626results about How to "Reduce computational overhead" patented technology

The invention relates to the technical field of environmental sound event detection, in particular to an environmental sound event detection method based on multi-modal data fusion, which utilizes a microphone array, a camera and a sensor to synchronously acquire audio, video and environmental data to form a multi-modal data source. And the collected multi-modal data is preprocessed, so that the data quality and the accuracy of subsequent analysis are improved. And key features are extracted from the processed data, and the features are fused to form multi-modal features. And a deep learning model is constructed, and multi-modal features are used for training, so that a model capable of identifying an environmental sound event can be obtained. And performing optimization such as pruning, quantification and knowledge distillation on the trained model, and deploying the model on edge equipment to realize real-time preliminary feature extraction and event detection. The problems that an environment sound event detection method in the prior art is poor in robustness, low in real-time performance and limited in generalization ability are solved.

The invention discloses a robot pose estimation method based on laser point cloud and visual SLAM, and relates to the technical field of robot sensing and positioning, and the method comprises the steps: in a data collection and integration link, installing a specific camera and a radar to collect and compress data; a data preprocessing and feature strengthening stage: processing the image and the point cloud to improve the quality and extract the features; a cross-modal network is constructed through feature fusion and accurate matching, and a special matching mode is adopted; space-time calibration and coordinate unification are realized by means of a high-precision circuit; multi-modal fusion and reinforcement learning are applied to loop detection and intelligent optimization; correcting filtering and visualizing uncertainty by combining pose smoothing and intelligent output with a prediction model; according to the method, the pose estimation precision, reliability and environmental adaptability of the robot are effectively improved, resource utilization is optimized, the pose uncertainty is visually evaluated, and the overall performance and the intelligent level of the robot are improved.

The invention discloses a satellite remote sensing image cloud classification method based on space-time coding guidance, and particularly relates to the technical field of satellite image processing and deep learning, and the method comprises the following steps: carrying out the preprocessing of the original data of a geostationary meteorological satellite and a Cloud Sat satellite; preprocessing operation such as cutting, projection conversion and feature selection is carried out on the geostationary meteorological satellite data, a top cloud type is selected for the Cloud Sat data to serve as a label, and space-time matching is carried out on the processed geostationary meteorological satellite data and the Cloud Sat data to obtain a training and testing data set; and establishing a GTCCT model, performing time and geocoding on the input data, inputting the coded feature data into a network for training, and finally obtaining the GTCCT model for cloud classification.

The invention relates to the technical field of Internet of Things security, and provides a security authentication method and system for an Internet of Things device, a medium and a device, and the method comprises the steps: responding to an authentication request of the Internet of Things device, the authentication request comprising a unique identifier, a current timestamp and a comprehensive feature vector of the Internet of Things device; after the current timestamp and the comprehensive feature vector are connected, generating a random challenge value through a random function and sending the random challenge value to obtain a returned Internet of Things equipment response value; after the authentication key, the comprehensive feature vector and the random challenge value are connected, an expected response value is generated through a secure hash function, and whether the Internet of Things equipment is successfully authenticated or not is judged through comparison; and if the authentication is successful, generating a session key, connecting the authentication key with the session key, and updating the authentication key through a secure hash function. The common security threats such as replay attack and man-in-the-middle attack are effectively prevented, and the uniqueness and security of each authentication are ensured.

The invention discloses a micro-expression recognition method and system based on a state space model and double-flow fusion, and relates to the technical field of expression recognition, and the method comprises the steps: extracting a starting frame image and a top frame image of a micro-expression continuous video frame sequence; inputting the initial frame image and the top frame image into a micro-expression recognition model based on a double-flow fusion network, extracting an optical flow image, and performing face key point detection and clipping operation on the top frame image and the optical flow image respectively; carrying out image block embedding operation to obtain texture features of the top frame image after cutting and motion features of the optical flow image; the two features learn the global dependency relationship and the local correlation between the image blocks from the positive direction and the negative direction through a parallel multi-layer bidirectional Mama fusion module, and deep image features and motion features are extracted after feature fusion; and the deep features of the two modes are interacted through a double-flow cross attention classification module, comprehensive features are learned, and a plurality of classifiers are combined to identify and output the micro-expression type more accurately.

