144 results about "Network segmentation" patented technology

The invention provides an unsupervised domain-adaptive brain tumor semantic segmentation method based on deep adversarial learning. The method comprises the steps of deep coding-decoding full-convolution network segmentation system model setup, domain discriminator network model setup, segmentation system pre-training and parameter optimization, adversarial training and target domain feature extractor parameter optimization and target domain MRI brain tumor automatic semantic segmentation. According to the method, high-level semantic features and low-level detailed features are utilized to jointly predict pixel tags by the adoption of a deep coding-decoding full-convolution network modeling segmentation system, a domain discriminator network is adopted to guide a segmentation model to learn domain-invariable features and a strong generalization segmentation function through adversarial learning, a data distribution difference between a source domain and a target domain is minimized indirectly, and a learned segmentation system has the same segmentation precision in the target domain as in the source domain. Therefore, the cross-domain generalization performance of the MRI brain tumor full-automatic semantic segmentation method is improved, and unsupervised cross-domain adaptive MRI brain tumor precise segmentation is realized.

Systems, methods, and other embodiments associated with logically partitioned networking devices are described herein. One example method includes receiving a message from a common interface. The message comprises a logical partition header (LPH) and a network segmentation header (NSH). The LPH may be associated with a logical partition of a networking device. The NSH is associated with a grouping (e.g., segmentation) of networking devices. The example method may also include forwarding the message to the grouping of networking devices based, at least in part, on the NSH and a virtual route forwarding (VRF) table. Forwarding the message to the logical partition of the networking device based, at least in part, on the LPH.

The invention discloses a CT image pulmonary nodule detection system based on 3D full-connection convolution neural network. The detection system comprises the following five steps: constructing a training set data; performing 3D convolution neural network classification network training; performing 3D convolution neural network segmentation network training; carrying out false-positive suppression in which the trained segmentation network and the false-positive are utilized to inhibit the network; and detecting the pulmonary nodule. The technical schemes of the invention can realize the full automatic detection without any human intervention. At the same time, the recall rate of pulmonary nodule detection can be increased effectively; the false-positive focus of infection is reduced considerably; and a pixel level positioning quantitative and qualitative result for the focus-of-infection area of the pulmonary nodule can be obtained.

The invention provides an MRI (Magnetic Resonance Imaging) brain tumor localization and intratumoral segmentation method based on a deep cascaded convolution network, which comprises the steps of building a deep cascaded convolution network segmentation model; performing model training and parameter optimization; and carrying out fast localization and intratumoral segmentation on a multi-modal MRIbrain tumor. According to the MRI brain tumor localization and intratumoral segmentation method provided by the invention based on the deep cascaded convolution network, a deep cascaded hybrid neuralnetwork formed by a full convolution neural network and a classified convolution neural network is constructed, the segmentation process is divided into a complete tumor region localization phase andan intratumoral sub-region localization phase, and hierarchical MRI brain tumor fast and accurate localization and intratumoral sub-region segmentation are realized. Firstly, the complete tumor region is localized from an MRI image by adopting a full convolution network method, and then the complete tumor is further divided into an edema region, a non-enhanced tumor region, an enhanced tumor region and a necrosis region by adopting an image classification method, and accurate localization for the multi-modal MRI brain tumor and fast and accurate segmentation for the intratumoral sub-regions are realized.

The invention proposes an image semantic segmentation method based on a super-pixel edge and a full convolutional network, so that a technical problem of low accuracy in the existing image semantic segmentation method is solved. The method comprises: a training sample set, a testing sample set, and a verification sample set are constructed; a full convolutional network outputting a pixel-level semantic mark is trained, tested, and verified; semantic segmentation is carried out on a to-be-segmented image by using the verified full convolutional network outputting a pixel-level semantic mark to otain a pixel-level semantic mark; BSLIC sub-pixel segmentation is carried out on the to-be-segmented image; and semantic marking is carried out on BSLIC super-pixels by using the pixel-level semantic mark to obtain a semantic segmentation result with combination of the super-pixel edge and the high-level semantic information outputted by the full convolutional network. Therefore, the original full convolutional network segmentation accuracy is kept and the segmentation accuracy of the small edge is improved, so that the image segmentation accuracy is enhanced. The image semantic segmentation method can be applied to classification, identification, and tracking occasions requiring target detection.

