24results about How to "Achieve high-precision recognition" patented technology

The application provides a radar working mode recognition method, device, equipment, medium and program product, and relates to the technical field of artificial intelligence. The method comprises the following steps: acquiring a pulse description word (PDW) sequence of a radar emitter; determining an intra-pulse feature of the radar emitter and an inter-pulse feature of the radar emitter based on the PDW sequence; inputting the intra-pulse feature and the inter-pulse feature into a working mode feature extraction model to obtain an intra-pulse feature vector and an inter-pulse feature vector output by the working mode feature extraction model; and inputting a feature fusion vector of the intra-pulse feature vector and the inter-pulse feature vector into a radar working mode recognition model to obtain a radar working mode recognition result output by the radar working mode recognition model. The intra-pulse feature and the inter-pulse feature are obtained first, and then the intra-pulse feature and the inter-pulse feature are subjected to feature extraction to obtain the intra-pulse feature vector and the inter-pulse feature vector with context information, so that high-precision recognition of the radar working mode can be realized for each scene.

The application discloses a kind of power forest fire intelligent monitoring method, device and equipment based on through remote fusion, and storage medium, it is related to forest fire intelligent monitoring technical field, comprising: obtaining the position information of first monitoring node, and obtaining the temperature and humidity data and smoke concentration data corresponding to position information as environmental data;Position information and environmental data are input into least square support vector machine model for processing, and fire risk probability is obtained;When fire risk probability is greater than preset risk threshold, position information, environmental data and fire risk probability are combined to obtain fire preliminary screening information, and fire preliminary screening information is uploaded to cloud server, so that cloud server fuses satellite remote sensing data to complete fire confirmation, realize the non-blind area monitoring coverage of power forest fire along transmission line, realize the preferential transmission of key fire data, reduce communication resource waste, improve system response speed and transmission reliability of key data.

The invention discloses a navigation mark visual identification and collision early warning method and system based on deep learning, and relates to the technical field of intelligent shipping, and the method comprises the steps: collecting a navigation channel monitoring video stream in real time, and carrying out the preprocessing of a key image frame; constructing a navigation mark detection model based on an improved Officient Det network, and outputting the position and category of the navigation mark in the key image frame; the method comprises the following steps: constructing a ship track prediction model through ship historical track data, predicting a short-time track of a ship, calculating the relative position, speed and course angle of the ship and a navigation mark, and constructing a dynamic collision risk field; and fusing the visual detection result, the radar ranging data and the AIS information to obtain a ship collision risk, and triggering graded early warning based on a preset risk threshold. According to the method, the improved deep learning model and the real-time calculation framework are combined, the robustness of navigation mark identification is improved, an efficient collision early warning mechanism is established, and technical support is provided for intelligent shipping.

The application discloses a vehicle cleanliness detection method, device and computer readable storage medium, the vehicle cleanliness detection method comprises: acquiring a to-be-detected vehicle image; segmenting the to-be-detected vehicle image to obtain a plurality of vehicle component images, the vehicle component image being an image in which a key component of a vehicle in the to-be-detected vehicle image is located; extracting features of the vehicle component images by using a feature extraction model in a cleanliness identification model to obtain global features and local features; fusing the global features and the local features by using a cleanliness prediction model in the cleanliness identification model to obtain fused features, and predicting the fused features to obtain a cleanliness score of each vehicle component image. In the foregoing manner, the application can improve the accuracy of vehicle cleanliness detection.

The present application relates to a kind of based on multi-modal fusion's weld magnetic detection system and method, system includes: multi-modal sensor component, for obtaining the two-dimensional image data and three-dimensional point cloud data of weld;Industrial control module, for according to two-dimensional image data and three-dimensional point cloud data, using instance segmentation model and three-dimensional point cloud feature extraction network to identify weld area, and extract weld three-dimensional structure information, according to weld three-dimensional structure information to generate detection path, based on detection path to carry out weld defect analysis and positioning;Mechanical arm actuating mechanism, for connecting detection probe, drive detection probe to move according to detection path, and detection probe real-time acquisition weld detection signal and transmission to industrial control module.The present application realizes the full process automation of weld identification, three-dimensional modeling, detection path generation and defect identification, compared with prior art has significant technical advantage and practical application value.

