237results about How to "Good distinction" patented technology

The invention relates to apparatus and methods for measuring local tissue impedance for subcutaneous tissue surrounding a needle tip inserted into a subject, impedance spectra and/or complex impedance values are determined. The invention applies a monopolar impedance measuring setup with a needle, a current-carrying electrode, an optional reference electrode. The setup is configured to eliminate contributions from the current-carrying electrode in order to measure local impedance of tissue in the close neighbourhood of the needle tip instead of an averaged value over the volume or current path between the needle and the electrode(s). The determined impedance can be correlated with either a tissue type or state, or with a position of the needle tip in the subject, and can thereby provide an insertion history to the operator in the form of impedance or corresponding tissue type as a function of insertion depth or time.

A system and method is described for saving power in a wireless network, using a physical layer address filtering protocol based on a partial address subset of the complete destination MAC address. The system comprises a PHY layer filtering protocol for generating the partial address and writing the partial address into a PHY layer header portion (e.g., PLCP header) of a sending station, or reading the partial address from the PHY layer header portion upon transmission of each frame. A receiving station receives and decodes these PHY layer header portion bits, in accordance with the protocol, and compares whether the subset of bits match that of the stations' own partial address. If a station finds a match, the station then continues further decoding the frame at PHY layer and send the complete frame to the MAC layer for further processing. The stations that do not have a match will not activate their MAC layer components. Thus, the stations of the network will avoid wasting power decoding a significant portion of the complete frame of other stations of the wireless local area networks and unnecessary MAC layer processing. When group addressed, control/management frames or other such frames are detected at the sending station, the address filtering protocol may be “disabled” using a partial address containing a predetermined value (e.g., all zeros).

This relates to methods and devices for improving treatment to a wall, a cavity or passageway with a medical device when used in tortuous anatomy.

A cutter for spine surgery is provided, especially for use in the area of the delicate cervical spine. The cutter includes a cylindrical cutter shank with cutter teeth formed at the distal end thereof. The cutter teeth at the distal end of the cutter shank are formed by grooves in the wall of the cutter shank. The grooves become deeper and expand from the outer radius of the cutter shank towards the distal end such that teeth narrowing towards the distal end with increasing height are formed between them. This guarantees especially gentle cutting in the area of the cervical spine, without surrounding delicate tissue being additionally jeopardized.

The invention relates to a knowledge graph and discloses a knowledge graph representation learning method for integrating text semantic features based on an attention mechanism. The method solves theproblems that semantic features are insufficient due to the fact that a translation model does not utilize description texts of entities and relations, semantic features cannot be fused into entitiesand relations at the same time by a multi-source information embedding method, and the text extraction effect is poor. The method comprises the steps of firstly obtaining and processing description texts of entities and relationships to obtain text semantic features of the entities and the relationships, then constructing a projection matrix of the entities by utilizing the semantic features of the entities and the relationships, projecting entity vectors into a relationship space, modeling in the relationship space by utilizing a translation thought, and carrying out representation learning,so as to model a many-to-many complex relationship. The method is suitable for representation learning of the knowledge graph.

The invention discloses a motion recognition method based on a three-dimensional convolution depth neural network and depth video. In the invention, depth video is used as the object of study, a three-dimensional convolution depth neural network is constructed to automatically learn temporal and spatial characteristics of human body behaviors, and a Softmax classifier is used for the classification and recognition of human body behaviors. The proposed method by the invention can effectively extract the potential characteristics of human body behaviors, and can not only obtain good recognition results on an MSR-Action3D data set, but also obtain good recognition results on a UTKinect-Action3D data set.