The invention relates to the field of graphic rendering, in particular to a high-efficiency graphic rendering system based on artificial intelligence core computing power, which comprises an intelligent task decoupling module for performing task decoupling on a graphic rendering task; the AI enhanced texture generation optimization module is used for learning a dynamic scene change rule to generate a texture map in real time, reducing calculation overhead in a repeated texture region by using a multiplexing algorithm, and performing super-resolution compensation on low-resolution textures by using a deep convolutional neural network; the AI driving illumination and shadow calculation module is used for predicting an illumination transmission path in a scene by using AI and reconstructing a dynamic shadow effect so as to realize high global illumination and shadow simulation; the task result synchronization module is used for synchronizing rendering data output by each subtask; according to the method, the high-quality texture map is dynamically generated, super-resolution compensation is carried out on the low-resolution texture, repeated calculation is avoided through a texture multiplexing algorithm, the calculation overhead is remarkably reduced, and the real-time rendering efficiency of a large-scale complex scene is improved.

The invention discloses an intelligent question and answer method and system based on semantic vector optimization and dynamic prompt, and the method comprises the steps: inputting a target question into a trained question and answer model, and obtaining an answer corresponding to the target question; wherein the trained question and answer model comprises an input layer, an embedded layer, a semantic vector generation module, a feature compression module, a dynamic prompt generation module, an answer generation module, a linear layer and an activation function layer which are connected in sequence; wherein the output end of the feature compression module is further connected with the input end of the RAG retrieval module, and the output end of the RAG retrieval module is connected with the input end of the dynamic prompt generation module; according to the trained question and answer model, the training process comprises the steps that a semantic vector generation module is constructed, and the semantic vector generation module is pre-trained; and setting the pre-trained semantic vector generation module in the question and answer model, training the question and answer model, and when the total loss function value does not decrease any more, stopping training to obtain a trained question and answer model.

The invention discloses an unmanned aerial vehicle target detection method based on feature fusion and detection head collaborative optimization, and the method comprises the steps: obtaining a multi-scale feature map through employing a backbone network formed by stacking a convolution module CBS and a residual module C2f, sequentially transmitting the feature maps of various scales to a multi-level feature fusion module (ML-FFM), carrying out the feature fusion from a deep layer to a shallow layer, and carrying out the detection of a target target. Semantic information of shallow layer features is enhanced; inputting the fused feature map into a refined feature enhancement module (RFEM), eliminating position conflicts and redundant information based on a space-time attention mechanism, and optimizing details and positioning capability of deep features; and the enhanced feature map is sent to a prediction head based on Transform for detection, and target classification and positioning of the aerial image of the unmanned aerial vehicle are completed. According to the method, the problems that the unmanned aerial vehicle aerial image is large in scale change, and small targets are high in proportion and dense in distribution are solved, so that the precision of target detection in the unmanned aerial vehicle aerial image is improved. The modules proposed in the method can significantly reduce the calculation amount, and are suitable for the real-time detection scene of the unmanned aerial vehicle.

The invention provides a cross-regional air conditioner regulation and control strategy modeling method and a cross-regional air conditioner regulation and control method and system, and is applied to the technical field of intelligent regulation and control of air conditioners. The method comprises the steps that clustering processing is conducted on air conditioner load data of users in all regions, and a clustering result is obtained; constructing a regional cold supply system simulation model corresponding to the clustering cluster based on the users in the same clustering cluster in the clustering result; the operation safety of an actual regional cooling system is considered, modeling is carried out on the regulation and control process of the simulation model based on the Markov decision process, and an optimization strategy model is constructed by taking expected return maximization of a value function corresponding to an operation strategy of the simulation model as a target; and training an optimization strategy model based on simulation operation data of the simulation model. According to the method, the problems that in the process that the DCS participates in power grid load regulation and control, a trial-and-error training-based model is prone to system damage in practical application, and single-model regulation and control cannot guarantee the load regulation and control effect of a whole power grid are solved.

