121 results about "Network complexity" patented technology

The invention provides a steel surface defect detection method and device, and belongs to the technical field of steel defect detection.The steel surface defect detection method comprises the steps that a defect detection network based on a YOLO architecture is constructed, and the defect detection network comprises a lightweight feature extraction network, a self-adaptive multi-scale feature fusion network and a head network; the lightweight feature extraction network comprises four layers of architectures, and each layer of architecture comprises a first convolutional layer and a space expansion feature module; the feature space expansion module comprises a second convolutional layer and an element multiplication operator module; the adaptive multi-scale feature fusion network comprises an adaptive fusion module and a feature space expansion module; training the constructed defect detection network based on the historical steel defect image, and performing channel pruning on the trained defect detection network by using LAMP (Loop-Mediated Isothermal Amplification); and acquiring a to-be-identified steel image, and inputting the to-be-identified steel image into the pruned defect detection network for defect detection. According to the method, the defect detection precision of the network can be improved while the network complexity is reduced.

The invention discloses an underwater target recognition method based on an improved YOLOv8 algorithm, and the method comprises the steps: inputting an underwater image into an improved YOLOv8n model for underwater target detection, and obtaining an output underwater target recognition result. The method has the advantages that the convolution blocks of the P5 layer of the backbone network and the last layer of the neck network adopt DSConv, so that the network complexity is reduced, and the reasoning speed is increased; a fourth C2f module of the backbone network adopts a C2f DiRMB module in which an inverted residual attention mechanism and dual-channel convolution are introduced, so that the capability of capturing key global information of the network is enhanced, training parameters are reduced, and the understanding of a complex scene is improved; and finally, a small target detection head for improving the small target detection capability is additionally arranged in the head network. According to the underwater target identification method, the mAP (at) is 0.5%, the mAP (at) is 0.5-0.95%, and the accuracy and the recall rate are respectively improved by 0.5%, 0.8%, 0.5% and 1.0%.

The invention relates to an intelligent equipment control method and device, electronic equipment and a storage medium. The method comprises the following steps: sensing an intelligent agent, obtaining multi-modal data, and processing the multi-modal data to obtain a sensing result; the decision-making intelligent agent is used for acquiring a sensing result and processing the sensing result to obtain a decision-making result; executing the intelligent agent, obtaining a decision result, and determining to-be-controlled intelligent equipment and to-be-executed control according to the decision result; the execution agent is used for controlling the intelligent equipment according to control and feeding back a control result and an equipment state of the intelligent equipment to the sensing agent; and compressing the pre-trained large model to generate an original agent, and dividing the original agent into a sensing agent, a decision-making agent and an execution agent according to the roles of the agents. In this way, the three types of lightweight intelligent agents are adopted for cooperation, the problem that the interaction real-time performance is poor due to the fact that a single large model network is high in complexity and large in parameter scale can be solved, and the interaction real-time performance is improved.

Systems, computer programs, devices, and methods that enable ML-based vision processing for low-power, embedded, and / or real-time applications. In one exemplary embodiment, smart glasses use classifiers that are based on machine-learned (ML) patch relationships. The ML patch features are determined during an offline training process. The ML patch features are grouped into weak classifiers, strong classifiers, and detectors to progressively improve prediction accuracy. An object detection architecture uses triggering logic, search management, and a classification neural network to enable event-based searching, interest-based searching, and / or dynamic search control. In some cases, pre-processing may also be used to minimize the neural network complexity (e.g., pre-processing for scaling, rotations, translations, etc.).

