38 results about "Cross model" patented technology

The invention relates to the field of data search and provides a cross-model method for searching for a 3D model with 2D pictures. The method comprises following steps: 1) establishing a 3D model library; 2) extracting characteristic vectors; 3) training convolutional neural networks; 4) inputting the characteristic vectors; 5) completing matching retrieval. According to the method, 3D objects and 2D pictures are projected into a new space for measuring the similarity of 3D and 2D; therefore, the problem that the similarity of 3D and 2D data cannot be measured and retrieved due to different data formats is solved; at the same time, the invention also provides an end to end solution scheme which is more efficient than other traditional frames and has better real-time performance.

The invention discloses a cross-model Hash retrieval method on the basis of cooperative training and matrix decomposition. By the aid of the cross-model Hash retrieval method, the similarity between models and the internal similarity of the models can be effectively constrained for unlabeled cross-model data. The cross-model Hash retrieval method includes implementation steps of acquiring original data and carrying out normalization processing on the original data; carrying out cooperative training to obtain constraints between the models; acquiring internal constraints of the models by the aid of neighbor relations; decomposing training data matrixes and adding the constraints between the models and the internal constraints of the models into the training data matrixes to obtain objective functions; carrying out alternate iteration to obtain expressions of basis matrixes, coefficient matrixes and projection matrixes; carrying out quantization to obtain Hash codes of training data sets and test data sets; computing the Hamming distances between every two Hash codes of the data sets; sorting the Hamming distances to obtain retrieval results. The cross-model Hash retrieval method has the advantages that constraints on the similarity between the models of the cross-model data can be obtained by the aid of cooperative training processes, accordingly, the image and text mutual retrieval performance can be improved, and the cross-model Hash retrieval method can be used for picture and text mutual search service of mobile equipment, internets of things and electronic commerce.

The invention discloses a method for automatically recovering cross-model GUI (graphic user interface) test scripts, which is used for carrying out automatic recovery on GUI test scripts recorded on an older-version GUI program aiming at a new-version GUI program. The invention provides a method implemented by using the complementation of two event reachability models so as to describe the GUI structure and event reachability models of an application program more completely and accurately, the legitimacy of statements of the GUI test scripts is examined by using the two event reachability models, and an automatic recovery scheme is given to incorrect GUI test scripts in the new-version GUI program. According to the invention, the legitimacy verification and recovery of scripts are completed by using the collaboration of the two models, and a self-built event reachability model is continuously perfected in the process of recovering without the overmuch confirmations of engineers; and due to the joint analysis and verification of the two models, the integral description on all event flows included in the GUI application program is ensured, and the automation of legitimacy verification and recovery of GUI test scripts is realized.

A system and method facilitating speech detection and/or enhancement utilizing audio/video fusion is provided. The present invention fuses audio and video in a probabilistic generative model that implements cross-model, self-supervised learning, enabling rapid adaptation to audio visual data. The system can learn to detect and enhance speech in noise given only a short (e.g., 30 second) sequence of audio-visual data. In addition, it automatically learns to track the lips as they move around in the video.

The invention provides a calculation method of a pipe joint hoop strain of a tide load sinking pipe tunnel. The method includes steps of, assuming a back silt layer on the pipe joint as an impervious bed, and regarding the change of the tide level as a large area load pO added on the impervious bed, wherein the location of an underground level is the top face of the impervious bed; not considering about influence of tide on the pipe joint structure, calculating water, soil pressure and subgrade reaction acted on the outer wall of the pipe joint; applying a winkler foundation module, regarding the loop strain calculation as the plane strain problem; analyzing the loop stress and deformation of the pipe joint structure by a load-structure method; through the calculation of cross model stress of the pipe joint, acquiring the relationship of the loop stress strain relationship of the pipe joint; finally, through calculation, acquiring the loop strain distribution of inner and outer walls of the pipe joint under one tide level and influence of the loop strain of every point of inner and outer walls of the pipe joint along with the change of the tide level. The calculation result can be used for predicting the loop strain of the pipe joint of the sinking pipe tunnel in operation stage, judging the part where may generate crack, and carrying out key monitoring and prevention.