The invention relates to the technical field of semiconductor device parameter extraction, in particular to a semiconductor device parameter extraction method based on a reinforcement learning enhanced equalization optimizer, and the method comprises the steps: constructing a semiconductor device model, and determining input, output and optimization targets based on the semiconductor device model; and starting a parameter extraction algorithm based on the input, the output and the optimization target, and obtaining the optimal parameter of the semiconductor device, the parameter extraction algorithm being constructed by combining reinforcement learning with an equalization optimizer. According to the method, the efficiency and the precision of MOSFET parameter extraction can be improved by adaptively adjusting the iteration parameters of the optimization algorithm.

The invention relates to a three-dimensional occupied grid network sensing method, which comprises the following steps of: acquiring a multi-view camera image shot by automatic driving of a vehicle, inputting the multi-view camera image into an image processing backbone network, and extracting two-dimensional image features; performing view angle conversion on the two-dimensional image features to obtain three-dimensional body features; constructing matching cost body features according to the geometric correspondence of the sampling points at different moments; a convolutional neural network is adopted to extract weights containing depth correction information from the matching cost body features, the weights act on the three-dimensional body features at the current moment, and the three-dimensional body features after depth information correction are obtained; and inputting the three-dimensional body features into an occupied grid decoder, projecting the three-dimensional body features to a discrete semantic space, and generating a three-dimensional semantic occupied grid. Compared with the prior art, by constructing the matching cost body, more accurate depth information and geometric understanding can be obtained, the calculation overhead under a long time sequence is reduced, and accurate and efficient modeling of the automatic driving environment can be realized.

The invention discloses a method and a device for detecting and positioning surface defects of a PCB (Printed Circuit Board). The method comprises the following steps: acquiring a reference image and a sample image of the PCB; processing the light and shade distribution of the sample image to obtain a sample image with uniform illumination; correcting a pixel difference between the sample image and the reference image to enable the reference image and the sample image to achieve pixel-level alignment; carrying out defect detection index calculation on the reference image and the sample image according to regions, and detecting and positioning defects. According to the method, the detection accuracy and efficiency are improved, the hardware cost and the calculation overhead are reduced, and deployment and implementation in the actual production and manufacturing process are easy; meanwhile, due to the high automation attribute, the detection reliability and the flexible adjustment capability, the method has wide application prospects and practical value.

The invention discloses a double neural network model predictive control method for a robot flexible joint, and belongs to the technical field of robot intelligent control, and the method comprises the steps: carrying out the modeling of a nonlinear dynamic system of a flexible joint robot, and building a dynamic model; setting a GRU neural network model to approach a dynamic model of the system; based on a GRU neural network model, constructing a hybrid control NNMPC controller including position, speed and torque; optimization of the NNMPC controller is optimized on the basis of a nonlinear optimization solver; and setting an NARX neural network model to approach a solution of an NNMPC controller, and forming a dual-neural network model predictive control method. According to the double-neural-network model prediction control method for the flexible joint of the robot, two neural networks are introduced to work cooperatively, one neural network is used for approaching the dynamics of the system, the other neural network is used for accelerating the optimization solving process of the model prediction controller, the calculation overhead possibly caused by a traditional method in a high-dimensional nonlinear system is reduced, and the calculation efficiency is improved. Therefore, real-time control is realized.

The invention discloses a liver tumor segmentation method based on SCAF-TransUNet, and the method comprises the steps: firstly, segmenting a liver image into semantically related superpixel regions through SLIC superpixel preprocessing, enabling a model to be more accurately focused on a tumor and other key regions, reducing the calculation redundancy, and improving the efficiency and precision of feature extraction; secondly, a CECA module introduced in jump connection replaces a traditional full connection layer through one-dimensional convolution, the attention mechanism between channels is simplified, capture of the channel dependency relation is enhanced, and meanwhile the feature extraction precision is kept; finally, the FPN is added in the decoder, and the fusion capability of the multi-scale features is further enhanced. According to the FPN, feature maps with different resolutions are fused layer by layer, so that the model can effectively combine global semantic information and local detail features before up-sampling, and the processing capacity for complex liver boundaries and small tumors is improved. Finally, the improvements significantly improve the accuracy and robustness of the liver tumor segmentation task.

