29 results about "Classification result" patented technology

The invention belongs to the technical field of data stream mining, and particularly relates to a method for handling missing values during data stream decision tree classification. The method includes reading data samples in data streams and updating sliding windows; updating missing handlers if the missing handlers corresponding to attributes are available when the detected attributes in the current data samples have the missing values, or adaptively selecting and creating missing handlers according to characteristics of data if the missing handlers corresponding to the attributes are not available; supplementing the missing values in the data samples by the aid of the missing handlers to obtain complete data samples, training the complete data samples according to a Hoeffding decision tree classification process and returning data stream decision tree classification results. Compared with existing methods, the method has the advantages that the method is superior in time performance, the classification accuracy of decision tree models can be sufficiently guaranteed, accordingly, the time expenditure can be reduced, the time performance can be improved, the data stream classification handling speeds can be increased, and requirements of actual data stream handling application can be met.

A signal processing apparatus includes a conversion unit, an encoding unit, and a control unit. The conversion unit converts first moving image data having a first frame rate into second moving image data having a second frame rate by inserting a new frame into the first moving image data having the first frame rate at predetermined timing. The encoding unit encodes the second moving image data having the second frame rate by selectively using intra-frame encoding and inter-frame prediction encoding. The encoding unit classifies each frame of the second moving image data as a first frame type which is not referred to in the inter-frame prediction encoding and a second frame type which is referred to in the inter-frame prediction encoding, and the encoding unit encodes the second moving image data according to the classification result. The control unit controls the encoding unit such that the frame inserted by the conversion unit among the second moving image data is encoded by classifying the inserted frame as a particular frame type.

The invention relates to a text classification method and device, computer equipment and a storage medium, and the method comprises the steps: obtaining to-be-classified text data, so as to obtain to-be-classified data; inputting the to-be-classified data into the target text classification model for classification to obtain a classification result; outputting the classification result to a terminal so as to display the classification result on the terminal; wherein the target text classification model is obtained by extracting a vector set from input text data, generating a label and then combining to form a training data set for training. According to the invention, the label is generated in a manner of automatically generating the label for the input text data; combining the generated label with an initial vector set; text data labels are corrected in an iterative mode, the initial text classification model is trained again, the training data quality is improved, the early-stage manual label labeling cost is reduced, and the requirement for a large amount of labeled data in a text classification task is quickly responded to, so that the text classification model is quickly established, and the text classification efficiency is improved.

The invention discloses a method for matching cured tobacco leaves by using a tobacco leaf beating and re-drying module. The method comprises the following steps: performing classified collection on tobacco leaf samples of an aroma module, a smoke module and a mouthfeel module to serve as modeling samples by adopting a sensory evaluation method; then preparing the modeling samples into powder, scanning and acquiring a near infrared spectrogram, preprocessing the spectrogram, then performing calculation on principal components, and creating a Naive Bayes classification model by using principle component scores; and then preparing to-be-matched tobacco leaf samples into powder, acquiring a near infrared spectrum, preprocessing the spectrum, then performing principle component decomposition operation on the spectrum of the to-be-matched tobacco leaf samples by using spectrum data loads of the modeling samples to obtain principle component scores of the to-be-matched tobacco leaf samples, classifying the to-be-matched tobacco leaves by using the principle component scores and the Naive Bayes classification model, and performing module matching according to a classification result. By adopting the method disclosed by the invention, the sensory evaluation is only needed at a modeling stage, manpower and time are saved, and the to-be-matched tobacco leaves can be classified and matched quickly and accurately.

The invention discloses an upper human body detection method based on an accumulative perceptron. The upper human body detection method based on the accumulative perceptron includes that cascading with the traditional perceptron algorithm, and using accumulated values of a sliding window at different stages of perceptron classification results to boost stage by stage in a rough-to-fine mode to finish the precise upper human body classification and locating so as to form a special accumulative perceptron algorithm; extracting a Gabor characteristic pattern through Gabor filtering, wherein the characteristic simulates the response of human eye cells to visual signals and well captures the global characteristic and local characteristics of an input image; using a sliding window detection method to traverse the image to be detected, and classifying in each sliding window through the accumulative perceptron. Because the position of the accumulative perceptron at each sliding window is independent, the algorithm can be performed with parallel optimization.

