101results about How to "Fast reasoning" patented technology

The invention discloses a PPY-YOLO-based steel surface defect detection method and a PPY-YOLO-based steel surface defect detection system. The PPY-YOLO-based steel surface defect detection method comprises the following steps: step 1, preprocessing a steel surface defect image; 2, a neck structure Faster-Neck, a deep learning module C3K2Faster and a spatial pyramid pooling module SPPFSim are introduced into the YOLOv11 network, and a PPY-YOLO target detection model is obtained; step 3, training a PPY-YOLO target detection model; and 4, acquiring a steel image, and inputting the steel image into the trained PPY-YOLO target detection model to obtain the category and positioning information of the steel surface defects. A new model is constructed, the backbone network and the neck network are improved, the backbone network is formed by connecting the input end of the C2PSA module with the output end of the SPPFSim module, and an up-sampling group and a down-sampling group are added to the neck network, so that the model has a relatively good detection effect.

The invention discloses a micro-expression recognition method and system based on a state space model and double-flow fusion, and relates to the technical field of expression recognition, and the method comprises the steps: extracting a starting frame image and a top frame image of a micro-expression continuous video frame sequence; inputting the initial frame image and the top frame image into a micro-expression recognition model based on a double-flow fusion network, extracting an optical flow image, and performing face key point detection and clipping operation on the top frame image and the optical flow image respectively; carrying out image block embedding operation to obtain texture features of the top frame image after cutting and motion features of the optical flow image; the two features learn the global dependency relationship and the local correlation between the image blocks from the positive direction and the negative direction through a parallel multi-layer bidirectional Mama fusion module, and deep image features and motion features are extracted after feature fusion; and the deep features of the two modes are interacted through a double-flow cross attention classification module, comprehensive features are learned, and a plurality of classifiers are combined to identify and output the micro-expression type more accurately.

The invention provides a data center machine room equipment indicating lamp identification method, a terminal and a storage medium, and relates to the technical field of machine room management, and the method comprises the steps: obtaining real-time image information containing all equipment indicating lamps in a machine room; inputting the real-time image information into a small target detection algorithm model, and obtaining the lighting state of each equipment indicating lamp; according to the lighting state of each equipment indicating lamp and a fault sensing information database, determining an equipment operation state so as to notify a machine room operator on duty of an equipment fault when the equipment operation state is a fault state; the small target detection algorithm model is obtained by training an improved RT-DETR algorithm model based on an enhanced historical image data set; the enhanced historical image data set comprises indicator light images of different sizes and indicator light images which are partially shielded by the shielding object. The small target identification efficiency can be improved, the indicating lamp can be accurately identified in a complex machine room environment with numerous devices and sheltering objects, and the operation state of the devices can be timely and effectively monitored.

The invention relates to the technical field of data assimilation, and discloses a data assimilation method, device, equipment and medium, the method comprises the following steps: firstly obtaining observation data and meshing, the observation data comprising at least one variable; or, a background field of at least one time and corresponding observation data are obtained, the observation data are gridded, and the background field and the observation data both comprise at least one variable; next, the constructed data assimilation model is trained by using the grids corresponding to all the variables to obtain a trained data assimilation model; and inputting the to-be-processed gridding observation data or the to-be-processed background field and the gridding observation data into the trained data assimilation model, and outputting a corresponding analysis field. The data assimilation method, device, equipment and medium provided by the invention have the characteristics of high reasoning speed, high assimilation precision and the like, and are suitable for data assimilation in the fields of weather climate, ocean, land surface and the like.

The invention relates to the technical field of image target detection, and discloses a dark light environment small target detection method and system, and the method comprises the steps: collecting original image information; performing dark light enhancement on the original image; using the improved Zero-DCE network to carry out brightness adjustment on the image after the dark light enhancement; and adding a simAM attention mechanism in a backbone feature extraction module of the yov11, and inputting the enhanced image into the improved YOLOv11 network to obtain a small target detection result. A dark light enhancement training data set is not needed, the data acquisition difficulty is reduced, and the occurrence of dark light enhancement model overfitting in the training process is avoided. The model training speed and the reasoning speed are high, and the requirement for computing power resources of equipment is low. Each pixel of the image has a brightness enhancement curve of the data, and the curves are obtained in training, so that the enhanced image is smoother.