The invention discloses an unmanned aerial vehicle group distributed management method based on a mobile active secret sharing technology, which supports off-line of a trusted institution and comprises the following steps: the trusted institution TA generates public parameters; a mobile base station MS and an unmanned aerial vehicle UAV participating in communication respectively register with a trusted mechanism in a registration stage, and acquire information such as a secret key and a certificate; the UAV joins the MS and is administered by the MS, and after mutual identity authentication, the MS distributes secret shares to the UAV; performing identity authentication on the unmanned aerial vehicle group; the unmanned aerial vehicles of the two groups share the share, group fusion is achieved, and after a new share is obtained, each unmanned aerial vehicle updates the certificate of the unmanned aerial vehicle. According to the method, the sub-secrets of the unmanned aerial vehicles are autonomously updated through a mobile active secret sharing technology, and fusion of completely decentralized unmanned aerial vehicle groups is realized. Secondly, an efficient certificate self-updating method is designed. The unmanned aerial vehicle can update the certificate only by calculating a threshold signature. Detailed safety certification and comprehensive experiments show that dynamic autonomous management of the distributed unmanned aerial vehicle group can be realized, and the method has good safety characteristics and relatively low calculation overhead.

The invention relates to an Internet of Vehicles multi-agent cooperative sensing method and system. The method comprises the following steps: constructing a collaborative sensing network, wherein the collaborative sensing network comprises a self-agent and a plurality of collaborative agents; acquiring and processing sensing data through the collaborative sensing network; performing feature extraction to obtain intermediate features; self-adaptive sparsification is carried out to obtain sparse features, and the sparse features are compressed and transmitted to a self-agent; performing time sequence feature enhancement on the features of all the agents at the self-agent end; fusing the features to obtain fused features; and constructing an Internet of Vehicles perception model, and realizing perception by the detection model according to the fused features. According to the method, the calculation complexity of traditional global attention is reduced from the square level to the linear level through an adaptive sparsification mechanism, the calculation overhead is remarkably reduced while the multi-agent feature interaction precision is kept, and the method is more suitable for real-time operation on the vehicle-mounted edge equipment with limited resources.

The invention discloses a capacitive touch screen touch control adjusting method and device and electronic equipment, and relates to the field of display screens. The method comprises the following steps: acquiring environmental parameters of a plurality of touch points of the capacitive touch screen; determining a calibration touch threshold value corresponding to each touch point according to the environmental parameters; obtaining the touch signal intensity of each touch point; the touch signal intensity of each touch point is compared with a corresponding calibration touch threshold value, the touch state of each touch point is obtained, and the touch state comprises an untouched state and a touched state; generating a first compensation touch point based on the touch state of each touch point, and generating a second compensation touch point based on the touch state of each touch point; and based on the touch state of each touch point and the compensation touch point, generating a touch event corresponding to the touch operation, and sending the touch event to the target program to update a user interface corresponding to the target program. By implementing the technical scheme provided by the invention, the accuracy of touch recognition is improved.

The invention discloses a multi-scale Transform image rain removal method based on residual channel prior guidance, and the method comprises the steps: obtaining a rain image, inputting the rain image into a trained image rain removal processing network, and obtaining a rain-removed image. The training process of the image rain removal processing network comprises the following steps: acquiring a rain image sample and a corresponding real rain-free image; extracting a prior feature map of the rain image sample by using a residual channel prior extraction module; extracting features of the rain image sample by using a context feature extraction module to obtain shallow layer features of rain fringes on different scales; performing feature fusion to obtain a first comprehensive feature map; processing the first comprehensive feature map by using L recursive Transform extraction modules (RSTMs) to obtain an image after rain removal; and calculating the loss between the image after rain removal and a real rain-free image corresponding to the input rain image sample, and optimizing network parameters through back propagation of the loss. According to the invention, rain stripes can be removed from different directions and scales, so that the rain-removed image is more consistent with a real image.