The invention discloses a CT image kidney segmentation algorithm based on a residual double-attention deep network. According to the method, the advantage that the residual error unit can repeatedly utilize the features and the excellent feature learning capability of the double attention mechanism are combined; a residual double attention module is designed; and a residual double attention moduleis used as a basic module to construct a U-shaped deep network segmentation model, and a loss function for segmentation is designed at the same time, so that the U-shaped deep network segmentation model can pay more attention to kidney region features, can effectively cope with shape changes of the kidney with cystic lesions, and can maintain robustness for the shape changes of the kidney with cystic lesions. Therefore, the boundary of the kidney area is accurately positioned, and automatic segmentation of the kidney area in the CT image is achieved, and a good segmentation effect is achieved.

The present invention enables edge components, such as Border Elements, of the service provider's network to capture performance data on all endpoints connected to them including registered devices (e.g., CPE gateways, IP phones, and terminal adaptors) and access links. The present invention enables the performance data to be sent to a centralized repository that consolidates information across the entire network, analyzes it, and segments it with respect to location of events that cause defects in calls. The performance data can then be graphically mapped into predefined network segmentations to enable faster identification and resolution of network problems.

The invention provides a fast hash vehicle retrieval method based on multi-task deep learning. The fast hash vehicle retrieval method includes a multi-task deep convolutional neural network used for deep learning and training recognition, a segmented compact hash code and instance feature fusion method for improving the retrieval accuracy and the practicality of the retrieval method, a local sensitive hash reordering algorithm for improving the retrieval performance and a cross-modal retrieval method for improving the robustness and accuracy of a retrieval engine. In the fast hash vehicle retrieval method, firstly, a method for segmented learning of hash codes through a multi-task deep convolutional network is proposed, image semantics and image representation are combined, the connectionbetween related tasks is used for improving the retrieval accuracy and refining image features, and at the same time, minimizing image coding is used for making learned vehicle features more robust; secondly, a feature pyramid network is used for extracting the instance features of vehicle images; then, a local sensitive hash reordering method is used for retrieving the extracted features; and finally, a cross-modal assisted vehicle retrieval method is used for the special case in which target images of inquired vehicles cannot be obtained.

Method for computing all-terminal reliability for arbitrarily interconnected networks such as the United States public switched telephone network. The method includes an efficient search algorithm to generate minimal cut sets for nonhierarchical networks directly from the network connectivity diagram. Efficiency of the search algorithm stems in part from its basis on only link failures. The method also includes a novel quantification scheme that likewise reduces computational effort associated with assessing network reliability based on traditional risk importance measures. Vast reductions in computational effort are realized since combinatorial expansion and subsequent Boolean reduction steps are eliminated through analysis of network segmentations using a technique of assuming node failures to occur on only one side of a break in the network, and repeating the technique for all minimal cut sets generated with the search algorithm. The method functions equally well for planar and non-planar networks.

Initially, customer time series credit files are acquired. The credit files are organized in a data mart environment for supporting a query system. Time series utilization attributes are created and a neural network time series segmentation process is applied and N.times.N dimension segments are generated for analysis. The chart may be modified to more accurately depict profitable credit revolvers. Credit data from each potential new customer is processed in a similar fashion by the neural network segmentation process. Profitable credit revolvers are identified by having credit utilization patterns belonging to profitable segments previously identified.

The invention relates to a lung nodule detection method and device, a computer device and a storage medium. The method comprises the following steps segmenting the lung CT image through a three-dimensional convolution neural network segmentation model, and segmenting a lung region image; detecting suspicious nodules from the lung region image through a three-dimensional U-Net detection model; classifying the suspicious nodules through a three-dimensional dichotomous network, and removing the fake nodules. According to the lung nodule detection method and device, the computer device and the storage medium, the lung CT image is segmented through the three-dimensional convolution neural network segmentation model; the lung region image is segmented, the segmentation speed is high, and the speed of subsequent detection of the lung nodules is accelerated; meanwhile, the invention can be applied to the segmentation of the lung region of all lung CT images.