The invention discloses a 3D attitude estimation method based on a monocular camera, electronic equipment and a storage medium, and the method comprises the steps: firstly constructing a data set composed of real data and synthetic data, then detecting the positions of 2D skeleton key points for the data set through YoloPose, obtaining estimated 3D data through GHUM fitting, obtaining a 3D attitude data set of 3D attitude estimation, and carrying out the 3D attitude estimation of the 3D attitude data set. 3D dimension improvement is added into YoloPose to obtain a 3D attitude estimation model, and the 3D attitude estimation model is trained by the three-dimensional data set and is used for 3D attitude estimation of a two-dimensional image obtained by the monocular camera. The method is low in cost and easy to deploy, does not depend on additional sensors or equipment, can stably work in various illumination and complex environments, and can realize detection and attitude estimation through one deep learning model.

The invention discloses a monocular depth estimation method and device based on a bidirectional state space model, and a medium, relates to an image processing technology, and solves the problems that an existing self-supervision method is insufficient in the aspects of long sequence modeling, training speed, reasoning speed and scene complex detail capturing, and the optimization effect of the model under sparse supervision is poor. The monocular depth estimation method comprises the following steps: constructing an encoder-decoder structure which comprises a continuous expansion convolution module and a bidirectional state space module, generating intermediate features of a training image by an encoder, modeling long-distance intermediate feature information, and generating output features by a decoder; constructing an attitude estimation model, performing joint reconstruction by using output features generated by a decoder and depth estimation of a training image, modeling a learning target, and designing a loss function of a target image and a reconstructed image to optimize a network; and training the constructed model by adopting a public KITTI data set to obtain a weight model, and evaluating the precision of monocular image depth estimation by the trained model. The method is used for image monocular depth estimation.

Coronary artery disease is the most common disease in the cardiovascular field and is also one of important factors causing death in the global range, and by conducting blood vessel segmentation and stenosis detection on coronary angiography images, doctors can diagnose the coronary heart disease more accurately. The invention provides a brand new deep learning architecture SAM-VMNet, and the model combines the powerful feature extraction capability of MedSAM and the linear complexity advantage of a visual state space model of VM-UNet, so that the model has higher reasoning speed and stronger data processing capability than Transformer, and higher segmentation accuracy and stability of coronary angiography images are realized. Moreover, a'dynamic queue 'is used for capturing the narrow part of the blood vessel, and the narrow part is compared with the normal blood vessel diameter before and after the narrow part, so that quantitative calculation of the blood vessel stenosis degree is realized. The algorithm is high in operation speed, and full-automatic blood vessel contour extraction and stenosis degree detection can be achieved. The kit can be applied to the clinical diagnosis process of coronary heart disease.

The invention discloses a road scene depth completion method based on semantic information and color image guidance, and the method comprises the steps: designing a double-branch network composed of a color image guidance branch and a semantic guidance branch, introducing semantic information into the network, and carrying out the depth completion of a road scene. According to the method, a multi-task learning thought is utilized, a backbone is shared with a depth map prediction layer in a color image guide branch, and semantic information can be obtained only by adding a semantic segmentation layer without adding a whole network. Meanwhile, a semantic label generated by the semantic segmentation layer serves as input to enter a semantic guidance branch, so that parameters of the layer can be adaptively adjusted during network training. According to the method, the sparse depth map provided by the laser radar is supplemented by integrating high-precision semantic information and the RGB image, so that the accuracy and the real-time performance of scene understanding in applications such as digital twinning, virtual reality, digital capital construction and intelligent transportation are improved.

The invention relates to the technical field of knowledge distillation, and discloses an over-parameterization knowledge distillation method, device, equipment and medium, which can map a two-dimensional parameter matrix in an original student model from a two-dimensional tensor space to a high-dimensional tensor space to obtain a student model comprising a high-dimensional tensor, increase trainable parameters in the student model and improve the knowledge distillation efficiency. The knowledge capacity of the student model is improved, the student model comprising the high-dimensional tensor is trained, the high-dimensional tensor in the student model is updated until the training end condition is met, namely the training purpose is achieved, the high-order tensor in the student model after training is completed is reduced into a target two-dimensional parameter matrix, and the target two-dimensional parameter matrix is obtained; the shape of the target two-dimensional parameter matrix is the same as that of the two-dimensional parameter matrix in the original student model, and obtaining the student model with good knowledge distillation. According to the invention, the knowledge distillation effect can be enhanced under the condition of maintaining the parameter quantity during the deployment of the student model.