The invention discloses an intelligent signal detection and identification method based on time-frequency feature fusion and an anchor-free detection head. The method comprises the following steps: capturing a wireless signal broadband; carrying out time-frequency transformation preprocessing; signal detection and identification based on a neural network; and converting the time-frequency information. The invention provides a network suitable for signal detection and identification. The network is divided into a backbone network, a neck network and a detection head network. A backbone network composed of a component-convolution block architecture is provided, and the network complexity can be flexibly adjusted to effectively extract features; a multi-scale time-frequency feature aggregation module is inserted into the neck network to efficiently perform feature fusion; the detection head network is changed into an anchor-free mechanism to improve the output efficiency; and the final output of the network is subjected to time-frequency information conversion to obtain the bandwidth, duration and other parameters of the signal. According to the method, support can be provided for rapid and automatic signal detection and identification, and the problems of high human participation degree, high signal detection and identification rate and the like in military communication intelligence reconnaissance are effectively solved.

Provided in the present application are a communication method, a terminal device, and a core network element. The method comprises: a first terminal device sending first information to a first core network element by means of a user plane message, wherein the first information is used for indicating information of a second terminal device. When the first core network element collects the information of the second terminal device, a network side can acquire the information of the second terminal device, so as to sense the second terminal device. In this case, the first terminal device may not need to acquire the information of the second terminal device by means of a private interface, thereby unbinding the first terminal device and the second terminal device. For a consumer, the consumer can purchase the first terminal device of any manufacturer, that is, the sensing of the information of the second terminal device by the network side can be realized. For manufacturers or suppliers of the second terminal device, the manufacturers or suppliers do not need to deploy servers by themselves, thereby reducing costs and network complexity.

The invention relates to a gravity data inversion method and system and a medium, and the method comprises the steps: generating tens of thousands of random single abnormal training sample pairs; based on all the random single abnormal training sample pairs, selecting a part of preset quantity of data as a training data set, and selecting the other part of preset quantity of data as a verification data set; training a series of width learning networks with different network complexities by using a training data set, wherein the target of gravity inversion is to estimate an optimal density model; identifying the width learning network with the most suitable network complexity by using the verification data set; and inputting the measured gravity data vector into the width learning network with the most suitable network complexity to obtain an optimal density model. According to the invention, the precision and efficiency of density structure modeling are improved by using the width learning network and the single abnormal training sample.

Systems, computer programs, devices, and methods that enable ML-based vision processing for low-power, embedded, and / or real-time applications. In one exemplary embodiment, smart glasses use classifiers that are based on machine-learned (ML) patch relationships. The ML patch features are determined during an offline training process. The ML patch features are grouped into weak classifiers, strong classifiers, and detectors to progressively improve prediction accuracy. An object detection architecture uses triggering logic, search management, and a classification neural network to enable event-based searching, interest-based searching, and / or dynamic search control. In some cases, pre-processing may also be used to minimize the neural network complexity (e.g., pre-processing for scaling, rotations, translations, etc.).

The invention discloses a seam network complexity determination method and device and a nonvolatile storage medium. The method comprises the steps that geomechanical parameters corresponding to a fracturing section are determined; fracturing transformation parameters and oil and gas productivity corresponding to the fracturing section are obtained, and the fracturing transformation parameters are parameters used for conducting fracturing treatment on the fracturing section to obtain oil and gas output; and based on the geomechanical parameters, the fracturing transformation parameters and the oil and gas productivity, the fracture network complexity in the fracturing section is determined. According to the invention, the technical problem of high time and material resource cost caused by high limitation and low universality of obtaining the complexity of the seam network in the prior art is solved.

The invention provides a predictive maintenance model management method for a communication information system, belongs to the field of artificial intelligence, and realizes predictive maintenance of equipment and network nodes by utilizing real-time data in a communication network and combining a PPO algorithm. The method is suitable for a dynamic environment of a large-scale communication network, integrates multi-source data through a data sharing platform, and optimizes the learning and prediction capabilities of the model. Comprising distributed data acquisition, dynamic model updating, maintenance strategy generation, model optimization and real-time data synchronization. The problem that traditional centralized model management cannot adapt to communication network complexity and dynamic change is solved. Through distributed data processing and adaptive optimization of the PPO algorithm, the prediction precision and response speed of the system are improved, and the operation efficiency and maintenance capability of the communication information system are significantly improved.