The invention discloses an NURBS curve robot trajectory offset point focusing fitting method based on adaptive optimization, which remarkably improves trajectory precision and calculation efficiency through adaptive optimization focused on an error offset point. The method comprises the steps of initial path planning, NURBS curve fitting, error evaluation and offset point extraction, adaptive optimization and real-time feedback adjustment. In error evaluation, double indexes of a root-mean-square error and a maximum absolute error are adopted, an offset point is accurately identified, a control point, a weight and a node vector are focused and optimized by using a genetic algorithm, and local optimum is effectively avoided. A regularization item is introduced into a fitness function, curve complexity and precision are balanced, and smoothness and feasibility of a track are achieved. And finally, real-time adjustment of the trajectory is realized through a multi-feedback mechanism, and the robustness of the robot in a dynamic environment is improved. Compared with the prior art, the precision, the real-time performance and the environmental adaptability of track generation are realized, and particularly, an efficient and accurate track fitting effect is achieved in a complex dynamic environment.

The invention discloses an image retrieval system and method based on color moments and deep learning features. The system comprises an image feature extraction module, an index construction module, a retrieval trap door generation module and an image retrieval module. The image feature extraction module extracts color moment features and deep learning features of images in the image data set to obtain image multi-level features; the index construction module is used for constructing an index tree from bottom to top for color moment clustering; the trap door generation module extracts hierarchical features of the query image to generate a retrieval image trap door; the image retrieval module traverses the index tree from top to bottom according to the retrieval trap door to obtain a similar image candidate set, calculates the similarity between the depth feature of the image in the candidate similar image set and the depth feature of the query image, and outputs top-k images similar to the query image; according to the method, the color moments of the images are extracted and clustered, and the index tree is constructed, so that the similarity between the comparison images and the query images does not need to be calculated one by one during retrieval, and the image retrieval efficiency is improved.

The invention discloses a lightweight rice disease prediction method based on SCL-YOLOv8n, and belongs to the technical field of rice disease prediction. The method comprises the following steps: constructing a rice disease data set; performing enhancement, labeling and division processing on the rice disease data set; in a GSConv module of a Slim-Neck network structure, before Shuffle operation, lightweight convolution is introduced to be connected with an original path in parallel to form a GSLConv module; the method comprises the following steps: based on a CSP-RFA structure, dividing data, and distributing the data to two branches to form an RFCA-CSP module; a lightweight sharing enhanced detection head LSCSBD is adopted; an SCL-YOLOv8n model is constructed according to the model; taking the processed data set as the input of an improved model SCL-YOLOv8n, and training the improved model SCL-YOLOv8n by using the training set; in the aspect of rice disease prediction under the complex background, the parameter quantity and the calculation complexity are reduced, the precision is improved, and an effective reference is provided for rice disease prediction under the complex background.

The invention discloses a lightweight deployment digital human system, and relates to the field of large model deployment, and the system comprises a cloud digital human module, a local digital human module, a local digital human model deployment module and a cloud edge cooperation module. The cloud digital human module is used for deploying the trained digital human model at the cloud; the local digital human module is used for carrying out compression optimization on the digital human large model to obtain a target local digital human model, and carrying out function splitting and recombination on the target local digital human model according to a local equipment demand to obtain a target sub-model; and the cloud edge collaboration module is used for establishing connection between the cloud digital human model and a target sub-model in the local equipment, dynamically dividing task boundaries of the digital human model and the target sub-model according to the task characteristics of the to-be-processed task and the resource use condition of the target sub-model, and performing task allocation on the to-be-processed task according to the task boundaries. The problem that local equipment deployment resources are limited is solved.