The application discloses a kind of agricultural laser weeding safety decision method and system, the method in which includes: obtaining the hyperspectral imaging data of weeding target area, ecological environment data and plant three-dimensional point cloud data;Based on hyperspectral imaging data, weed identification is carried out using pre-trained plant phenotype classifier;Based on hyperspectral imaging data, ecological environment data and plant three-dimensional point cloud data, the minimum lethal energy of laser for killing weeds is calculated using pre-trained energy optimization model, and laser emission parameter result is generated according to the minimum lethal energy of laser;Based on plant three-dimensional point cloud data, plant three-dimensional model is constructed, and weed positioning result and operation path planning result are generated;Operation equipment positions laser emission end to target weed according to positioning result and operation path planning result, and controls laser emission end to emit laser corresponding to laser emission parameter result.The method in the application realizes the synchronization of weeding efficiency, crop protection and ecological safety.

This application discloses a method, device, and storage medium for air gesture recognition, relating to the field of gesture recognition technology. The method includes: acquiring reflected signals after a hand gesture reflects a target acoustic signal through at least one microphone in a wearable device, wherein the target acoustic signal is constructed from a signal emitted by at least one speaker in the wearable device; performing frequency domain processing on each reflected signal to obtain target time-series spectrum information; performing differential processing on each target time-series spectrum information to obtain differential spectrum information; extracting potential gesture segments based on each differential spectrum information; inputting each potential gesture segment into a pre-constructed gesture recognition model, and outputting gesture recognition results. This application can improve the accuracy of gesture recognition during interaction.

The invention discloses an ionized layer electron density profile abnormal fluctuation detection method, system and equipment. The method comprises the following steps: calculating a relative electron density change rate between adjacent height points of each effective electron density profile; the height axis is divided into a plurality of intervals, for each effective electron density profile, the robust percentile index of the relative electron density change rate is counted in each height interval, and a noise upper limit threshold value, a sawtooth amplitude threshold value and an extreme sudden change threshold value are generated; formulating a three-level progressive judgment process comprising a high noise background criterion, an extreme single-point jump criterion and an effective sawtooth structure criterion based on a noise upper limit threshold, an extreme sudden change threshold and a sawtooth amplitude threshold; and if any criterion is triggered, determining that the electron density profile is invalid. According to the method, non-physical sawtooth oscillation and extreme jump can be efficiently and accurately identified, and the quality control capability of data before ionosphere modeling is remarkably improved.

The embodiment of the invention provides a product disassembly and repair detection method and device, equipment, a storage medium and a program product. The method comprises the following steps: acquiring an appearance quality inspection report and an X-ray image of an internal structure of a to-be-detected product; performing multi-modal anomaly detection processing on the appearance quality inspection report and the X-ray image to obtain a detection result of the to-be-detected product; wherein the detection result comprises the maintenance processing condition of the to-be-detected product; and if it is determined that the detection result represents that the to-be-detected product has been repaired, performing deep repair processing on the to-be-detected product. The method can improve the efficiency and accuracy of product maintenance detection.

The present application relates to a safety helmet wearing recognition detection method and system, an electronic device and a medium, and belongs to the technical field of oil and gas station recognition. The method comprises model establishment and recognition detection of on-site safety helmet wearing images. The model establishment comprises collecting on-site safety helmet wearing images, labeling and dividing the collected on-site safety helmet wearing images into a data set, establishing a safety helmet target recognition model based on the divided data set, and establishing a personnel posture estimation model based on the divided data set. The present application utilizes the automatic recognition function of the safety helmet target recognition model and the recognition function of the personnel posture estimation model on human body posture, realizes high-precision recognition of on-site safety helmet wearing, solves the problems of low monitoring efficiency and high labor cost of artificial management, solves the problem of low detection robustness of the existing safety helmet recognition model in the application scenario of oil and gas stations, and improves the accuracy of recognition.