The invention discloses an image denoising method based on a transform blind spot network and knowledge distillation. The method comprises the following steps: obtaining a training set; constructing a blind spot denoising network model based on a transform, and carrying out the training of the blind spot denoising network model; changing the size of a blind convolution kernel of the blind spot denoising network model to obtain a group of pre-trained blind spot denoising network models; respectively inputting the noise images into the group of pre-training blind spot denoising network models to obtain pseudo clean images with different weights; constructing a lightweight denoising network model based on the improved U-net, and training the lightweight denoising network model by using the noise image and the pseudo clean images with different weights; and processing a noise image by using the trained lightweight denoising network model to obtain a final denoised image. According to the method, image details are reserved, meanwhile, the complexity of the network is reduced, and the visual performance, the peak signal-to-noise ratio (PSNR) and the structural similarity (SSIM) of the image are improved.

The invention relates to the technical field of communication networks, in particular to an intention-driven satellite-ground convergence network on-demand arrangement system and method, and the method comprises an intention input layer which is used for receiving a multi-mode intention of a user; the intention analysis layer is used for analyzing the intention of the user to generate a recognizable network strategy; the arrangement control layer comprises a design state and an operation state; wherein the design state is used for dynamically generating a resource arrangement strategy based on an identifiable network strategy, and verifying the feasibility of the resource arrangement strategy in the digital twin network; the running state is used for generating an executable scheme based on the verified resource arrangement strategy; and the infrastructure layer comprises a satellite-ground unified atomic power abstraction module which is used for abstracting heterogeneous resources into standardized atomic power and executing an executable scheme. The scheme aims to provide a brand-new intention-driven satellite-ground fusion network arrangement architecture with dynamic generation, thorough decoupling and active prediction capabilities, so that the technical problem that complexity and dynamics of a 6G network cannot be handled at present is solved.

The application discloses an underwater target recognition method based on an improved YOLOv8 algorithm, which comprises the following steps: inputting an underwater image into an improved YOLOv8n model for underwater target detection, and obtaining an output underwater target recognition result. The method has the following advantages: DSConv is used in the convolution block of the P5 layer of the backbone network and the last layer of the neck network, which reduces the network complexity and improves the inference speed; the C2f_DiRMB module is used in the fourth C2f module of the backbone network, which introduces an inverted residual attention mechanism and a double-channel convolution, enhances the ability of the network to capture key global information, reduces the training parameters, and thus improves the understanding of complex scenes; finally, a small target detection head is added in the head network to improve the small target detection capability; and the underwater target recognition method improves the mAP@0.5%, mAP@0.5-0.95%, precision and recall by 0.5%, 0.8%, 0.5% and 1.0% respectively.

The invention discloses a near infrared spectrum reconstruction method based on stage perception mixed prior and a two-channel optical imaging system, and the method comprises the steps: constructing a deep expansion network model of stage perception mixed prior based on RGB guidance, and the network model comprises k cascade reconstruction stages, each reconstruction stage comprises a degradation perception residual gradient descent module and a near-end mapping module. According to the method, a degradation sensing residual error gradient descent module is introduced, the difference between a sensing matrix and an actual degradation process is effectively reduced through degradation learning, and the iteration process of a deep expansion network is divided into two stages of spectral space preliminary feature extraction and space detail deep extraction by applying a segmented optimization strategy; the network reconstruction precision is guaranteed, the optimization of the calculation efficiency is improved by adjusting the network complexity in stages, and the completeness and quality of the reconstructed image are guaranteed while the calculation overhead is reduced.

There is provided a method performed by a first entity. The method comprises obtaining decoding information and conveying the obtained decoding information towards a second entity, wherein the decoding information includes at least one neural network, NN, complexity value, and the at least one NN complexity value indicates NN decoding capability for using one or more NN models in a decoding process of a video bitstream.

