnormal probability exceeds a preset threshold value, determining that the traffic is abnormal. A gating mechanism is introduced into a bidirectional long-short-term memory layer, a gating coefficient is calculated at a time step level, the influence weight of time step information on final output is dynamically adjusted, feature expression of key time steps is strengthened, noise or irrelevant information is suppressed, and the sensitivity of a model to time sequence data is improved.

The invention relates to the technical field of boundary recognition, in particular to a glioma boundary recognition method and system based on image fusion, and the method comprises the following steps: obtaining a multi-modal brain image, constructing a fusion matrix, extracting the gray features of an edge region and an adjacent region, recognizing signal-noise abnormal points, and revising a judgment standard. And adjusting the path direction and reconstructing an edge communication structure, and generating a glioma boundary region map under fusion. According to the method, high-precision alignment among modals is realized through multi-modal image gray scale unification and registration processing, key details are expanded and focused by enhancing edges and regions, the recognition accuracy is improved, gray scale comparison between the edges and outer adjacent regions is introduced, the signal distinguishing capability is enhanced, misjudgment is avoided, judgment conditions are dynamically revised according to the signal-noise difference, and the accuracy of recognition is improved. The method enables the recognition standard to have local adaptability, combines the path change trend to reorder and connect edge points, guarantees the continuity of a boundary structure, integrally improves the accuracy and integrity of fuzzy boundary recognition, and enhances the glioma contour extraction effect.

The invention discloses a modulation and signal category identification method based on multi-scale attention and residual error, and relates to the technical field of signal type and modulation mode identification, and the method comprises the steps: obtaining signal data sets under different signal types and modulation modes, and dividing the signal data sets into a training set and a verification set; constructing an end-to-end deep learning model based on an input preprocessing module, a shared convolutional feature extraction module, a multi-task branch module and a joint loss optimization module; the shared convolution feature extraction module comprises a channel expansion convolution layer, a multi-scale attention residual module and a down-sampling module; an end-to-end deep learning model is trained; and inputting the data of the to-be-tested signal into the model to obtain the signal type and the modulation mode of the to-be-tested signal, and through end-to-end deep learning model processing, the problems of model redundancy, computing resource waste, insufficient inter-task information utilization and the like can be solved, and the model complexity and the training overhead are reduced while the identification accuracy is improved.

The invention discloses a robot obstacle avoidance method and system based on millimeter wave radar sparse point cloud, and relates to the technical field of obstacle avoidance recognition. A robot obstacle avoidance system based on millimeter wave radar sparse point cloud comprises a point cloud acquisition module, a negative obstacle identification module, a weak obstacle identification module, a point cluster identification module, a risk map module, a tentative verification module and an obstacle avoidance decision module. According to the invention, suspected obstacle point clusters are extracted based on a reflection intensity threshold and a spatial proximity relation in an enhanced point cloud, a theoretical parallax model of a real static obstacle is constructed under the constraint of a robot motion trajectory, and Doppler velocity distribution of each frame is combined with a static obstacle Doppler physical law for comparison. And classifying the point clusters which do not meet the multi-view geometric consistency or Doppler physical law, and distinguishing multipath false point clusters from dynamic point clusters.

The invention discloses a multi-modal data sensitivity grading method and system based on semantic risk map diffusion perception, and belongs to the technical field of artificial intelligence security and information content identification. Aiming at the problems of weak cross-modal linkage capability, insufficient context modeling, poor interpretability and the like of the existing multi-modal sensitivity identification method, the method realizes multi-modal fragment association by constructing a semantic risk unit, utilizes a semantic risk graph to model a cross-modal cooperative relationship, and adopts a thermal diffusion mechanism to simulate a risk propagation process, so that the method has the advantages of high sensitivity and high reliability. And finally, generating a structured interpretation path through the sensitive tag atlas. According to the method, high-precision multi-mode sensitive information identification and traceable grading judgment are realized, and the method is suitable for automatic compliance review of AIGC generation content.