The invention discloses an image searching method, which comprises a training part and a searching part, wherein the training part comprises the following steps of: the extraction of characteristic points, the supplementation of the characteristic points and the determination of matching relationships, the generation of similar point set, the clustering of the characteristic point sets and the generation of characteristic vectors of each image in an image database; and the searching part comprises the following steps of: extracting the characteristic points of a picture to be retrieved and generating the characteristic point sets; calculating distances between each characteristic point descriptor vector and corresponding cluster centers, and determining a cluster where a current characteristic point belongs by using a smallest distance; calculating the frequency ni of each cluster where the characteristic points of the picture to be retrieved belong; based on the frequency ni of the clusters where the characteristic points of the picture to be retrieved belong, and the probability logarithm wi of each cluster, generating and unitizing the characteristic vector; and calculating Euler distances between the characteristic vector of the picture to be retrieved and the characteristic vectors of each image in a picture library, and selecting the image output with the smallest distance as a searching result.

The invention discloses a video semantic representation method and system based on a multi-mode fusion mechanism and a medium. Feature extraction: extracting visual features, voice features, motion features, text features and domain features of a video; Feature fusion: performing feature fusion on the extracted visual, voice, motion and text features and domain features through the constructed multi-level hidden Dirichlet distribution topic model; And feature mapping: mapping the fused features to a high-level semantic space to obtain a fused feature representation sequence. The model utilizesthe unique advantages of the theme model in the semantic analysis field, and the video representation mode obtained through model training on the basis of the model has ideal distinction in the semantic space.

The invention discloses an adaptive action recognition method based on multi-view and multi-mode characteristics. The adaptive action recognition method specifically comprises the steps of: preprocessing videos; carrying out multi-view description on a target movement variation change; extracting equal-hierarchical pyramid characteristics; constructing a multi-view depth and RGB (Red Green Blue) model; selecting a multi-view model, deducing and integrating multi-mode characteristic results. According to the adaptive action recognition method, firstly, aiming at the difficulties of illumination variation, shadow and the like usually occurring in a process of recognizing visible image actions, action recognition is carried out on the basis of multi-view and multi-mode characteristics; then aiming at the limitation of the single view, multi-view description in the target movement variation process is provided, and is capable of more completely capturing variation processes of a target in a depth and RGB image sequence; then, the equal-hierarchical pyramid characteristics also have a spatial resolving power and a detail description power, thereby having very good robustness and discrimination property; finally, multi-mode characteristics are adaptively integrated according to the variation change of ambient light, and the performance and the stability of the action recognition method are further improved.

The invention discloses a face recognition method based on distinguishability feature fusion, comprising the following steps: A, cutting a global image and at least two local images in each training sample image; B, carrying out model training on each intercepted image by adopting a multi-loss function to obtain a corresponding model, wherein the multi-loss function is obtained by combining the a-softmax loss function with the center loss function. C, carrying out fusion and dimension reduction of each model obtain by training by using a ternary loss triplet loss function, and obtaining the final depth feature of the training sample image. The technical proposal disclosed in the application can solve the problem of data fusion, face posture and model fusion in the process of face recognition by using CNN, and achieve better face recognition effect.

The invention relates to an audio scene recognition method and device based on long-term and short time feature extraction. The method comprises: preprocessing an inputted to-be-recognized audio signal; d carrying out short-time audio feature extraction on the pre-processed to-be-recognized audio signal and carrying out long-term audio feature extraction; and carrying out long-term and short-timeaudio feature combination of the to-be-recognized audio signal, inputting the features into a classification model and a fusion model, carrying out classification and identification, and outputting anidentification label of an audio scene. According to the invention, the long-term features of the audio scene are combined based on the conventional short-time feature extraction and complex audio scene information can be represented; the features are inputted into the classification model and the fusion model to carry out classification and identification; and the identification label of the audio scene is outputted. The method and device have advantages of high robustness and good distinguishing performance; the overall characteristics of the scene data can be represented to the great extent; and the recognition efficiency and the stability are high.