The invention discloses a nuclear magnetic resonance image automatic classification method and device based on an MDCLSTM-LdenseNet network, and belongs to the field of medical image processing. The method comprises the following steps: obtaining nuclear magnetic resonance images of different types of subjects; preprocessing the nuclear magnetic resonance image to obtain a training data set, a verification data set and a test data set; training is carried out on the constructed MDCLSTM-LDenseNet network model, so that the MDCLSTM-LDenseNet network model can be obtained; inputting the test dataset into the trained MDCLSTM-LDenseNet network model to carry out a test; obtaining classification results of the nuclear magnetic resonance images of the three types of subjects with normal cognition, mild cognitive impairment and Alzheimer's disease and classification accuracy of the MDCLSTM-LdenseNet network model; according to the method and the device, used data sets are strictly divided; data leakage is avoided; the method is advantaged in that the training data set, the verification data set and the test data set are not intersected, reliability of classification results of normal cognition, mild cognitive impairment and Alzheimer's disease is guaranteed, classification accuracy is higher, the missed diagnosis rate and the misdiagnosis rate are lower, and doctors can be more effectively and reliably assisted to diagnose Alzheimer's disease.

The invention discloses an identity recognition method based on multi-modal information. The identity recognition method comprises the following steps: 1, making a multi-modal video data set with a label; 2, respectively constructing and training a face detection model and a head detection model; 3, constructing and training a feature extraction model of the face, the head and the sound; 4, performing feature extraction on the face, head and sound information through the trained feature extraction models; 5, constructing and training a classification model to respectively classify the three extracted features; 6, performing result prediction by using the three features through the classification model; 7, performing information fusion on the classification result according to the formulated multi-modal information fusion strategy; 8, arranging the fused result and then outputting an identity recognition result. Based on the identity recognition network model based on multi-modal information provided by the invention, the identity recognition network model has a wide application prospect in the fields of human-computer interaction, information security, security monitoring and the like.

The invention relates to the technical field of safety protection, and discloses a face detection model based on face super-resolution. The detection method of the detection model comprises the following steps: S1, extracting features of an input image through a detection network, carrying out the downsampling, and obtaining feature layers of a plurality of scales; and S2, performing detection inthe feature layers generated in the step S1 by the detection network to obtain detection and classification results, and storing the detection and classification results. According to the face detection model based on face super-resolution, the performance of face detection is improved through face super-division. The bottom layer features have high resolution. The model is advantaged in that extraction of the face frame positioning information, especially small face positioning, is facilitated. The high-level characteristics have strong semantic information, extraction and determination of whether the face is the face classification information are facilitated. The resolution of the low-level characteristics is enhanced through super-division. The low-level information is enhanced. The high-level information is utilized. The small face and blurred face detection capability is improved.

The invention relates to a power distribution network fault classification method and system based on deep learning, and a medium. The method comprises the steps: obtaining a plurality of original fault waveform data sets; respectively processing each original fault waveform data set to obtain target sample data; making all target sample data into a data set, dividing the data set into a trainingset and a test set, constructing a deep learning network model, and training the deep learning network model by using the training set to obtain an original fault classification model; performing parameter tuning on the original fault classification model by using the test set to obtain an optimized fault classification model; acquiring a real-time fault waveform data set, processing the real-timefault waveform data set to obtain to-be-detected fault data, and performing real-time identification on the to-be-detected fault data by utilizing the optimized fault classification model to obtain areal-time fault classification result. According to the method, the faults in the power distribution network are quickly and reliably recognized and classified by utilizing the strong classificationadvantage of deep learning, the recognition efficiency is high, and the classification accuracy is high.

The invention provides a DenseNet-based CT image classification method and a DenseNet-based CT image classification device for COVID-19 patients, which are used for classifying according to computed tomography images of suspected COVID19 patients to obtain a classification result. The method includes the following steps of: storing medical image information; preprocessing by using a preprocessing method to obtain preprocessed data; obtaining deep information and shallow information from the preprocessed data by using the trained dense connection neural network model, and fusing the deep information and the shallow information to obtain a fused feature vector; mapping the fusion feature vector to a low-dimensional space by using an activation function to obtain a classification probability prediction value; obtaining a CAM activation graph based on the internal parameters of the densely connected neural network model and the computed tomography image; displaying a computed tomography image, a classification probability prediction value, and a CAM activation map to assist a doctor in diagnosis. The method and the device are suitable for an early screening stage of an epidemic situation area, the problem of too high false negative of accounting detection can be solved, and the diagnosis efficiency can be improved.