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 provides a farmland surface disturbance and desertification monitoring method and system based on multi-modal remote sensing, and the method comprises the steps: obtaining a remote sensing image, carrying out the preprocessing of the remote sensing image, dividing the obtained multi-modal dual-temporal remote sensing image, constructing a Token sequence, inputting the Token sequence into a Transform backbone network, and carrying out the feature extraction, and obtaining an optical modal feature and an SAR modal feature; a low-rank adapter is introduced into an attention module of the network, gradient contribution is regulated and controlled based on modal deviation fractions, and cross-time-phase differential change features are constructed and fused to obtain fusion features; and generating a change probability graph, applying an optical and SAR prediction result consistency constraint, and outputting a final change detection result. According to the method, the fusion balance of optical and SAR features is remarkably improved through a modal depolarization LoRA fine tuning mechanism, so that weak modal information is fully expressed; and meanwhile, a cross-modal consistency constraint is introduced to ensure that a change detection result has consistency and credibility.

This invention discloses a pure vision-based 3D reconstruction and adaptive completion method in a collapse environment, belonging to the field of computer vision and emergency rescue 3D reconstruction technology. The method completes initial 3D reconstruction by acquiring data through binocular vision, constructs an anisotropic point cloud density distribution field to achieve accurate detection of low-density blind spots, establishes safe and reachable spatial constraints based on an implicit distance field, and optimizes the completion viewpoint by fusing an observability model. Map iteration is completed through local reconstruction and point cloud updates, and a closed-loop completion mechanism is formed using point cloud density as feedback until the reconstructed model meets the integrity requirements. This invention solves the problems of permanent voids in reconstruction caused by viewpoint blind spots, poor consistency between completion results and the real scene, and insufficient safety in the reconstruction process in existing technologies in collapse environments. It can achieve complete and high-precision autonomous reconstruction of 3D models of collapse disaster sites, providing reliable spatial data support for emergency rescue decision-making.

The application relates to a pantograph-catenary arc detection method and device, electronic equipment and a storage medium. The method comprises the following steps: acquiring current information in the process of train contact current collection with a pantograph-catenary through a current sensor arranged in a power supply circuit at the rear end of a pantograph insulator; acquiring arc radiation information in the process of train contact current collection with a pantograph-catenary through an antenna arranged at the top of a pantograph frame; decomposing the current information and the arc radiation information to obtain corresponding intrinsic mode function components and corresponding residual components; dividing each intrinsic mode function component into a plurality of time windows, and extracting time-frequency domain characteristic values of each time window; inputting the time-frequency domain characteristic values into a pre-established pantograph-catenary arc detection model, and outputting a detection result of whether an arc exists in the time window by using the pantograph-catenary arc detection model. The implementation of the application can improve the safety of train operation.

This invention belongs to the field of satellite remote sensing and red tide detection technology, and discloses an integrated method and system for generating and identifying broadband 3D samples of dominant red tide species. The method constructs an integrated model for generating broadband 3D red tide samples and identifying dominant species based on deep learning. It utilizes a red tide image generation module based on spectral feature transfer to obtain expanded samples of different dominant red tide species. A red tide dominant species identification module is constructed by integrating multi-scale feature extraction and cross-spatial attention mechanisms. The module is trained using a student-teacher semi-supervised learning model. Experimental results show that the model has good adaptability and can effectively identify three dominant red tide species: *Noctiluca scintillans*, *Noctiluca chloroticus*, and *Hacochloa erythrophora*, with an overall identification accuracy of 94.48%, which is 3.52%–8.66% higher than other comparative methods, providing a basis for red tide prevention and management.