The invention relates to the technical field of unmanned aerial vehicle detection and recognition, in particular to an unmanned aerial vehicle detection system and method based on vision, laser radar and sound waves, and the system comprises a sensor unit, a data processing fusion unit and a target recognition tracking unit. The data processing fusion unit performs multi-source data alignment and weighted fusion to extract texture, shape and color features of the image, point cloud generates a depth map, sound wave signals are mapped into a two-dimensional feature map, and weights are dynamically distributed based on sensor reliability to generate a unified feature map; the target identification and tracking unit identifies the type of the unmanned aerial vehicle by using the optimized deep learning model and realizes accurate prediction and updating of position and speed states in combination with Kalman filtering, and the decision response unit triggers sound-light alarm and wireless alarm for the non-cooperative unmanned aerial vehicle in real time and transmits target dynamic information to the ground station. And the target detection accuracy of the unmanned aerial vehicle in a complex environment is improved.

The invention relates to the technical field of traditional Chinese medicinal material identification, in particular to a traditional Chinese medicinal material intelligent identification and grading system based on deep learning, which integrates image acquisition, feature extraction, expression optimization, identification evaluation and origin traceability into a whole. Curvature, structure tensor and spectral features are extracted in combination with a differential geometry theory; constructing a Riemannian manifold representation space and performing isometric embedding dimension reduction optimization; identifying the types of the medicinal materials by using a deep convolutional neural network, and comparing with a standard model to evaluate the quality grade; the origin discrimination is realized based on the multi-scale feature comparison of geodesic distance, the category, quality and traceability information of the medicinal materials are comprehensively output, the surface visual features and internal component information of the traditional Chinese medicinal materials are comprehensively utilized through a multi-source data fusion technology, and the feature expression ability and discrimination precision of the recognition system are comprehensively improved.

The invention belongs to the technical field of automatic driving vehicle identification, and discloses an automatic driving vehicle point cloud identification method and system based on a laser radar. According to the method, an automatic driving vehicle point cloud recognition model is built, modular dynamic edge convolution based on a feature sensitivity layer number selection strategy is provided in the model, local geometric information is better captured by dynamically learning a local geometric structure, and the extraction capability of geometric feature information is enhanced. In order to prevent the problem of insufficient information extraction caused by layer number simplification of modular dynamic edge convolution, a selective kernel attention mechanism is introduced, a feature fusion mode is adjusted, residual connection is added, more comprehensive statistical information is captured while original information is reserved, and the multi-scale feature capture capability is improved, so that the classification performance of the model is improved. Especially in the face of noise, sparse data and complex object shapes, high classification accuracy can still be kept, and a more accurate environment perception capability is provided for an automatic driving system.

The invention discloses a machine learning-based weld defect intelligent identification system, relates to the technical field of weld defect intelligent identification, solves the technical problems of multi-modal data fusion precision and robustness optimization and defect shielding or overlapping feature deficiency, and provides a machine learning-based weld defect intelligent identification method based on PSNR dynamic parameter adjustment and gradient weight optimization. The limitation of existing fixed parameter denoising is solved, the edge feature retention rate of cracks, air holes and other defects is improved, the omission ratio is reduced, improved DeepLabv3 + segmentation semantic masks are introduced and mapped to point cloud voxels, geometric + semantic double-attribute enhanced point clouds are formed, the defect area positioning accuracy is improved, and through a cross-modal attention module, the defect area positioning accuracy is improved. Weights are dynamically distributed according to illumination intensity and workpiece materials, feature waste caused by fixed weights is avoided, depth mutation and a shielding area with semantic defects are positioned by utilizing depth information of enhanced point cloud, real overlapping and projection overlapping can be effectively distinguished by combining an improved Poisson fusion algorithm, and the overlapping defect recognition accuracy is improved.

The invention discloses a multi-modal communication signal intelligent identification method based on a neural network, and relates to the technical field of multi-modal signal identification. Comprising the following steps: acquiring original signals acquired by a sound sensor and a camera; inputting the processed original signal into a cross-modal auto-encoder to obtain a purification feature vector: processing the purification feature vector to obtain a quality scoring scalar; performing time sequence quality evaluation processing on the original signal to obtain a time sequence quality scoring sequence; fusing the time sequence quality scoring sequence and the quality scoring scalar to obtain a joint quality coefficient; based on the joint quality coefficient, adjusting a basic weight value of the original signal through weight nonlinear scaling processing, and generating a weight vector; processing the purified feature vector based on the weight vector to obtain a fused feature vector; and processing the fused feature vector to obtain a probability value representing the equipment failure rate. And through dynamic weight adaptive adjustment based on quality scoring, the high-quality signal occupies a larger proportion in the fusion process.

