21results about How to "Real-time recognition" patented technology

The invention discloses a plant disease and insect pest detection device and method based on unmanned aerial vehicle high-altitude texture imaging, and belongs to the technical field of intelligent agriculture. The plant disease and insect pest detection device comprises an unmanned aerial vehicle body, a holder, a high-altitude texture camera and a small host; the high-altitude texture camera is arranged below the holder, the holder is arranged below the small host, the small host is arranged below the unmanned aerial vehicle main body, the unmanned aerial vehicle main body is provided with a front high-definition camera, a battery and an illumination sensor, and the high-altitude texture camera is wirelessly provided with a handle and a flight remote controller through Bluetooth; according to the plant disease and insect pest detection method, high-precision recognition of disease and insect pests is achieved on the basis of high-altitude texture feature fusion in combination with an environment perception driven data enhancement module, a self-adaptive multi-scale image convolution feature extraction and fusion module and a disease and insect pest area attention fusion module. The technology effectively supports precise detection of diseases and insect pests in agricultural production, and provides a scientific basis for agricultural management and pesticide application.

The invention provides a track fastener state recognition method based on a three-section type parallel network architecture. The method comprises the steps that a track inspection image is acquired; constructing a convolutional neural network (CNN) backbone network, and performing multi-scale feature extraction on the image of the track fastener; performing convolution processing and up-sampling processing on the fastener features by using the local feature branches to obtain local fastener features; performing global average pooling and convolution processing on the fastener features by using the global feature branches; the method comprises the following steps of: processing the characteristics of the fastener by using a mask generation branch of a framework based on Vision Transform, and generating a prototype mask of the track fastener; local, global and mask fastener features are fused; performing non-maximum suppression and mask generation on the fastener feature map; and analyzing the health state of the track fastener. According to the method, the CNN backbone network and the Vision Transform are combined, a new three-section parallel network architecture is provided, accurate identification and state analysis of the track fastener are realized through instance segmentation, and the identification accuracy in a complex environment is improved.

The invention belongs to the technical field of unmanned aerial vehicle intelligent traffic, and discloses a road event identification method and system based on an unmanned aerial vehicle real-time video stream. The method comprises the following steps: marking vehicles in collected unmanned aerial vehicle images, and training an accumulated unmanned aerial vehicle vehicle marking data set by using a YOLOv8 algorithm; the trained YOLOv8 model is applied to a real-time video collected by an unmanned aerial vehicle for vehicle identification, and position information of a vehicle in an identification video frame is intercepted; determining a driving area and a driving direction of the vehicle based on vehicle movement data in two unmanned aerial vehicle pictures which are associated front and back in the video; the direction vector of the inspection center line is set as the forward advancing direction of the unmanned aerial vehicle, the included angle between the direction vectors of the vehicle driving on the left lane and the right lane is calculated in the same coordinate system, and the vehicle on the lanes is judged to move reversely according to the included angle. Compared with traditional ground individual road section arrangement monitoring and manual checking, the monitoring efficiency is greatly improved.

The invention discloses a survey method and system for ocean subsurface vortex and a sub-mesoscale process thereof, and the method comprises the steps: obtaining three-dimensional temperature-salinity-depth flow data of a designated sea area, and drawing a vertical distribution diagram of temperature, salinity, density and a flow field; screening an equal-density surface according to the vertical density abnormal structure and the maximum flow velocity occurrence position; identifying the secondary surface vortex based on the determined potential height contour map and the vertical flow field map of the equal-density surface; acquiring satellite altimeter data of a specified sea area and drawing satellite orbit distribution diagrams of different dates; according to the identified subsurface vortex and the satellite orbit distribution map, carrying out route design and drawing a route; and drawing a site distribution map for hydrological survey according to the route distribution map. The invention aims to provide a scheme for identifying and positioning the subsurface layer vortex, investigating and analyzing the subsurface layer vortex sub-mesoscale process and designing an observation route aiming at the ocean subsurface layer vortex so as to solve the problem that the ocean subsurface layer vortex and the sub-mesoscale process thereof are difficult to position and observe.