The invention discloses a semi-supervised learning-based pedestrian detection method. The method includes the following steps that: the training samples of a source image set and the categories of the training samples are obtained, pedestrian labeling is performed on a part of images in a target scene image set, and training samples and sample features corresponding to target scene images are obtained; a decision-making forest is generated through training based on the training samples of the source image set, training samples of which the categories are known in the target scene image set are adopted to screen decision-making trees in the decision-making forest, and after the decision-making trees are reorganized, a new decision-making forest can be generated; the new decision-making forest is adopted to score training samples of which the categories are unknown in the target scene image set, and training samples with high confidence are labeled as pedestrian training samples; the training samples of which the categories are known in the target scene image set and the pedestrian training samples are adopted to train a neural network; and test samples are inputted into the new decision-making forest, test samples with high confidence are made to pass through the neural network, so that a pedestrian detection result is obtained. The semi-supervised learning-based pedestrian detection method is advantageous in high pedestrian detection accuracy.

A system receives, based on processing of an inspected frame of an inspected image generated by collecting signals indicative of a pattern on an article, at least one candidate defect location in the inspected frame. The system defines a candidate patch within the inspected frame. The candidate patch is associated with the candidate defect location. The system identifies at least one similar patch in the inspected frame using a predefined similarity criterion and determines whether a defect exists at the candidate defect location based on a comparison of at least a portion of the candidate patch with at least a corresponding portion of the at least one similar patch.

The invention relates to a real-time online multi-target tracking method adopting a temporal dynamic appearance model. The method comprises the following steps: taking appearance information of each target in a multi-target tracking process as an appearance sequence of temporal orderliness; expressing the temporal dynamic variation characteristics of the appearance sequence by establishing an appearance model with strong discriminating ability by considering the rule of change of the target appearance information in the tracking process following the change of time; supplying accurate appearance matching information related to the data between target and detection results through structural diversity of the modeling of hidden state observation models in a characteristic space; relating the detection results with the target frame by frame, thereby finishing the real-time online multi-target tracking task and accurately real-time tracking a plurality of interesting targets in a tracking video. The tracking property is ensured while calculation efficiency is ultra-high. The real-time online multi-target tracking method can be practically applied to various real-time systems.

The invention relates to a rumor-contained user generated content recognition method and device. The method includes: acquiring a content identification, wherein the content identification is used foridentifying a user generated content to be recognized; querying spread user identifications corresponding to the content identification; querying spread user attributes corresponding to the spread user identifications; acquiring a spread relation between the spread user identifications corresponding to the content identification; extracting spread features from the spread user attributes and thespread relation through a neural network model; and recognizing if the user generated content to be recognized is the rumor-contained user generated content or not according to the spread feathers. The rumor-contained user generated content recognition method and device can improve the recognition efficiency of the rumor-contained user generated content.

The invention discloses an animal voiceprint feature extraction method, device and an electronic device, relates to the technical field of animal identity recognition. According to the invention, animal voiceprint features can be accurately extracted, so as to improve an animal identity recognition effect. The method provided by the invention comprises the following steps: obtaining animal voice data; extracting an animal voice feature vector from the animal voice data; inputting the animal voice feature vector to a convolutional neural network model for training, and then obtaining the animalvoiceprint features for recognizing animal identity.

The invention provides a massive data processing method and a massive data processing device, and aims to solve the problem of unremarkable data discrimination effects caused by higher sparsity of original data. The method comprises the following steps of storing the massive original data into an (m*n)th-order original matrix A, wherein m and n are positive integers, and the original data is operating data for each user; when the original matrix A is subjected to singular value decomposition, distributing the original data in the original matrix A to a plurality of processing nodes for processing; reconstructing a first unitary matrix U, a first diagonal matrix S and a second unitary matrix V, which are obtained by the singular value decomposition, to obtain a corresponding reconstructed matrix B; and clustering data in the reconstructed matrix B to discriminate the data of different types. According to the method and the device, the problem of higher sparsity of the original data is radically solved, so that higher data discrimination performance during subsequent massive data processing is ensured.