A packet transfer apparatus for a network system is disclosed. The apparatus comprises a packet receiver that accepts an input of packets from a first network segment, a packet classifier that classifies packets based on their respective process flows, a packet discarder to discard packets and a packet sender that sends packets to a second network segment. Another aspect of the invention is a network system comprising a plurality of terminal nodes, at least one packet transfer unit effectively connected between at least two of said terminal nodes, said at least one transfer unit further comprising a packet classifier that classifies packets into their respective process flows. Yet another aspect of the invention is a method of transferring packets in a network comprising accepting an input of packets from a first network segment, classifying the packets based on their process flows, providing a unique process flow identification to packets belonging to a same process flow, discarding at least a packet, providing information to a packet classifier regarding the discarded packets, and stopping further transfer of packets having a same PFID as the discarded packet.

The invention relates to a lightweight network real-time semantic segmentation method based on an attention mechanism, which is used for solving the problems that the segmentation precision and the segmentation efficiency are difficult to balance and the practical application cannot be met. The method comprises: preparing image data; and constructing a lightweight real-time semantic segmentation network based on an attention mechanism, providing a new asymmetric encoding and decoding network structure. In an encoder, a lightweight module-separable asymmetric module is used, and the module combines the advantages of depth separable asymmetric convolution and hole convolution. The calculation amount is greatly reduced while the precision is kept; an attention feature fusion module is designed in a decoder, features in the encoder and features in the decoder are fused, the fused features are selected and combined through an attention mechanism, the features useful for recovering image information are enhanced, and the network segmentation precision is effectively improved. Finally, semantic segmentation is achieved by using the trained segmentation network.

The invention discloses a semi-supervised medical image segmentation method based on adversarial collaborative training. A neural network segmentation model is trained by using a small amount of labeled medical image data and a large amount of unlabeled medical image data, so that the model performance is improved. The model uses two decoder branches with different structures, the two decoder branches share the same encoder, and the two decoder branches can learn each other through a cooperative training method. Meanwhile, the model also uses an adversarial learning method to train a discriminator, and the discriminator can learn the high-order continuity between the segmentation result and the real label, so that the output of the segmentation network is closer to the real label visually.And meanwhile, the discriminator can also select the part with higher confidence in the unlabeled data pseudo tags to train the segmentation model. The method provided by the invention is not limitedby diseases and focus types, can be used for medical image segmentation of diseases of various parts such as the liver and the oral cavity, and has very good universality and universality.

The invention relates to a road scene segmentation method based on a full convolutional neural network, and the method comprises the following steps: 1, carrying out the median filtering of an original road scene image through a KSW two-dimensional threshold value and a genetic algorithm, and obtaining a training set; 2, constructing a full convolutional neural network framework; 3, taking a training sample obtained at step 1 and an artificial segmentation image discriminated and identified through human eyes as the input data of the full convolutional neural network, and obtaining a deep learning neural network segmentation model with the higher robustness and better accuracy through training; 4, introducing to-be-segmented road scene image test data into the trained deep learning neuralnetwork segmentation model, and obtaining a final segmentation result. An experiment result indicates that the method can effectively solve a segmentation problem of a road scene image, has the higherrobustness and segmentation precision than a conventional road scene image segmentation method, and can be further used for the road image segmentation in more complex scenes.

The present invention provides a detection method of an unstructured point cloud feature point and an extraction method thereof. The extraction method includes (1) calculating the Harris response value of a sampling point in different scale space; (2) selecting the Harris response value of the optimal scale space as the Harris response value of the sampling point to obtain a feature point set Q; (3) selecting one maximum point of the Harris response values possessing maximality in both of the scale space neighborhood and a geometric neighborhood as a candidate feature point, at last, selecting the optimizing strategy to draw the final feature point. A tangent plane of the gained feature point is subjected to network segmentation under a polar coordinate system, and then a neighborhood point of the feature point is projected to the tangent plane, a feature information statistical matrix is generated by voting projected length corresponding to projective points from each grid to four peaks of the grid, then both of a row vector and a column vector are respectively subjected to the DCT transform and the DFT transform, the elements of the upper left corner after transform is a character description vector.

The invention provides a network function end node prediction method based on cascading failure. The method comprises steps that 1, data pre-processing on the infrastructure network is carried out, and actual network abstraction is carried out to establish a network model; 2, an initial weak node is determined based on a key node identification method or historical data, and a load capacity model is further established; 3, network segmentation points during cascading failure are calculated; and 4, a propagation distance of a function end node is predicted according to an overload cascading failure propagation distance. The method is advantaged in that the function end node in a cascading failure process can be discovered in advance at a protection stage before cascading failure, a key node or a not-easy-to-restore node is designed and distributed at the function end node in advance, through propagation prediction of the function end node, real-time control in the cascading failure process is carried out, and thereby cascading failure control and repair work afterwards are facilitated.