This invention discloses a method and system for measuring single-photon orbital angular momentum Bell states based on a diffractive optical neural network. Based on time-energy correlated photon pairs generated by a spontaneous parametric downconversion process, an interference light field is generated by passing a signal photon modulated into a single-photon orbital angular momentum Bell state through a Bell state evolution system. A diffractive optical processor is constructed using a diffractive optical neural network to map the interference light fields of different Bell states into Gaussian spots in different detection regions within the output plane. Fast and high-precision identification of single-photon orbital angular momentum Bell states is achieved by correlating the imaging results with an enhanced charge-coupled device (CCD) camera. The proposed Bell state measurement method possesses a signal-to-noise ratio and anti-interference capability exceeding the classical limit, maintaining high-precision Bell state identification capability even under high background noise. It exhibits low energy consumption and high efficiency, meeting the needs of green computing and sustainable development. It can be applied to quantum state measurement in the field of quantum information processing.

The invention discloses a converter transformer saturation protection optimization method based on L-M algorithm neural network fitting, and the method comprises the steps: initializing system parameters after a protection device is started; the position state of a circuit breaker is monitored in real time, and after a closing signal is detected, instantaneous values of three-phase current and neutral current of the valve side of the converter transformer are collected; performing digital filtering processing on the acquired three-phase current data, and extracting a fundamental component; then calculating a peak value curve of a fundamental component of each phase, and identifying a fastest attenuation phase; recording the peak value of the fundamental component and the time required for attenuating from the peak value to 5%; taking the characteristic quantity data set as input and output, and training an L-M algorithm neural network model based on the data set; and according to an input characteristic quantity data set, obtaining an output value through a neural network, querying an inverse time limit characteristic curve, and delaying a corresponding operation time to perform a protection action. According to the method, the magnetizing inrush current and the direct-current magnetic bias are accurately distinguished by quantifying the direct-current magnetic bias, so that the reliability of protection action is improved.

A distributed optical fiber drainage pipeline leakage event identification method based on a distillation visual converter comprises the following steps that 1, a DAS system is prepared, and leakage data are collected; step 2, signal preprocessing and feature map construction; 3, designing and initializing a teacher model network; 4, inputting sufficient label data to train and verify the teacher model; step 5, constructing a DeiT student model; and step 6, carrying out distillation training on the DeiT student model based on the teacher model. According to the method, a distillation visual converter structure and a training mechanism are introduced and improved, and the characterization and classification capability of leakage event characteristics is remarkably improved under the condition of limited labeled samples, so that the recognition accuracy of tiny leakage is improved, and the false alarm rate in a complex noise environment is reduced.

The invention belongs to the technical field of unmanned vehicle environment perception and obstacle avoidance, and discloses an unmanned vehicle transparent obstacle visual touch fusion perception and obstacle avoidance method, and the method comprises the steps: carrying out the transparent object edge detection of an image collected by a vehicle-mounted around-view camera through an improved Transform semantic segmentation network, and outputting an initial position coordinate and a credibility score; a tactile verification and enhancement module is built, flexible multi-mode tactile sensors fusing triboelectric sensing and visual tactile sensing are deployed on the edges of a bumper and a hub of the unmanned vehicle, and close tactile features are generated in the non-contact stage; when the visual credibility is relatively high, the visual data is dominated, the tactile data assists correction, and when the visual credibility is relatively low, a tactile signal dominant mode is triggered; and generating and executing an obstacle avoidance path. According to the invention, the detection precision of the unmanned vehicle on transparent obstacles in extreme scenes such as urban roads, indoor venues, rainy days and nights is obviously improved, the collision risk is reduced, and the unmanned vehicle is adapted to complex moving working condition requirements.