The invention discloses a method for detecting scouring depths of a soil body around a marine pile foundation. The method comprises the steps: (1) a pile-soil system is built; (2) the pile foundationis divided into n joints through a concentrated mass method to be numbered, the joints are connected through massless elastic rods with the length being d, each joint and a lower connecting rod serveas a unit, and the rod length d serves as the unit thickness; (3) the soil body around the foundation is layered, the thickness of each layer is the same as the unit thickness, namely d, the soil bodyaround the foundation is simplified to be massless spring constraint around the joints, and the elastic coefficient is determined; (4) an overall rigidity matrix and a mass matrix under the pile-soilsystem are built; (5) the inherent frequency and vibration modes of all orders of the pile foundation under the different scouring depths are detected; and (6) the rigidity correction coefficient isobtained through a cross-model and cross-mode method (CMCM method), the joints, namely damaged units, corresponding to mutation positions of the correction coefficient are found, the number of the units between the damaged units and pile foundation units on the mud surface is determined, and the range of the scouring depths is determined by multiplying the number of the units and the unit thickness.

Presenting data from different data providers in a correlated fashion. The method includes performing a first query on a first data set controlled by a first entity to capture a first set of data results. The method further includes performing a second query on a second data set controlled by a second entity to capture a second set of data results. The method includes receiving a selection of one or more results from the first data set. The method further includes using the one or more selected results, consulting a relationship ontology that correlates data stored in different data stores controlled by different entities, to identify one or more relationships between data in the selected results set and the second data set.

The invention discloses a multi-particle hybrid modified thrombus simulation method based on a Cross model. A multi-particle hybrid modified SPH method based on the Cross model is utilized to simulatethe blood, so a problem that the phenomenon that large-mass particles sinks downwards and small-mass particles rise upwards in the simulation process does not conform to the normal blood physiological phenomenon is solved, a particle-by-particle polymerization method is utilized to simplify a thrombus model to simulate thrombus formation, and a problem that a traditional method can not simulate thrombus formation of the blood is solved. The method is advantaged in that the blood can be more realistically simulated, and thrombus formation of the blood can be further simulated. Through correctly simulating thrombus formation, the thrombus study is facilitated, the method can be applied to medical applications such as virtual surgery and medical assistant research, moreover, needs of relatedaspects in game and film industries can be satisfied.

The invention discloses a DEA cross model energy efficiency analysis method based on principal component fusion. The method comprises the following steps: obtaining 20 groups of production data of a PTA (purified terephthalic acid ) production device; obtaining input indexes of a phthalic acid production device according to the 20 groups of production data, wherein the input indexes comprise 17 factors influencing acetic acid consumption; obtaining a hierarchical model of the 17 factors according to a preset interpretative structural model; according to a preset analytic hierarchy process andthe hierarchical model, forming fusion data; and obtaining output indexes according to a preset data envelopment analysis cross model and the fusion data, wherein the output indexes comprise tower topconductivity. The method optimizes input data of the data envelopment analysis cross model, and improves distinction degree of analysis results; and besides, by carrying out analysis and evaluation on production efficiency of the PTA production device, relative effectiveness of the production efficiency is obtained, so that energy saving and consumption reducing direction of the PTA production device is obtained, and meanwhile, effectiveness and applicability of the analysis conclusion above can be proved.

The invention relates to a crossing model considering internal cost and supporting rapid way finding. The crossing model adopts a cellular node model, and has a three-layer structure comprising an outermost layer, a middle layer and an innermost layer; the outermost layer is an interface layer and is a dynamic expansion layer, and the quantity of the interfaces of the outermost layer is increased or decreased in the road network logicalization process, wherein the interfaces are divided into incoming interfaces and outgoing interfaces, and are used for preserving all topology connection relation information of one node and other nodes; the middle layer is a channel layer, is a logic layer and is used for indicating correspondence of the interfaces to practical roads; and the innermost layer is an internal cost layer, and is used for preserving the internal connection cost of crossings. The crossing model fully considers the internal passing costs of the crossings, solves the problems of low way finding speed and inconsistence of the way finding result with the practical driving road, and can be used to construct vehicle navigation electronic maps.