The invention relates to a simulation testing method and device of a whole vehicle electrical system, and aims to solve the problems that in the prior art, the testing is incomplete and the testing cost is high due to the real object-based testing of the whole vehicle electrical system. According to the simulation testing method of the whole vehicle electrical system, the whole vehicle electricalsystem comprises a plurality of independent subsystems. The method comprises the steps of performing equivalent circuit simplification on each subsystem; classifying all parts; performing model building on the parts in each subsystem, and dividing part models of each subsystem into a plurality of part model libraries according to a classification result; assigning a value to each part model in thepart model libraries; establishing a plurality of model testing racks according to the classification result; and applying all the value-assigned part models in each type of the part model library tothe model testing rack of the corresponding type, and performing part model testing, subsystem model testing and system model testing.

The invention relates to a feedback type classification method integrating unsupervised learning and supervised learning. The method comprises the steps of classifying an original feature set throughunsupervised learning to obtain a classification result with label information; randomly and equally dividing the classification result into a training group and a control group, wherein the traininggroup serves as input of supervised learning; adopting a feature selection algorithm to enable the classification accuracy of the supervised learning to be the highest, and obtaining a corresponding feature subset; reconstructing a control group feature set, namely extracting a feature subset corresponding to the control group feature set according to the feature category of the feature subset, taking the reconstructed control group feature set as the input of the supervised learning model obtained by the training group, and calculating the classification accuracy of the supervised learning model; setting a classification accuracy threshold value as an iteration termination condition, terminating iteration if the classification accuracy is higher than a preset threshold value, obtaining aclassification result, otherwise, reconstructing an original feature set, and carrying out iteration in sequence until the preset condition is met. The classification method has good adaptability andaccuracy for the classification problem of unknown classification labels.

The invention discloses an advertisement recommendation method and apparatus, an electronic device and a storage medium. The method comprises the steps of obtaining archive data and generation data of a target user, wherein the target user belongs to an inactive user, and the generated data is associated data generated between the user and the advertisement; splicing archive data of the target user and the generated data and then performing dimension reduction processing to obtain a feature vector corresponding to the target user; inputting the feature vector corresponding to the target user into a pre-trained recommendation model to obtain a classification result corresponding to the target user, wherein the recommendation model is obtained through training by taking seed users as positive sample users in advance and taking users which are active and are not interested in the target advertisement as negative sample users, and the seed users at least comprise converted users interested in the target advertisement in the active users; and if the classification result is positive, recommending a target advertisement to the target user, wherein the classification result is that the similarity between the forward characterization target user and the seed user meets a preset condition.

The invention provides a flow analysis method of an IPv6 network. The method comprises the following steps: collecting flow data and user data of the IPv6 network; predicting the traffic bandwidth of the user based on the traffic data and the user data; acquiring real-time stream data of the user whose traffic bandwidth satisfies a preset condition; classifying the real-time streaming data; and displaying the prediction result of the traffic bandwidth and the classification result of the real-time streaming data. The invention further provides a flow analysis device of the IPv6 network, electronic equipment and a storage medium.

The embodiment of the invention discloses a network attack determination method and system, electronic equipment and a storage medium. The method comprises the following steps: processing access data corresponding to each website based on a network attack detection model, and determining a classification result; if the classification result is an attack classification result, acquiring attack packet data and attack address information corresponding to the corresponding website; analyzing and processing the attack packet data and the attack address to determine the risk level of the website; and when the risk level is consistent with a preset risk level, determining a network attack and carrying out risk early warning. The problem that whether the network attack exists or not cannot be accurately judged in the prior art is solved, rapid detection of the network attack behavior is achieved, and the effect of improving the network security level is achieved.

The invention discloses a UNET3 + and full-connection conditional random field fusion-based post-earthquake landform segmentation method. The method comprises the steps of S1, collecting historical post-earthquake related image data; s2, data preprocessing; s3, constructing an improved UNET < 3 + > model; s4, constructing an improved full-connection conditional random field model, inputting data obtained by the improved UNET3 + model as a first-order potential function, introducing a Gaussian kernel function including two linear combinations, namely a shape kernel and a smoothing kernel second-order potential function, and respectively controlling the closeness degree and the similarity degree of two pixels and the overall smoothness degree of the image; the method comprises the following steps of: introducing and adding region constraint, improving into a full-connection conditional random field, generating each superpixel through a mean-shift segmentation algorithm, then calculating a posterior probability mean value, and further correcting a classification result of each pixel; and S5, evaluating the segmentation effect of the output of the improved full-connection conditional random field model.