A method includes obtaining first information indicative of instability of a data communication network. The method also includes isolating a first portion of a network from a second portion of the network responsive to the obtained first information. After a predetermined period of time, second information indicative of instability of the first portion is obtained. The method further includes isolating a first segment of the first portion from a second segment of the first portion responsive to the obtained second information.

The method comprises: the managing device gets the tree topology structure information about all managed network devices in the network; according to the tree topology structure of the managed network devices, the network device at the top layer of the tree topology structure is appointed as the agent device; allocating a public network address for the agent device; allocating an internal private IP address located in same network segmentation with the agent device for the agent device and each subset connected to the agent device; starting the Web relay server of proxy server, which will open a port used for forwarding of message for each subset connected to the agent device, and informs the port number corresponding to each subset to the managing device; the relay server module in the agent device receives the request message from the managing device and forwards the message to the visited subset according to the port number of the visited subset in the message; the message returned from the visited subset is transmitted to the managing device.

The present invention provides a method for communication, an RRM server, a RSU, and a node. The method includes: performing, by a Radio Resource Management RRM server, segmentation of an area in a communication network to form at least one segment for at least one Roadside Unit RSU respectively; allocating, by the RRM server, channels for the at least one segment; sending, by the RRM server, network segmentation information and channel allocation information to each of the at least one RSU, wherein the network segmentation information indicates a segment for the RSU and the channel allocation information indicates channels allocated for the segment so that RSU communicates with a node which enters the segment through the channels allocated for the segment.

The method features the following points: connecting an agent device in series between the terminal and the slave device embedded-in a TELNET server; the agent device has serial-port management function or TELNET management function and a forwarding module for data forwarding; public IP address and internal IP address are allocated to the agent device, and for the slave device only a internal IP address located in same network segmentation with the internal IP address of the agent device is allocated. The method thereof comprises: building a connection between the terminal and the agent device, which can use the serial port or TELNET to make remote connection; after the terminal sends a request for managing the slave device, the TELNET connection will be built between the agent device and the slave device.

According to the eyeball segmentation method and device based on the convolutional neural network and the mixed loss function, the segmentation precision of eyeballs in a CT image can be improved. The method comprises the following steps: (1) in a data set manufacturing stage, drawing an eyeball segmentation gold standard through manual labeling, carrying out preprocessing operations of taking two-dimensional slices, downsampling and standardizing on original three-dimensional CT image data, and then integrally dividing a data set into three parts, namely a training set, a verification set and a test set for training and testing a network; (2) in a network training stage, establishing a convolutional neural network model cascaded by a coarse segmentation module and a U-shaped residual error fine tuning module, and performing multi-level supervised optimization on a network segmentation result by using a mixed loss function formed by cross entropy, intersection-to-union ratio and structural similarity measurement; and (3) in a test stage, feeding a test data set into the optimal segmentation model obtained by training for segmentation, and restoring an output result into three-dimensional data to obtain a final eyeball segmentation result.

The invention discloses a method and equipment for detecting a driving area, relating to the technical field of automatic driving. The method is used for solving the problem that accurate three-dimensional distance information cannot be obtained in current driving area detection, the driving area cannot be accurately detected. The method comprises the steps that first feature information of road surface points and second feature information of road shoulder points in the bird's-eye view feature map are determined through a neural network segmentation model according to the average reflection intensity and height coding features of grids in the bird's-eye view feature map, and the bird's-eye view feature map is obtained by conducting rasterization processing on a point cloud map; road surface points corresponding to the road surface points in the bird's-eye view feature map in the point cloud map is determined according to the first feature information, and road shoulder points corresponding to the road shoulder points in the bird's-eye view feature map in the point cloud map is determined according to the second feature information; the road surface points and the road shoulder points in the point cloud map are subjected to geometric model fitting to determine the driving area, and the driving area is detected by adopting a deep learning method, so that the accuracy is high.