The invention provides a mechanical water meter automatic meter reading method based on a deep learning OCR model, and belongs to the technical field of water meter automatic meter reading, and the method comprises the following steps: collecting and marking mechanical water meter pictures, and forming a data set; a mechanical water meter character detection model, a mechanical water meter character direction judgment model and a mechanical water meter character recognition model are constructed, and the data set is used for training; obtaining a mechanical water meter picture needing to be read, sequentially inputting the mechanical water meter character detection model, the water meter character direction judgment model and the mechanical water meter character recognition model, and finally obtaining a meter reading number by determining a meter reading area position and a character direction. According to the method, automatic meter reading of the mechanical water meter is realized, the positioning precision of a meter reading area is improved, the processing speed is high, the problem of large identification error caused by diversity of shooting angles of compatible water meter pictures by identifying character directions is solved, a half-character mapping rule of a reading label is introduced, the interference of half characters on the identification precision of full characters is avoided, and the character identification precision is improved.

The invention discloses a pavement crack segmentation and quantification method based on pyramid region attention, and relates to the technical field of computer vision. The road maintenance mileage is increased year by year, the capital and human input required for maintaining the road network is quite huge, and the efficient, accurate and robust pavement crack segmentation and quantification method has important significance in monitoring the pavement damage condition and reducing the road maintenance expenditure. According to the method, the long-distance dependency relationship can be effectively captured while the low calculation cost is kept, the segmentation precision of the pavement crack is improved, the crack damage rate is calculated according to the segmentation result, a basis is provided for evaluation of the road damage condition, the pavement maintenance decision-making efficiency is improved, and the maintenance expenditure is reduced.

The invention relates to the technical field of dam underwater structure defect detection, in particular to a dam underwater structure defect real-time detection method and device under a weak computing power condition, and the method comprises the steps: obtaining dam underwater structure defect video data, and extracting defect images to construct a dam underwater structure defect data set; performing defect category and defect area information labeling on the extracted defect image, and dividing a labeled data set into a training set, a verification set and a test set; constructing a dam underwater structure defect detection model by taking YOLOv8n as a basic model, training and evaluating the dam underwater structure defect detection model by utilizing the training set and the verification set, and checking by utilizing the test set; and analyzing the dam underwater structure defect real-time image by using the dam underwater structure defect detection model to obtain the category and position information of the underwater structure defect. According to the method, the defect image of the underwater structure of the dam can be rapidly detected under the condition of weak computing power, and the cost and the safety risk of manual detection are greatly reduced.

The invention provides a model searching method for a sorting machine, a sorting method and electronic equipment, and belongs to the field of data processing.The model searching method comprises the steps that searching sampling is conducted on a preset searching space, and a neural network structure and a hyper-parameter and an optimization algorithm corresponding to the neural network structure are obtained; selecting candidate network structures from the neural network structures according to the scale values of the neural network structures; based on the hyper-parameters of the candidate network structures and an optimization algorithm, performing optimization evaluation on the candidate network structures by using the plurality of data sets of different materials to obtain performance indexes of the candidate network structures; and determining an optimal network structure from the candidate network structures according to the performance indexes. Therefore, the optimal network structure is a model which has small scale values such as calculated amount and parameter amount and has optimal recognition performance on the multiple materials of the sorting machine, so that the optimal network structure has wide adaptability and has higher recognition precision and higher reasoning speed.