The invention relates to an image classification method based on a multi-core dense connection network. The method comprises the following steps: S1, establishing an image set; S2, constructing a multi-core density connection network model, wherein the multi-core density connection network model comprises an intensive connection unit, an attention unit and a classification unit, the intensive connection unit comprises at least two intensive connection modules, each intensive connection module comprises a plurality of bottleneck layers, each bottleneck layer comprises two convolution layers which are arranged in sequence, and each convolution layer comprises at least two convolution layers which are arranged in sequence, wherein the convolution kernels of the second convolution layer in thebottleneck layers in different dense connection modules are different in size; S3, training the multi-core dense connection network model to obtain a trained multi-core dense connection network model; and S4, inputting the test set into the trained multi-core dense connection network model, and outputting an image classification result. Compared with the prior art, the method has the advantages that depth features of different scales on the extreme image can be effectively extracted through convolution kernels of different sizes, and a better classification effect is achieved.

The invention discloses an obstacle avoidance method, device and equipment and a storage medium, and the method comprises the steps: carrying out the clustering operation of to-be-detected laser radar data in response to the obtained to-be-detected laser radar data, and obtaining at least one to-be-detected clustering data set; for each to-be-detected clustering data set, determining human leg feature data based on each to-be-detected laser radar data in the to-be-detected clustering data set; inputting the human leg feature data and the to-be-detected laser radar data into a target classification model to obtain a classification result which is output by the target classification model and corresponds to the to-be-detected clustering data set; and if the at least one classification result contains the human leg classification label, controlling the robot to execute obstacle avoidance operation based on a preset human leg obstacle avoidance strategy. According to the embodiment of the invention, the problem that the existing robot laser radar technology cannot identify human legs is solved, the risk of collision between the robot and people is reduced, and the obstacle avoidance performance of the robot is improved.

The invention relates to a sea island remote sensing image geological classification method based on a Deeplabv3 + network model, and the method is technically characterized in that the method comprises the steps: obtaining an original remote sensing image of a coral island, and carrying out the data preprocessing of the original remote sensing image; establishing a geological classification system of the coral reefs, and performing feature extraction and classification on the preprocessed remote sensing images to obtain a geological classification data set of the coral reefs; training the Deeplabv3 + convolutional neural network and the ResNet residual network model by using the geological classification data set of the coral island reef to obtain a trained identification and classification model; and performing prediction by using the trained identification and classification model to obtain a sea island reef remote sensing geological classification result map. The method is reasonable in design, is used for high-resolution remote sensing image classification of ocean coral reefs, has relatively high classification precision, can better distinguish details among different ground features, has the overall classification precision and the Kappa coefficient of 97.57% and 0.9643, and can be widely applied to remote sensing geological classification of sea reefs.

A garbage classification and recognition system for a smart city comprises an acquisition module used for acquiring characteristic information of to-be-classified garbage, a main control module used for recognizing the to-be-classified garbage according to the characteristic information to obtain a classification result, an execution module used for sorting the to-be-classified garbage according to the classification result and converting the to-be-classified garbage into classified garbage, a monitoring module used for monitoring the classified garbage to obtain the garbage heap amount, and a management module used for generating a scheduling instruction according to the garbage heap amount. According to the garbage classification and recognition system and method for the smart city, the to-be-classified garbage can be automatically recognized, classification of the to-be-classified garbage is completed according to a recognition result, then recycling is conducted according to the classification result, and the accuracy and efficiency are higher.

The invention relates to the technical field of image processing, in particular to an image classification model construction method and device, an image classification method and device and electronic equipment, and the construction method comprises the steps of obtaining a first classification model and a second classification model, where the first classification model and the second classification model are obtained by training a sample image set, the complexity of the second classification model is greater than that of the first classification model, and the sample image set comprises sample images of a target category and sample images of other categories; connecting the first classification model with the second classification model by using a threshold judgment module to obtain an image classification model, where the threshold judgment module is used for determining whether the second classification model needs to be started based on the output of the first classification model. According to the invention, for the images which can achieve a better classification result by using the first classification model, the second classification model does not need to be started for classification, fusion based on lightweight and heavy models can be realized, the image classification time is greatly shortened, and high-efficiency image classification is realized.