The invention relates to a robust image hashing method and device based on Radon transformation and invariant features, and belongs to the field of information safety. In terms of the problem that hashing cannot resist geometric attacks well, normalized preprocessing operation is carried out on images firstly, invariant feature points are generated by utilizing an unchanged centroid algorithm, the circular area around an unchanged centroid is selected, Radon transformation is carried out on the circular area to generate a coefficient matrix, multiple lines of coefficients are selected randomly from a transformation domain by utilizing a chaotic system, robust features of each line are extracted, the features of all lines are combined with the invariant moment features of the whole matrix to generate image hashing, and similarity comparison is carried out by utilizing Euclidean distance. By the adoption of the robust image hashing method and device based on the Radon transformation and invariant features, the problem that the false drop rate rises due to geometric attacks can be solved effectively; the problems that computation complexity is too high and hashing is too long can be solved according to hashing steps and hashing lengths. The method and device can be applied to the field of image content authentication, and can also be applied to image retrieval, image identification and other information safety fields.

The embodiment of the invention provides a determining method and device for user natural attributes. The method comprises the steps of determining a sample containing the natural attributes of a first user and user feature vectors of the first user according to the natural features and behaviors of the first user; normalizing features in the user feature vectors of the first user; obtaining a logistic regression machine study model corresponding to each natural attribute by the sample containing the normalized user feature vectors of the first user; determining user feature vectors of a second user according to behaviors of the second user, wherein the second user is a user with the natural features to be determined; normalizing the features in the user feature vectors of the second user; applying the normalized user feature vectors of the second user to the logic regression machine study model corresponding to each natural feature, and determining the natural attributes of the second user. The accuracy for determining the natural attributes of the user is improved.

The present invention provides a discriminant local information distance preserving projection-based speaker confirmation method and belongs to the voiceprint recognition, pattern recognition and machine learning field. According to the method, speech data are obtained in a training phase; the i-vector of each training speech datum is extracted; the i-vector of each speaker is extracted accordingto training speech data corresponding to each speaker; training is performed, so that a discriminant local preserving projection matrix is obtained; during a speaker confirming phase, speech data to be tested are acquired, one speaker of the training speech data is selected, a distance between the speech data to be tested and the i-vector of the speaker is calculated; and if the distance is smaller than a preset distance threshold value, it is determined that the speech data to be tested belong to the speaker; and the confirmation of the speaker is completed. The method of the invention has high applicability, focuses on heterogeneous neighbor points, enhances the discrimination of the easily-confounded speech of speakers, has better distinguishing capability and improves the accuracy of speaker confirmation.

The invention relates to a filter bank training method. The filter bank training method comprises the steps that first, a training image which has a target position mark is preprocessed to obtain a denoising training image; second, initial clustering is conducted on the denoising training image, and the image is decomposed into K training sets; third, an ideal filter output model is designed according to the target position mark in the training image; fourth, K total filter models are obtained by training according to the ideal filter output model to constitute a filter bank; fifth, whether an image sample set is convergent or not is judged, if yes, the seventh step is executed, and otherwise the sixth step is executed; sixth, whether the convergent frequency reaches a preset threshold value or not is judged at present, if yes, the seventh step is executed, otherwise, classification is conducted again to obtain K new training sets, the K new training sets replace the K training sets, and the fourth step is executed again; seventh, the filter bank is stored, and the training process of the filter bank is completed. The filter bank training method has better distinguishing performance to targets, and improves the accuracy and the precision of positioning to a certain extent.

The invention discloses a target tracking method based on a three-branch neural network, and belongs to the technical field of computer vision. Visual target tracking belongs to video analysis and serves as an important branch in the field of computer vision, and the basic task of the visual target tracking is to predict the position, area and motion track of a target in a video sequence accordingto given position information of the target in an initial frame. The precision and speed of visual target tracking are low, and the visual target tracking is easily affected by shielding, backgroundconfusion, size change, severe appearance change, illumination change and the like. The invention provides a target tracking method based on a three-branch neural network. Compared with the traditional visual target tracking technology, the three-branch neural network is utilized to track the target, so that the target can be represented with high robustness, the obvious change of the appearance of the target can be handled, the background can be well distinguished, and meanwhile, the drifting of the algorithm can be effectively avoided. And the tracking speed is far higher than that of otheralgorithms.