The invention discloses a dental CBCT three-dimensional tooth segmentation method based on deep learning. According to the invention, the method includes performing semantic segmentation on teeth in each image in a CBCT image sequence, so that noise is removed; constructing a deep supervision coding-decoding network for denoising an oral CT image, mutually connecting coding and decoding sub-modules through a series of nested dense jump paths, and reducing semantic deficiency of feature maps in the coding and decoding sub-modules; the method specifically comprises the following four stages: stage 1, collecting and preprocessing a dental CBCT image; stage 2, constructing a model training set; stage 3, constructing a network segmentation model of a coding-decoding structure; and step 4, performing model training and evaluation. Experimental results show that the Dice similarity coefficient for individual three-dimensional tooth segmentation is 95.64%. Results show that the method provided by the invention provides an effective clinical application framework for digital dentistry.

The invention discloses a rock core FIB-SEM image segmentation method based on a convolutional neural network, and mainly relates to an image segmentation technology of rock core sequence images. The method comprises the following steps: (1) establishing a core FIB-SEM image data set; (2) constructing a convolutional neural network: embedding a channel attention module into a coding stage, extracting multi-scale features by using an improved feature pyramid attention module, extracting a fine boundary by using multi-scale space attention in a decoding module, and recovering an original resolution by using a sub-pixel convolution module in an up-sampling stage; (3) performing network training and parameter optimization to obtain a model with the best effect; (4) performing a network segmentation result test by using the test set obtained in the step (1); the FIB-SEM pores of the rock core are extracted by using the convolutional neural network, manual operation is not needed, and the segmentation precision is improved.

The invention discloses user equipment, a method for cell reselection therefor, and a computer readable medium. The method comprises the steps: obtaining a cell reselection mode for the cell reselection based on the network segmentation information through the preset information or an instruction sent by a network side; obtaining all neighbor cells with the network segmentation information; and performing the cell reselection based on the network segmentation information and a signal intensity measurement value for all obtained neighbor cells with the network segmentation information. According to the above technical scheme, the method can enable the user equipment to quickly perform reselection to obtain a network segmentation cell of a required service, and achieves the diversified services of an NR network.

The invention belongs to the technical field of medical image processing, and particularly relates to a multi-channel head magnetic resonance imaging tissue segmentation method. However, the existingdeep learning method does not utilize anatomical structure information of relatively fixed brain. The invention provides a multi-channel head magnetic resonance imaging tissue segmentation method, which comprises the following steps of: 1, matching a pre-segmentation label most similar to each image to form four channels; And 2, inputting the obtained four channel data into a convolutional neuralnetwork, training the network through a real label of the input data to obtain a training model, and enabling the test data to pass through the trained model to obtain a segmentation result. The priortexture information of brain tissue is fully utilized, a new channel is added, accurate segmentation of the network is promoted, and the network segmentation precision is improved. The method provided by the invention is simple and high in robustness, and can be added to any segmented network for segmentation without changing the original network structure.

A network element of a cable television (CATV) network, said network element comprising a distributed access node comprising a core network interface for receiving a plurality of broadcast multiplexes; means for dividing the plurality of broadcast multiplexes into at least a first transmission content and a second transmission content, wherein the first and the second transmission content comprise at least partly different multiplexes; and means for transmitting the first transmission content to a first network segment and the second transmission content to a second network segment.

The invention provides a method and device for calculating the space loading rate of a carriage, and the method comprises the steps: building a carriage XYZ-axis coordinate system, enabling a Z axis to be the distance direction of infrared distance measuring sensors, enabling the infrared distance measuring sensors to be arranged in the distance direction, and enabling the infrared distance measuring sensors to carry out the distance measurement sampling in the carriage; s2, constructing a triangulation network segmentation space according to the sampling values of the infrared distance measuring sensors on the Z axis, performing linear interpolation on the space by using the sampling values on the Z axis, and constructing a plurality of minimum triangulation networks on XY coordinates ofthe coordinate point set established in the step S1; calculating the volume of each minimum triangulation network by using a double definite integral method; traversing all the minimum triangulation networks obtained through segmentation, summing the volume of each minimum triangulation network, calculating the carriage volume, and calculating the cargo volume according to the carriage volume. Theinvention is high in calculation efficiency, real-time calculation can be achieved, and calculation precision is more accurate than that of a manual mode.