According to the wheat leaf pest detection method based on YOLOv10, data processing and model architecture can be improved, the detection precision and speed can be improved, and the problems of data set limitation and insufficient model optimization in the prior art can be solved, so that the wheat pest recognition accuracy can be improved, and agricultural producers can be helped to monitor and prevent pests in real time. Compared with the prior art, the detection precision is high, the reasoning speed is high, the generalization ability is high, and computing resources are saved. According to the improved YOLOv10 model, a CoordAttention mechanism is introduced, so that the accuracy of disease and pest detection is remarkably improved. The model optimizes the reasoning process by removing the NMS, the reasoning speed of each image reaches 0.8 millisecond, and the real-time detection efficiency is greatly improved. Training is carried out in combination with multiple data sets, the model has good performance in different field environments, complex scenes of different illumination, angles, shielding and the like can be processed, and the applicability in actual production is improved. The optimization of the model enables the calculation resource consumption to be lower during reasoning.

The application provides a large model security management method, system, device and medium based on a core particle architecture, which can be applied to the technical field of semiconductor integrated circuits. The method comprises the following steps: setting a physically isolated special security core particle as a hardware trusted root in a heterogeneous integrated system, and connecting the special security core particle with at least one target core particle running a large model through a bus; deploying a lightweight security supervision model with a smaller parameter quantity than the large model in the special security core particle; collecting behavior characteristic data of the target core particle running the large model in real time; performing inference by using the lightweight security supervision model, calculating an abnormality metric between the current behavior of the target core particle and a preset normal behavior model; updating a dynamic trust state of the target core particle based on the abnormality metric; and performing a hardware-level security response operation by the special security core particle in response to the dynamic trust state reaching a predetermined threshold.

The present application relates to the field of digital image processing and information security, and discloses a document tampering detection method and system based on image data processing, comprising the following steps: first, extracting the noise residual and microscopic penetration characteristics of the document image, and constructing a physical potential energy field and a virtual viscous resistance field; then, using a Darcy law variant model for dynamic evolution, generating a virtual flow velocity vector field to simulate the sliding behavior of fluid in heterogeneous media; subsequently, constructing a heterogeneous graph based on the flow field divergence singular point and streamline trajectory, using a graph neural network to aggregate the node dynamics characteristics for deep reasoning, and finally generating a tampering positioning mask. The present application innovatively introduces fluid mechanics field theory, converts hidden static texture differences into significant dynamic flow field anomalies, solves the problem that the prior art is difficult to capture microscopic tampering traces, and significantly improves the detection accuracy and generalization ability in complex document scenarios.

The application discloses a single-frame far-field wavefront recovery method and system based on a non-rotationally symmetric special-shaped pupil, and belongs to the technical field of wavefront recovery. The method comprises the following steps: a non-rotationally symmetric special-shaped pupil mask is arranged on an entrance pupil surface of a to-be-detected optical system, so that the pupil of incident light has non-symmetry; after the incident light is modulated by the special-shaped pupil mask, the to-be-detected optical system is passed through to form a first single-frame far-field light spot image; the normalized first single-frame far-field light spot image is input into a trained FFT-UNet wavefront recovery network, the trained FFT-UNet wavefront recovery network outputs a first pupil surface wavefront phase image corresponding to the first single-frame far-field light spot image; and wavefront recovery is realized based on the first pupil surface wavefront phase image. The application encodes phase information of a to-be-detected wavefront into a non-symmetrical diffraction structure of a single-frame far-field light spot, and helps to alleviate the conjugate solution degeneracy problem in single-frame phase inversion.

The application discloses a kind of textile flaw detection methods based on machine vision, and the textile RGB image is obtained by image acquisition and preprocessing, and size adjustment, normalization operation is carried out, core detection process includes: feature extraction is carried out using ResNet50 network, and abnormal area recognition is realized in pixel level in combination with SPADE algorithm and k nearest neighbor, on this basis, using transfer learning strategy, specific textile flaw data set is fine-tuned and optimized, to build high-precision detection model, CFLOW-AD model is used for multi-scale feature extraction and the generation of abnormal score chart, accurately identify potential flaws, and integrated real-time feedback mechanism, system according to preliminary detection result, dynamically adjust detection parameter, and suspected area is re-detected, through TP, FP, ROC curve index carries out model evaluation, generates detailed flaw report, the method effectively improves the automation level and accuracy of textile flaw detection.