The invention discloses a methane point source emission remote sensing recognition and plume extraction method and a methane point source emission remote sensing recognition and plume extraction system, belongs to the technical field of methane point source recognition, and solves the problems of incomplete methane plume extraction and lack of morphological characteristics of an extraction result in a traditional signal filtering method by introducing threshold segmentation, spatial continuity and geographic space constraints. And the method has good applicability to the multi-point dissipation condition in the after-mine activity process of a coal mine area.

The invention belongs to the technical field of image processing, tracking and recognition, and relates to a multi-mode re-recognition method based on semantic-style decoupling distillation. The method depends on a multi-modal re-identification model which comprises a multi-modal feature extractor comprising a teacher branch module and a student branch module, a decoupling distillation module and a hierarchical self-supervised learning module, and comprises the following steps: constructing a mixed multi-modal feature extractor sharing a shallow layer and an independent deep layer to extract mixed features; performing dual supervision of semantic distillation and style distillation, modeling modal-invariant semantic information and modal-specific style information, and realizing effective decoupling of a feature space; a hierarchical self-supervised learning space is constructed, and in combination with intra-modal and cross-modal comparative learning, images under local damage and style disturbance conditions are scrambled; according to the method, recognition performance and reasoning efficiency are both considered, semantic features and modal specificity styles are effectively separated, semantic consistency, feature robustness and network learning efficiency are cooperatively improved, and modal specificity is also reserved.

The invention provides a multi-sensor fusion cloth online flaw identification method and system, and relates to the technical field of cloth flaw identification. The method comprises the following steps: acquiring specified maps of the front and back sides of current detected cloth; performing space-time alignment on the data in the specified atlas to obtain aligned multi-source data; key information is extracted from the multi-source data; according to the cloth type parameter of the current detected cloth, determining the fusion weight of each data in the key information; performing weighted fusion on data in the key information according to the fusion weight to generate a fusion feature vector; and comparing the fusion feature vector with the fusion feature of the normal cloth, and determining and positioning a chromatic aberration area. The method provided by the invention aims at solving the technical problem of performing color difference defect detection on the cloth which has double-sided property and light transmission and is subjected to an after-finishing process on a cloth manufacturing production line, and realizes more accurate identification and positioning of the color difference defect of the complex cloth.

The invention discloses a mine earthquake intelligent identification method based on multi-mode deep learning and signal processing, and belongs to the technical field of mine earthquake identification. Aiming at the problems that a traditional mine earthquake signal processing method cannot fully utilize spatial relevance and frequency domain characteristics of signals in a sensor network, the automation level is relatively low, the response speed is slow, and the emergency processing effect is influenced, seismic waveform data and multi-modal characteristic input are adopted, so that the emergency processing effect is influenced. Based on text semantic features generated by a large language model (LLMs), frequency domain features extracted by Fourier transform and time sequence features after obspy processing, a dynamic graph recognition model of a fusion graph neural network (GNN) and Transform is constructed; graph structure edge weights are dynamically updated through cross correlation between signals, and the structure change of complex environments such as mine roadways is self-adapted.

The invention relates to the field of traffic control, in particular to an intelligent network connection inductive control platform for road traffic safety facilities. The traffic data acquisition module is used for acquiring a data flow of a traffic sensor and outputting traffic data with a traffic data source identifier through an identification technology; the traffic data processing module is used for obtaining standardized traffic parameters according to a traffic data source identifier adaptive analysis protocol; evaluating a data reliability index according to the index system; the situation fusion module is used for forming a vehicle driving track through a graph neural network, generating a global traffic situation map by using a data reliability index, and constructing a traffic network digital twinborn model; and the decision and control module is used for analyzing traffic states from the traffic network digital twin model, predicting traffic events and collision risks and generating traffic control instructions. According to the platform, through the equipment fingerprint and protocol reverse technology, an information island is broken, the comprehensiveness and high credibility of a data source are ensured, and the road traffic safety and passing efficiency are remarkably improved.