The invention discloses a water chilling unit small sample fault detection method and system based on transfer learning, and the method carries out the fault detection through an expansion causal convolution module, a dense neural network module and a classification layer which are connected in sequence. A sparse connection strategy is introduced into dense blocks of the fault detection model, and representative connection between far and near layers is only reserved in each dense block, so that the network complexity is reduced, the feature redundancy is reduced, and the feature multiplexing advantage is kept; meanwhile, the method utilizes a transfer learning strategy to transfer labeled data knowledge of a source domain to a target domain, and introduces a meta-learning thought in a fine tuning stage to carry out gradient updating. In addition, feature subsets with high discrimination ability are screened out from an original high-dimensional feature space in combination with importance scores and correlation analysis, so that the detection precision of model training is improved.

The embodiment of the invention provides a network model knowledge distillation method, an image depth estimation method and electronic equipment, and the method comprises the steps: obtaining training data and a depth label corresponding to the training data, and the training data comprises a plurality of pieces of image data; inputting the training data into a first model to obtain a first intermediate output feature and a first depth estimation result, the first model being a trained convergence network model; inputting the training data into a second model to obtain a second intermediate output feature and a second depth estimation result, the network complexity of the second model being lower than the network complexity of the first model; calculating loss corresponding to the second model according to the first intermediate output feature, the first depth estimation result, the second intermediate output feature, the second depth estimation result and a depth label; and training the second model based on the loss to obtain a training convergent target network model. Therefore, the result precision when the electronic equipment executes the image depth estimation task can be improved.

The invention provides an OCTA image retinal vessel segmentation method based on a full-resolution network, and the method comprises the steps: carrying out the preprocessing of an input OCTA image through a data preprocessing module, wherein the preprocessing comprises the optical transformation and geometric transformation, and obtaining the preprocessed data; the feature coding module uses five feature extraction units which are arranged in sequence to extract image features of five levels and outputs the image features to the cascade feature enhancement module; a cascade feature enhancement module CFEM combines low-level detail information and high-level semantic information through cross-layer guidance for feature fusion, and then outputs a segmentation result; calculating loss and optimizing the OCTA image retinal vessel segmentation model based on the full-resolution network to obtain an optimized segmentation model; using the optimized segmentation model to obtain a retinal blood vessel image; according to the method, high-precision and high-efficiency segmentation of the retinal blood vessel can be realized, the network complexity is relatively low, and the computing resource consumption is relatively low.

The invention discloses a digital self-interference cancellation method and device based on an enhanced neural network structure, and the method comprises the steps: carrying out the complex value splitting operation and the non-linearity removal operation of a first stepped grid structure, obtaining a second stepped grid structure, carrying out the complex value splitting operation and the non-linearity removal operation of a first sliding window grid structure, and obtaining a second sliding window grid structure; obtaining a second sliding window grid structure; obtaining a first output of the dynamic phase layer according to the second stepped grid structure or the second sliding window grid structure; constructing a vector decomposition structure to separate amplitude information and phase information; feeding the amplitude information into the nonlinear hidden layer to obtain a second output; fusing the phase information and the second output; feeding the fused data into a full-connection output layer, and outputting a simulation result; and according to the simulation result, performing self-interference cancellation operation on the received signal to obtain a target signal. The method can improve the digital self-interference cancellation performance and reduce the network complexity, and can be widely applied to the technical field of full-duplex communication.

The invention discloses a communication network complexity evaluation and optimization method and device, equipment, a medium and a product. The method comprises the following steps: acquiring communication network complexity original information of each segmented network in an end-to-end communication network; the original information comprises evaluation data and weight of each preset first evaluation element and evaluation data of a preset second evaluation element; the first evaluation element is an evaluation element related to network complexity, and the second evaluation element is an evaluation element related to communication service volume; respectively calculating the network complexity of each segmented network according to the original information; and according to the segmented network weight and the network complexity of each segmented network in the end-to-end communication network, calculating the network complexity of the end-to-end communication network for optimizing the communication network. By adopting the method and the device, the cross-domain coordination, scheduling and optimization requirements of the end-to-end network are considered, the network complexity evaluation of the end-to-end communication network is realized, and the optimization of the communication network is facilitated.