The application discloses a ramjet lightweight flow field reconstruction method guided by multi-scale feature mapping, and belongs to the technical field of engine combustion monitoring and artificial intelligence application, and the method comprises the following steps: S1, acquiring wall surface pressure time sequence signals and schlieren images, and constructing a combustion flow field reconstruction data set under different inflow conditions; S2, performing data preprocessing on the combustion flow field reconstruction data set to obtain a preprocessed combustion flow field reconstruction data set; S3, inputting the preprocessed combustion flow field reconstruction data set into a space-time multi-scale feature alignment and fusion model to obtain a lightweight flow field reconstruction model; and S4, real-time receiving wall surface pressure time sequence signals and inputting the wall surface pressure time sequence signals into the lightweight flow field reconstruction model to reconstruct a high-time-resolution two-dimensional transient flow field. The method realizes high-precision and high-efficiency reconstruction of extreme combustion non-steady-state multi-physical fields under high-speed flight conditions, and solves the problems of data sparseness and inability to perceive combustion characteristics at future time points in advance in the traditional method.

The invention discloses a battery screening method, electronic equipment and a storage medium. The battery screening method comprises the following steps: acquiring a plurality of test voltages of a plurality of to-be-tested batteries in a formation process; for any to-be-tested battery, determining a voltage deviation between each test voltage of the to-be-tested battery and a plurality of test voltages of other to-be-tested batteries based on the plurality of test voltages of the to-be-tested battery and the plurality of test voltages of the other to-be-tested batteries; and based on the voltage deviation corresponding to each test voltage of the plurality of to-be-tested batteries, screening a target battery from the plurality of to-be-tested batteries. The accuracy of battery screening can be improved.

This invention provides an autonomous control method for a drone swarm based on VGG16 and virtual game theory. The drones utilize a pre-trained attitude recognition model to identify the historical movement information of their opponents and calculate the opponent's strategy tendency at the previous moment based on this information. Then, the drones select a candidate action from a set of potential actions, calculating the opponent's strategy tendency and probability of choosing that action according to the game theory formula. A reward function is designed, and the drones select their own action based on the obtained opponent's strategy probability using the optimal response decision rule, observing the changes in the opponent's actions until the entire drone swarm reaches a stable state. This enables the analysis and game theory of multiple surrounding drones. When encountering signal interference, this invention allows some drone units to malfunction while the drone itself can adjust its movement attitude, maintaining relative stability for the entire swarm.

The application discloses a kind of based on unmanned aerial vehicle's image recognition method, device, equipment, storage medium and product, relate to image recognition technical field, the method includes: obtaining the inspection image data collected by unmanned aerial vehicle;Inspection image data is input to lightweight identification model, and lightweight identification model is obtained by replacing the backbone network in standard target detection framework with lightweight backbone network and quantization compression;The position information of each target object in inspection image data is obtained by carrying out forward inference calculation to inspection image data by lightweight identification model;Image recognition result is generated according to the position information of each target object.Compared with the prior art, the application replaces cloud recognition by end-side lightweight model, realizes the real-time identification of large-scale image of unmanned aerial vehicle end side.

The application discloses a kind of fast moving face recognition method and system based on multi-frame image enhancement, comprising the following steps: through imaging device acquisition multi-frame image sequence, obtain multi-frame image after pre-processing;The multi-frame image after pre-processing is input into the improved MonST3R network, carries out spatial alignment and structure analysis, obtains the image after alignment and three-dimensional structure information;The above result is input into CRNet network, separates and enhances multi-scale feature, and obtains enhanced feature;Enhanced feature is input into space-time attention fusion module, and multi-frame feature representation is obtained by weighted fusion;Fusion feature is input into identification network, and dimension reduction is carried out, and face embedding feature vector is obtained;Similarity comparison is carried out between embedding feature and database template, and recognition result and confidence prompt are output.The application can realize efficient, accurate and resource-friendly face recognition under dynamic complex environment.

The invention discloses a double-wheel milling construction geology identification method based on a random forest and a support vector machine. The method comprises the following steps: S1, data preprocessing: data acquisition, sample marking, filtering processing and correlation analysis and feature selection; s2, geological recognition model training: S21, overburden and rock stratum recognition model training; s22, training a rock stratum identification model; during model training, taking a plurality of slot segment data in all slot segment data as a test set, and taking the data of the remaining slot segments as a training set; and S3, real-time geological recognition: after the model is obtained through training in the S2, real-time recognition of the residual groove section double-wheel milling grooving construction geology is carried out. According to the method, real-time recognition of the double-wheel milling construction geology can be realized, the geological judgment efficiency and accuracy are improved, and beneficial help is provided for quality improvement and efficiency improvement of underground diaphragm wall grooving construction.