The invention provides an engine intake and exhaust system state evaluation method and system based on heterogeneous data. The method comprises the steps that a vibration signal and a thermotechnical signal under a diesel engine intake and exhaust system fault simulation test are acquired; performing wavelet transform on the vibration signal to obtain a combined wavelet feature, extracting a time domain feature of the vibration signal, and performing feature combination on the wavelet feature, the time domain feature and the thermotechnical signal to obtain a diesel engine motion state feature vector; and obtaining an optimized SVM model based on a particle swarm algorithm, and carrying out online state evaluation on the intake and exhaust system of the engine based on the SVM model. According to the method, the problems that the signal interference of a high-power multi-cylinder diesel engine is large and the accuracy rate of an existing intake and exhaust system state evaluation method taking a single information source as a diagnosis basis is low are solved to a certain extent, meanwhile, end-to-end intelligent online evaluation can be achieved, few hardware resources are consumed, and the method has good engineering application prospects.

The present application provides a model miniaturization method, device, electronic device and computer-readable storage medium, wherein the method comprises: obtaining online application data generated by the teacher model to be miniaturized in the process of performing a variety of different segmentation tasks; wherein the online application data includes the input data and output data of the teacher model; determining the online application data generated by the teacher model in the process of performing the same segmentation task as the data set corresponding to the segmentation task; for each segmentation task, using the online application data included in the data set corresponding to the segmentation task to train the initial model, and obtaining a student model used only for performing the segmentation task. This method is conducive to ensuring the reasoning accuracy of the miniaturized student model.

The application discloses a dialect self-adaptive learning method and system, comprising a structured dialect domestication interface, a speech intent recognition unit, a grammar framework domestication unit, a cross-dialect analogy reasoning engine, a dialect-scene-customs linkage memory module and a distributed dialect knowledge federation network; the robot can continuously update dialect recognition capability through daily interaction without intervention of manufacturers. Users can teach the robot to learn dialects in a natural way such as demonstration, error correction and praise, just like teaching children to speak. The system can continuously optimize the model according to the feedback of users, and adapt to the language habits and dialect variants of different users.

A lightweight image super-resolution method based on dynamic reparameterization includes: S1 inputting a preprocessed low-resolution original image into a CNN-based lightweight image super-resolution backbone network and image reconstruction module to obtain a super-resolution image; S2 calculating a loss function based on the super-resolution image and the corresponding high-resolution original image after preprocessing, and updating the network weights through backpropagation; S3 expanding the convolutional layer with the largest contribution score in the iteration cycle into an enhanced edge-oriented reparameterization branch module according to the iteration cycle and initialization conditions, and pruning all reparameterization branches with the smallest contribution score that have been expanded into reparameterization branch modules in the iteration cycle; repeating steps S1-S3 until the backbone network converges. This method improves reconstruction performance without changing the lightweight network parameters of the backbone network and maintaining fast inference speed. Compared to other reparameterization methods, it can significantly reduce training time and memory consumption.

The invention relates to an image processing model generation method and device, a medium and electronic equipment, and the method comprises the steps: obtaining a first decoder and a trained original image processing model, the original image processing model comprises an original self-coding module and other modules except the original self-coding module, the original self-encoding module comprises an original encoder and an original decoder, and resources used when the original decoder works are higher than resources used when the first decoder works; replacing an original decoder in the original self-encoding module with the first decoder to obtain a target self-encoding module; training the target self-encoding module to obtain a trained second decoder; and replacing an original decoder in the original image processing model with the second decoder to obtain a target image processing model, thereby reducing resource use of the target image processing model in actual reasoning.

The application discloses a kind of based on improved YOLOv10m light weight small target enhancement detection model noodle adhesion identification method, belong to food industrialization intelligent production and machine vision target detection cross technical field. Including: the original image of acquisition noodle industrialization cutting link;Establish noodle adhesion state detection model and training;Using the model detects noodle adhesion state.The noodle adhesion state detection model of the application, with YOLOv10m as baseline model, adopts light weight improved NAD-MobileViT main network, to adapt to the extraction demand of noodle slender structure characteristics;Introduce CEECA module in Neck feature fusion layer, realize the channel-space joint reinforcement of core feature and redundancy information suppression;In Head detection head part, add new high-resolution small target detection head.It can realize the real-time accurate monitoring of noodle edge form after cutting, can effectively identify interlaced state.