The application discloses a pet and shadow distinguishing false alarm suppression method of a security camera, belongs to the technical field of intelligent security, and is based on an improved lightweight deep learning framework to construct a multi-target fusion recognition model, synchronously extracts the contour, texture, motion trajectory and gray value distribution multi-dimensional features of a human body, a pet and a shadow, innovatively adopts double judgment logic of "shadow followability + pet feature correlation" to perform secondary verification on a classification result, realizes accurate classification of three types of targets and automatic filtering of non-threat events, and solves the technical short board of traditional mobile detection technology that only captures dynamics and cannot identify targets, so that the invalid alarm rate in a household scene can be greatly reduced, the alarm accuracy is effectively improved, the model is small in size and fast in edge running speed, function deployment can be realized through OTA upgrading without replacing hardware, is compatible with most mainstream household security cameras in the market, and does not increase the existing hardware load in running power consumption.

The invention relates to the technical field of mineral separation, and discloses an underground in-situ efficient intelligent fine separation device and method for coal and gangue, and the device comprises a multi-layer screen, a material conveying system, a coal and gangue recognition system, a multi-stage separation system and a chute system. The multi-layer screen is used for grading the raw coal according to the size fraction through a first-stage screen mesh, a second-stage screen mesh and a third-stage screen mesh of which the sizes of screen holes are sequentially reduced; the material conveying system conveys screened materials through all levels of belts and is provided with a gangue returning belt and a well outlet conveying belt. The coal and gangue recognition system adopts image recognition, ray recognition and the like to adapt to recognition of materials of different size fractions; the multi-stage sorting system adopts a mechanical sorting device and an air-flowing sorting device which are respectively adaptive to materials with different particle sizes for sorting, so that low energy consumption and accurate sorting are realized; the chute system ensures efficient collection of coal and gangue. Through multi-stage screening, intelligent recognition and self-adaptive sorting, the problem of energy consumption caused by back-and-forth transportation of gangue in the prior art is solved, and efficient and intelligent sorting of underground coal gangue is achieved.

The application discloses an overhead power transmission line joint pipe identification and positioning method, which comprises collecting joint pipe images, expanding the number of target samples, labeling the joint pipe and making a joint pipe rotation target data set; a joint pipe rotation target detection model is constructed, a model rotation boundary box loss function is improved, a feature extraction and fusion network is optimized, an anchor frame parameter is adjusted in combination with a K-means clustering algorithm; a rotation target algorithm identifies an RGB image, outputs joint pipe target confidence and a rotation frame information, and maps the information to a joint pipe depth map; a joint pipe center point pixel coordinate depth value is extracted, a joint pipe physical width and height are estimated according to a pixel coordinate and a world coordinate corresponding relationship, and a center line on a depth point through a joint pipe center point and parallel to a long side is linearly fitted, and a joint pipe is reconstructed in a three-dimensional depth space. The application can provide a terminal joint pipe autonomous identification and accurate positioning reference scheme for a multi-rotor unmanned aerial vehicle carrying a digital X-ray imaging device to implement X-ray photographing detection on the joint pipe.

The application provides a half-height-width region identification method and system based on linear interpolation. The method comprises the following steps: obtaining two-dimensional topography data of a target structure prepared by a scanning probe nanometer direct writing; extracting a height data sequence of a single-row topography from the two-dimensional topography data; identifying key feature points in the height data sequence; calculating left half-height position height values and right half-height position height values based on the key feature points; determining left half-height-width boundary positions corresponding to the left half-height position height values and right half-height-width boundary positions corresponding to the right half-height position height values by using a linear interpolation method; and taking a difference value between the left half-height-width boundary positions and the right half-height-width boundary positions as a half-height-width value corresponding to the current single-row topography. The technical scheme provided by the application realizes automatic and high-precision identification of nanometer structure feature sizes, and significantly improves the measurement efficiency and the objectivity of process analysis.

The application provides a ship type identification method and device based on motion characteristics of a track point set, which comprises the following steps: collecting track point data of a ship by a ship positioning device, and preprocessing the track point data; constructing a track point set according to the preprocessed track point data, and extracting motion characteristics of the ship from the track point set; and identifying the ship type by using a deep learning algorithm according to the motion characteristics of the ship, and obtaining the ship type. The application has high accuracy and stability, and has good expansibility.