The invention discloses a hydrogen damage test device and a detection method thereof.The hydrogen damage test device comprises an electrolyte container, an electrochemical workstation, an acoustic emission probe, an acoustic emission detector and a computer, and a metal sample, an auxiliary electrode and electrolyte are arranged in the electrolyte container; the metal sample and the auxiliary electrode are soaked in the electrolyte and are both connected with the electrochemical workstation, the non-hydrogen-filling surface of the metal sample is provided with the acoustic emission probe, the other surfaces of the metal sample except the hydrogen filling surface are sealed through resin, the acoustic emission probe is connected with the acoustic emission detector, and the electrochemical workstation and the acoustic emission detector are both connected with the computer. According to the method, acoustic emission signals are collected in real time in the electrochemical hydrogen charging process, synchronous analysis is conducted in combination with parameters of the hydrogen charging process, elastic wave signals generated by a local damage event can be captured at the moment of hydrogen-induced microcrack initiation, and therefore real-time recognition of the hydrogen crack initiation moment is achieved; and the accuracy and reliability of hydrogen damage early recognition are improved.

The invention discloses a welding penetration control method and system based on LabView, and relates to the technical field of welding process control. The method comprises the steps that a welding voltage signal in the welding process is collected; a LabView processor is used for conducting data format conversion, noise reduction processing and frequency domain analysis on the welding voltage signals, and the oscillation frequency of a molten pool is extracted; the oscillation frequency of the molten pool is compared with the critical penetration frequency in an expert database, and the penetration state of the current welding process is judged; and the welding system and / or the motion control system are / is controlled to execute corresponding control actions according to the judgment result, so that welding heat input is adjusted, and closed-loop control over the penetration state is achieved. The noise reduction treatment adopts a self-adaptive multi-scale correlation weighted generalized threshold wavelet noise reduction method, and the oscillation frequency of the molten pool is obtained through peak group proportion analysis. According to the method, online recognition and real-time regulation and control of the welding penetration state can be achieved, and the method has the advantages of being high in anti-interference capacity, stable in feature extraction and high in control precision.

The present application relates to the field of battery thermal safety performance test, specifically to a kind of energy storage device thermal safety performance test test device, including test box, controller, temperature detector and voltmeter, the controller, temperature detector and voltmeter are sequentially installed from top to bottom in test box front left side, and controller is electrically connected with temperature detector, heater electrically connected with controller is installed in the left and right inner walls of test box, controller controls the start-stop of heater, let heater heat test box, temperature measuring mechanism is installed in the top of test box inner chamber, and fire extinguishing mechanism is installed around the top of test box inner chamber.Effectively solve the technical pain points that existing test device cannot detect battery bulging, test dimension is single, fill the blank of structure failure early warning in battery thermal safety test, provide more comprehensive, rigorous data support for battery thermal safety performance evaluation, and then provide accurate direction for structure optimization and performance improvement of battery.

The invention relates to the technical field of quality evaluation, in particular to an animal product quality safety state intelligent evaluation method and system, and the method comprises the following steps: collecting a plurality of indexes to construct a monitoring matrix, generating an offset trend and an abnormal path, evaluating the offset grade distribution, judging the quality safety state, and outputting an animal product quality safety state intelligent evaluation result. According to the method, multi-point parameter synchronous acquisition is carried out on animal products in key links of cold chain transportation, processing and sorting, dynamic segmentation analysis is carried out in combination with trend change of an index offset rate, tracking and difference calculation of a continuous offset phenomenon are realized, and the accuracy of the continuous offset phenomenon is improved on the basis of offset intensity grading division and quantitative proportion statistics. According to the method, the quality states of the whole animal products are classified and collected, the quality state grades are determined on the basis of an offset grade distribution structure, so that the quality fluctuation of the animal products is recognized in real time, and the continuity, sensitivity and timeliness of quality evaluation of the animal products are effectively improved.

The present invention relates to the technical field of gangue classification and identification, and in particular to a gangue identification method and system based on hyperspectral imaging and laser ranging, the method comprising: mapping hyperspectral image data onto a three-dimensional point cloud of laser ranging data to obtain a mapping result; using a trained network model to extract multi-scale spatial features and multi-scale spectral features from the mapping result; fusing the extracted features, and then generating candidate target areas based on the fused features; classifying the candidate target areas to obtain a classification result of gangue; constructing a three-dimensional model of the coal surface based on laser ranging data; generating a triangular mesh model for the three-dimensional model of the coal surface; smoothing the triangular mesh model to obtain a reconstructed model; determining the particle size of the coal block based on the reconstructed model; and identifying the petrographic features of the coal based on the spectral data to distinguish different types of coal. The accuracy and efficiency of gangue sorting are improved.