This invention discloses an arbitrary-shape text detection method based on iterative optimization and cross-knowledge enhancement, belonging to the field of natural scene text detection technology. Compared with previous text information detection methods, this invention improves upon the problems of existing boundary iteration methods, which have complex processing flows, only apply to features at a single scale, have high computational costs, and cannot meet the requirements of practical applications. It obtains representational prior information through semantic segmentation and generates initial boundary points. Furthermore, it fuses representational prior semantic information with cross-scale feature information through feature enhancement, giving the features rich semantic information. Finally, it uses a unified boundary iteration Transformer that can fully utilize multi-scale information to predict boundary offset values, thereby continuously optimizing the initial boundary points. The simple and efficient network process of this method can greatly reduce network complexity and efficiently utilize cross-scale information and representational semantic information, thus meeting the speed and performance requirements of real-time detection.

According to the miner personal protection equipment detection method, system and equipment and the storage medium, the backbone network model is constructed based on the visual transformation network model, so that the feature extraction capability is better, the detection precision and efficiency are improved, the calculation complexity is reduced while the detection precision is ensured, the real-time performance of the algorithm is improved, and the detection efficiency is improved. The neck network is constructed based on a multi-scale space attention mechanism, multi-scale feature processing can be realized, the feature representation capability and the target detection performance of Shengloo are improved, the network complexity can be reduced by constructing a detection model based on a space reconstruction unit and a channel reconstruction unit, and the detection precision is also improved. According to the detection method, light weight is achieved while detection precision is guaranteed, ideal balance is achieved between calculation speed and recognition precision, and powerful support is provided for coal mine safety production.

The present invention relates to a communication signal recognition method and system based on an adaptive feedforward neural network, belonging to the technical field of signal recognition, and solves the problems of inaccurate recognition of existing small-sample signals and difficult adjustment of model parameters. It includes randomly generating the activation states of each hidden-layer neuron of binary type, the input weights of each hidden layer of real number type, and the hidden-layer bias based on a single-hidden-layer feedforward neural network, and forming multiple initial network individuals in a hybrid coding form as the parental population; using the root mean square error and network complexity as the objective function, and based on the preprocessed signal sample set, an improved NSGA-III algorithm is used to obtain the Pareto optimal population; according to the root mean square error, multiple network individuals are obtained from the Pareto optimal population as base classifiers, and the preprocessed communication signals collected in real time are input into the base classifiers for ensemble learning. After weighted summation of the output results, the category corresponding to the maximum value is taken as the communication signal recognition result. Accurate signal recognition is achieved.

The application discloses a billet detection method based on deep learning, and belongs to the technical field of billet detection, which is mainly used for the identification and positioning of billets in the process of discharging from a hot continuous rolling heating furnace. The multilayer feature extraction structure of the method is embedded with a CSA module, a SPPELAN network and an improved PANET network, the complex environment in which the billets are located is fully considered, end-to-end network construction from input of a detection picture to output of a prediction result is realized, and the problems of low precision and poor real-time performance of traditional methods are solved. Moreover, a residual network is used to extract image features, the network complexity and feature effectiveness are well balanced, and the model attention to the target and the prediction accuracy are improved. Meanwhile, the improved path aggregation network also makes the model pay more attention to billet features of different scales, and improves the detection precision. Therefore, the automation and intelligence degree of the method is high, and the industrial application prospect is good.