The application provides a method and system for image data recognition model construction. The method comprises: obtaining an image data set comprising true value labels; based on the image data set and a teacher network model, using an FSP method to initialize the weights of each layer of a student network to obtain at least two student network models with the same network structure; based on the image data set comprising true value labels, using a DML distillation method to train each student network model, and when the trained student network models all meet preset conditions, taking the trained student network models as an image data recognition model. Through the method, at least two student network models with the same structure and completed weight initialization of each layer are obtained by transferring learning from a teacher network model, which can avoid problems such as the inability to well initialize a neural network due to insufficient data, and can reduce the training difficulty. The DML distillation method is used to obtain an image data recognition model with guaranteed recognition ability, and the training cycle is short, the speed is fast, the deployment is easy, and the applicability is wide.

The present invention relates to the technical field of control systems, and discloses a remote monitoring method and system for a skin effect electric heating system based on big data. The present invention first introduces a linear auxiliary model to construct a minimization criterion function, and optimizes the parameters in the minimization criterion function based on an artificial rabbit optimization algorithm to obtain a skin effect electric heating system model; then, by training a fuzzy RBF neural network, the output control quantity controls the duty cycle and frequency and feeds back to a PID controller to regulate the temperature of the skin effect electric heating system; then, based on a graph neural network, time and space extraction is performed to obtain a multi-time scale time feature set and a multi-time scale space feature set, respectively, and temperature prediction is achieved based on a Bi‑GRU neural network; finally, a monitoring platform and a database are established to achieve data monitoring and storage of the skin effect electric heating system. The present invention achieves the purpose of remote monitoring by regulating the temperature of the skin effect electric heating system, and the remote monitoring results are accurate and objective.

The application relates to an image super-resolution reconstruction method and device based on anchor point guidance and boundary refinement cooperation, and belongs to the field of computer vision. The method comprises the following steps: performing shallow feature extraction on an input low-resolution image to obtain initial features; a deep feature enhancement network comprising multiple series cooperative reconstruction units is constructed, the initial features are input, and the initial features are sequentially enhanced by each unit; each unit performs anchor point routing global interaction and boundary perception region refinement on the input features to obtain global enhanced features and local refined features, the global enhanced features and the local refined features are fused through adaptive gating, and then the input features are fused through residual fusion, current cooperative enhanced features are output and are transmitted to the next unit, and finally, deep enhanced features are obtained; upsampling and pixel reconstruction are performed on the deep enhanced features, and a super-resolution image is output. The application considers long-distance structure dependence modeling and local boundary detail recovery, and realizes the cooperative optimization of global structure consistency and local detail fidelity under the constraint of lightweight deployment.

The invention discloses a deep learning-based railway wagon foreground and background side frame axle box spring loss fault identification method, and relates to the technical field of image processing. The invention aims to solve the problem that the detection precision of a deep learning network is reduced due to the fact that missing detection or false detection is easy to occur when the loss of the foreground and background side frame axle box spring is detected by adopting the existing image processing and deep learning technology. The invention discloses a deep learning-based railway wagon foreground and background side frame axle box spring loss fault identification method, which comprises the following steps of: inputting a detected image into a YOLOv5s network to obtain an identification result; according to the YOLOv5s network, a SimAM parameter-free attention mechanism is embedded at the tail end of each CSP1 unit of a Backbone part and after Focus slicing operation, an AFF attention feature fusion module is adopted in a Neck part to replace a Concat module, and the positioning loss function of the YOLOv5s network is SIoU.

The invention discloses a tumbling behavior detection method based on YOLOv5s and ResNet-50 improvement, and the method comprises the steps: S1, collecting a human body data set, forming a tumbling data set after recognition processing, dividing the tumbling data set, and marking tumbling attributes; s2, building a ResNet-50 model, training a fall data set, extracting features, and outputting a fall behavior detection model; and S3, adopting a multi-frame logic judgment strategy to perform falling behavior detection on the test set through the trained ResNet-50 falling behavior detection model, judging whether falling occurs or not according to feature changes, outputting falling attribute results of test set pictures, and performing classification according to different falling attributes. The method has the advantages of high accuracy, strong anti-interference capability and strong environmental adaptability, and can accurately identify the falling behavior under the conditions of different illumination, backgrounds and posture changes.