The invention discloses a product quality inspection management system and method, and belongs to the field of product quality inspection and control. The system comprises the following modules: a network topology construction module which dynamically constructs a network topology structure based on multi-source inspection data and realizes time-space correlation analysis of a quality state; the anomaly detection module is used for identifying an abnormal node and positioning a risk source through an anomaly detection algorithm; the influence degree calculation module responds to the anomaly identification output by the anomaly detection module and quantifies the diffusion range and intensity of the anomaly association relationship in the whole quality analysis system based on a propagation model; the evaluation index module is used for constructing a multi-dimensional quality evaluation index system and calculating the relative weight of each index by adopting an analytic hierarchy process; and the early warning monitoring module monitors the change of the network topology structure and an influence degree calculation result in real time, identifies abnormal association of different severity degrees in combination with a graded early warning threshold value, and starts a differentiated dynamic monitoring and handling mechanism.

The embodiments of the present disclosure disclose a method, device, equipment and readable medium for identifying live broadcast behaviors. The method includes: constructing facial action features in the current live broadcast frame based on the target key points detected in the current live broadcast frame; identifying the teaching behaviors in the current live broadcast frame from the constructed teaching behavior library based on the position deviations between the target key points and the facial action features. The technical solution provided by the embodiments of the present disclosure realizes real-time recognition of live broadcast behaviors and improves the timeliness of live broadcast behavior recognition; at the same time, various teaching behaviors that do not meet the requirements of teaching specifications can be flexibly configured in the teaching behavior library to realize customized recognition of live broadcast behaviors and improve the flexibility and comprehensiveness of live broadcast behavior recognition.

The invention relates to a lemon yellow vein disease recognition method based on a hyperspectral image and a lightweight neural network, and the method comprises the steps: collecting lemon leaf hyperspectral images with different wavelengths, processing the lemon leaf hyperspectral images with different wavelengths, inputting the processed leaf hyperspectral images into a 3D-2D-LcNet model, and carrying out the recognition of the lemon yellow vein disease based on the 3D-2D-LcNet model. Obtaining a lemon yellow vein disease recognition result; wherein the 3D-2D-LcNet model is obtained by combining a lightweight neural network ShuffleNetV2 with a Hybrid SN (Software Node), and the 3D-2D-LcNet model is obtained by combining the lightweight neural network ShuffleNetV2 with the Hybrid SN. According to the invention, the lemon yellow vein disease can be identified with high efficiency and high accuracy.

The invention relates to the technical field of gas pipe network monitoring, and discloses a natural gas pipeline full-life-cycle risk early warning method based on data fusion, which comprises the following steps: step 1, full-life-cycle multi-source data acquisition; 2, preprocessing edge node data; 3, cloud multi-modal data fusion is carried out; step 4, AI intelligent analysis and risk identification; step 5, risk grading and early warning generation; step 6, early warning pushing and work order dispatching; according to the method, through the full-chain design of full-cycle data penetration, edge, cloud collaboration, AI dynamic analysis and closed-loop optimization, the core pain points of lagging, low efficiency and insufficient precision of traditional natural gas pipeline risk early warning are solved, and real-time identification, accurate judgment, rapid disposal and continuous optimization of risks are achieved; and the intelligent level and the operation and maintenance efficiency of safety management of the natural gas pipeline are greatly improved.

The invention discloses an injection mold for cleaning electric appliance structural parts, and relates to the technical field of injection molds, the injection mold comprises an injection molding assembly, the injection molding assembly comprises a supporting base, a lower mold body is fixed to the top of the supporting base, a height adjusting part is fixed to the surface of the lower mold body, and an upper mold body is arranged at the top of the lower mold body; the height adjusting piece is fixed to one side of the upper die body, and a buffering flow guiding piece is fixed to the top of the lower die body. Through the collaborative design of the injection molding assembly and the auxiliary assembly, the defects that an existing injection mold for cleaning the electric appliance structural part is low in production efficiency and poor in defective product screening convenience are effectively overcome, on one hand, the mode of manually checking and screening defective products after injection molding is completed is abandoned, the operation process is simplified, and the labor intensity of workers is reduced; the problems of missing detection and false detection of defective products due to artificial visual fatigue and judgment deviation are avoided, and the production quality of the cleaning electric appliance structural parts is guaranteed.

The application provides a construction method and system of an abnormal action recognition model for stroke patients. The abnormal action patterns and kinematic characteristics of stroke patients and their actions in different rehabilitation stages are determined. Then, a neural network algorithm is used to construct an abnormal action recognition model suitable for stroke patients. The optimal model for different sitting Taijiquan action is selected by comparing the performance of each model, and it is applied to a traditional upper limb rehabilitation robot to realize real-time recognition of abnormal actions, promote the learning of normal upper limb movement patterns by patients, and improve the upper limb movement function of the patients.