The invention relates to a nano tackifying fluid and supercritical CO2 fluid synergistic fracturing method. In the target stratum development area, nano tackifying fluid and supercritical CO2 fluid are adopted for collaborative fracturing, sand-carrying fluid and displacing fluid are combined to form a fracture network, and the corresponding relation between the construction parameter combination and the fracture flow conductivity and the corresponding relation between the construction parameter combination and the fracturing construction fluid amount are established; based on the corresponding relation, a multi-target optimization model with the fracture conductivity maximization and the fracturing construction liquid amount minimization as the target is constructed, and through population initialization, non-dominated sorting, crowding distance calculation, crossover variation selection and iterative updating, a construction parameter optimization result is output. The shale reservoir fracture flow conductivity and the fracture network complexity can be improved, the construction liquid amount is reduced, and the fracturing transformation effect and economical efficiency are improved.

The invention discloses a voice noise reduction method and device, a medium, a product and equipment, and the method comprises the steps: determining a voice linear spectrum frequency and a noise linear spectrum frequency based on noisy voice through employing a pre-trained neural network, and enabling the neural network to be obtained through training based on sample voice data; determining a filter gain based on the speech linear spectrum frequency and the noise linear spectrum frequency; and based on the filter gain, calling a linear filter to determine a noise reduction voice matched with the noisy voice, so that the noise reduction is completed through the filter, and the neural network only needs to be used for determining the voice linear spectrum frequency and the noise linear spectrum frequency corresponding to the noisy voice in real time to obtain the current filter gain. The method does not need to extract complex voice features capable of directly completing voice noise reduction, thereby greatly reducing the network complexity, improving the voice noise reduction response timeliness, and finally improving the voice quality in different scenes.

The invention relates to the technical field of multi-node ranging, and particularly discloses a multi-node ranging method and system based on code division multiple access, and the method comprises the steps: S1, calculating the network complexity, obtaining the bandwidth when the network complexity exceeds a certain range, and calculating the channel capacity; s2, calculating transmission time, calculating a transmission time ratio, correcting the bandwidth based on the transmission time ratio to obtain a corrected bandwidth, and obtaining corrected time; and S3, setting the bandwidth as a modified bandwidth, sending bytes with the same number to the test nodes, obtaining actual transmission time, screening qualified nodes, presetting a byte number gradient, obtaining a ladder coefficient when the requirement is met, sorting the test nodes, retaining efficient nodes and adjusting the corresponding bandwidth as the modified bandwidth. According to the invention, the byte data transmission frequency in the multi-node ranging process can be reduced, the positioning delay is reduced, and the response speed is improved.

Aspects of the disclosure are directed to a randomly determined first hop for network routing in an N-dimensional grid network. The first hop for a packet traversing the network is randomly determined based on a probability distribution for potential first hops. Based on a destination for the packet, a router in the network can search for potential paths and corresponding probabilities for the first hop, select a path based on the first hop, and forward the packet along that path to its destination. Network routing with a randomly determined first hop can improve fault tolerance without adding network complexities to the N-dimensional grid network.

This invention discloses a speech recognition method based on a photonic spiking neural network, comprising: preprocessing the original speech dataset to obtain FBank feature values ​​corresponding to the training set and the test set, respectively; training a convolutional spiking neural network using the FBank feature values ​​corresponding to the training set; processing the FBank feature values ​​corresponding to the training set and the test set using the trained convolutional spiking neural network to obtain high-dimensional features corresponding to the training set and the test set; training the photonic spiking neural network using the high-dimensional features corresponding to the training set; and processing the high-dimensional features corresponding to the test set using the trained photonic spiking neural network to obtain the speech recognition result. This invention provides several speech recognition methods based on photonic spiking neural networks that not only have the advantages of low power consumption, high speed, and short latency, but also have high network complexity, enabling classification and recognition of large-scale standard datasets, and are suitable for large-scale networks.

This disclosure relates generally to image coding and particularly to methods and systems for neural image compression (NIC). The disclosed NIC decoder / encoder may include various neural network components that are configured to achieve a balance between network complexity and coding efficiency. Such a NIC decoder / encoder implementation particularly include a core decoder and a hyper decoder each including a neural network architecture adapted for achieving a lightweight decoder / encoder.