The invention relates to the technical field of rural image inspection and recognition, and discloses a multi-source image collaborative inspection and recognition analysis system and method for a digital rural area, and the method comprises the steps: collecting multi-source image data in real time; obtaining a plurality of characteristic parameters corresponding to each image data item in the image data set, and carrying out space-time registration and multi-scale fusion processing on the plurality of characteristic parameters of each image data item; performing target detection and identification analysis on the plurality of feature parameters in the fusion feature parameter set, and constructing an abnormal point identification model; setting a plurality of abnormal point change thresholds according to the inspection coordinate data set for classification processing to obtain a plurality of abnormal point categories; and setting a corresponding co-processing scheme according to the plurality of abnormal point categories, and setting early warning information corresponding to the change trends of the plurality of abnormal point categories based on the co-processing scheme. According to the invention, the intelligent degree and response efficiency of rural inspection are improved, and the safety and sustainable development of digital rural construction are effectively guaranteed.

The invention relates to the technical field of landslide geological disaster hidden danger point identification, in particular to a landslide hidden danger point identification method based on an air-space remote sensing fusion neural network, which comprises the following steps: acquiring satellite SAR data, a digital elevation model, a vector boundary map and a precision orbit ephemeris of a research area; carrying out interference processing on the InSAR data, and carrying out deformation rate inversion, terrain residual error correction and atmospheric delay phase modeling and correction; preprocessing an original high-resolution remote sensing image, interpreting a high-resolution image, and outputting spatial distribution of suspected landslide hidden danger points in a research area; planning an aerial survey route of the unmanned aerial vehicle, and performing distortion correction, feature matching, dense point cloud generation and orthoimage generation processing; performing multi-dimensional feature analysis, and performing comprehensive interpretation in combination with an expert knowledge base; according to the method, the key effect of the multi-source remote sensing fusion technology in landslide hidden danger monitoring is displayed from data acquisition to hidden danger recognition.

The invention discloses a PDF engineering drawing structured recognition method and device based on multi-dimensional feature fusion and a storage medium, and belongs to the technical field of engineering drawing intelligent processing. The method comprises the following steps: PDF analysis and preprocessing: analyzing a PDF engineering drawing file, extracting a vector graph drawing instruction, text block contents and respective original coordinate information, and carrying out image rendering and standardized preprocessing on a drawing page to generate a binary image with a standard resolution; multi-dimensional feature extraction: extracting the following four types of features from the binary image: a visual depth feature, a text semantic feature, an industry specific symbol feature and a spatial topology feature between elements; feature fusion and relation reasoning: inputting the visual depth features, the text semantic features, the industry specific symbol features and the spatial topological features into a graph neural network fusion module, and constructing a graph structure taking each identified element as a node and taking a spatial topological relation between the elements as an edge, performing feature interaction and fusion among nodes through an attention mechanism, and reasoning a semantic association relationship among drawing elements; and generating structured data: classifying and grouping the drawing elements according to the semantic association relationship, and outputting a structured data file according to a predefined structured format. Therefore, the technical problems of low recognition efficiency and accuracy in the prior art are solved, and the technical effect of efficient and accurate engineering drawing structured recognition is achieved.

The invention discloses an optical cable micro-disturbance identification method based on physical simulation and self-supervised time sequence decoupling, and relates to the technical field of optical cable identification, and the method comprises the steps: constructing a physical digital twin simulator, and generating a high-fidelity training set; constructing a deep learning model, wherein the deep learning model adopts a lightweight time sequence decoupling network; training the model by adopting a staged training strategy, and sequentially carrying out self-supervised noise distribution pre-training, simulation supervised training and spectral domain physical consistency fine tuning operation; inputting DAS time sequence data collected in real time into the trained model, and outputting the data as an optical cable identity ID and a physical position; the lightweight time sequence decoupling network comprises a physical guide preprocessing module, a lightweight U-Net separation module, a sparse gating module and an intelligent parallel decoding module. Through the technical means of simulation-driven data generation, staged training strategies and the like, the defects of the prior art in the aspects of reducing the data cost, improving the detection capability in a low SNR environment, realizing multi-source blind source separation and the like are overcome.

The invention discloses a traditional Chinese medicine quality tracing and identifying method based on a block chain two-dimensional code, and particularly relates to the technical field of quality tracing and identifying. The method comprises the following steps: extracting effective component contents, trace element contents and biological activity indexes of traditional Chinese medicinal materials, constructing a recessive feature data set reflecting the real quality state of the traditional Chinese medicinal materials, and further generating unique corresponding medicinal material quality block chain tracing identification information by adopting feature dimension reduction, Hash mapping and block chain binding means; in combination with an under-chain recessive feature database and an intelligent contract technology, accurate mapping and two-way verification of the traditional Chinese medicine recessive features and the block chain two-dimensional code are realized, and the authenticity of traditional Chinese medicine quality tracing and the accuracy of an identification result are improved.

The invention discloses a gas recognition method based on multi-source information fusion and environmental perception, and the method comprises the steps: constructing a deep feature learning framework of multi-source fusion through combining the spatial response features, time sequence features and external environmental information of gas; the method comprises the following specific steps: preprocessing collected gas data, and respectively extracting features of an image mode, a sequence mode and an environment mode; fusing the image features and the sequence features through a cross attention fusion module, and capturing the space-time correlation of the data; a cross-modal attention compensation module is introduced, so that main-modal gas data adaptively gathers key information in an auxiliary-modal environment, and effective compensation of environmental factors on gas recognition performance is realized; and finally, gas category prediction is performed through a classification decision head. According to the method, the problems that the detection is easily interfered by environmental factors, the stability is poor or the qualification is inaccurate due to the fact that modeling depends on single modal data in the existing gas identification technology are solved.

The invention relates to the technical field of semantic recognition, can be applied to business scenes such as financial science and technology and medical health, and discloses a business response method, device and equipment based on communication theme recognition and a medium. And calculating the distribution density of the words in the plurality of text units, generating word weights in combination with word frequencies and the distribution density, identifying concerned topics, generating user feature tags or content classification identifiers, and associating with a service information system to trigger service operation. According to the method, the communication record is converted into the text data, the word frequency and the word distribution density are calculated, and the word weight with the discrimination degree is generated, so that a user concerned theme is automatically identified, and a label or a classification identifier is further constructed; automation of user behavior understanding, standardization of label generation and linkage of an operation process and a business system are realized, and personalized service response capability and operation efficiency are improved.

The invention belongs to the technical field of abnormal behavior person recognition, and particularly relates to an abnormal behavior person recognition method based on multi-source monitoring image integration. The method comprises the steps of obtaining original video data of a plurality of cameras in a target area; judging whether a person completely appears or partially disappears through an attitude estimation and semantic segmentation model, carrying out inter-frame sampling by taking the judgment as a starting point and a stopping point, and retaining time-space information; preprocessing the image frames, and extracting static posture and dynamic behavior characteristics; fusing dynamic behaviors, appearance and position information to realize cross-camera personnel identity association and construct a behavior track sequence; and matching the trajectory with a preset template, and calculating a deviation score in combination with time consistency, position path similarity, an action matching degree and an abnormal fragment confidence aggregation value to recognize abnormal personnel. The method breaks through the limitation of a single visual angle, improves the low-illumination recognition precision, solves the problem of cross-visual-angle identity breakage, and improves the recognition accuracy and traceability of abnormal behaviors.

The invention relates to the technical field of detection and identification, and discloses a mobile ship dynamic tracking and locking method and system based on target detection and identification, and the method comprises the steps: constructing a manifold embedded network, and mapping the multi-modal features of a ship to a Riemannian manifold space; constructing a continuous evolution modeler of the Shenchang differential equation model, and predicting a continuous evolution trajectory of the features under time, view angle and scale changes; a multi-scale bridging network is constructed, and bidirectional conversion between long and short distance feature representations is realized; constructing a differential geometric attitude encoder, and representing a ship attitude in an SO (3) Lie group space; a feature memory and reconstruction mechanism is realized, and the problem of short-time disappearance target recovery is solved; the components are integrated to construct a unified tracking and locking system. According to the method, the problems of discretization processing limitation, feature representation fragmentation, cross-condition consistency deficiency and the like in the prior art are solved, remarkable technical effects are achieved in the aspects of cross-condition recognition capability, continuous feature evolution capability, short-time disappearance target recovery capability and the like, and the method is suitable for the scenes of Yangtze River shipping monitoring, maritime affair safety supervision and the like.