The invention relates to the technical field of computer data management, and discloses a data management method and system capable of reducing computing power occupancy, and the method comprises the steps: S1, based on data distribution analysis, employing a sparse representation algorithm and a principal component analysis method, carrying out the operation elimination of the redundancy of high-dimensional data features through setting an L1 regularization constraint condition, and obtaining an L1 regularization constraint condition; decomposing the data correlation matrix in combination with a matrix operation mode, extracting feature vectors with top-ranked contribution values, and converting the feature vectors into a sparse low-dimensional matrix; and generating a sparse low-dimensional data matrix. Through combination of sparse representation and principal component analysis, L1 regularization is creatively applied to perform feature screening on high-dimensional data, redundant data items are removed, and principal component features with high correlation are extracted, so that the data dimension is effectively reduced, consumption of computing resources is reduced, and meanwhile, main information features of the data are reserved; the efficiency and accuracy of subsequent processing are improved, and distributed scheduling of tasks and efficient utilization of equipment computing power are realized.

The invention relates to the field of heterogeneous federal basic model adjustment, in particular to a clustering-based heterogeneous federal basic model adaptive fine adjustment method and a computer device. According to the scheme, a representative client is selected for each cluster through a multi-factor heterogeneous sensing clustering module, and the representative client selects a corresponding model as a cluster model according to own computing power limitation; searching an optimal sub-model based on the cluster model for all the clients in each cluster through a knowledge perception model architecture search algorithm, and deploying the optimal sub-model at the client; the method comprises the following steps of: uploading parameters to representative clients, aggregating corresponding parameters on the representative clients, issuing the aggregated parameters to clients in clusters, transmitting knowledge of each cluster to a server model through a knowledge transfer module perceived by the clusters, and returning the knowledge of the server model and updating the representative clients of each cluster through reverse knowledge distillation. The method is suitable for self-adaptive fine tuning of the heterogeneous federal basic model.

The invention discloses an aerial view target detection method and system, a storage medium and equipment, and belongs to the field of BEV visual detection map reconstruction.The method comprises the steps that S1, images from different camera visual angles are input, and features of the different camera visual angles are extracted through a backbone network; s2, redesigning a time encoder through deformation and splicing by using long-period time information, and processing the characteristics of different camera visual angles through the designed time encoder; s3, based on a processing result of the step S2, further utilizing dynamic deformable attention to extract spatial features of the image; s4, processing the features obtained in the step S3 by using a feedforward network, and generating a final BEV feature map; and S5, performing target detection by using the final BEV feature map. According to the method, dynamic deformable attention is fused in aerial view angle detection, so that image query is more comprehensive, the calculation overhead is low, meanwhile, time sequence coding is improved, long-time historical information can be better integrated, and the robustness and accuracy of detection are improved.

The invention discloses a fuzzy evolutionary algorithm-based distributed photovoltaic and electrical load decoupling bus load prediction method and system for a power grid planning theme dynamic iteration method, and relates to the technical field of power dispatching. Comprising the steps of obtaining historical operation data of a to-be-predicted bus region; establishing a photovoltaic power generation and electrical load decoupling model of the access interval; dividing the bus load into time intervals according to the installed capacity proportion; constructing an electrical load prediction reference model of the non-accessed section; performing bus load prediction based on the decoupling model; and performing accuracy evaluation and iterative optimization on the prediction model. According to the method, the reliability and accuracy of prediction are ensured by acquiring the historical operation data and establishing the reference model, the data utilization efficiency is improved, the time intervals are divided according to the installed capacity proportion, the decoupling model of the distributed photovoltaic and electrical load is established, corresponding prediction strategies are adopted in different time intervals, and the adaptability of prediction is improved.

The invention discloses a fast scoring spiking neural network architecture optimization method and system, and the method comprises the steps: constructing a data set, and determining a multi-objective optimization equation; initializing a population, and obtaining a sub-network structure score through a training-free method; performing non-dominated sorting and crowding degree calculation, and selecting a tournament; performing selection, crossover and mutation operation; and training the optimal sub-network again. According to the method, the optimal channel pruning structure of the pre-trained spiking neural network can be quickly identified, the optimal pruning SNN model is obtained on the premise of ensuring the minimum precision loss, and the lightweight SNN model can be customized according to user requirements.

The invention discloses a lightweight vehicle intranet identity authentication and key negotiation method based on a PUF (Physical Unclonable Function). The method comprises a system initialization stage, a PUF registration stage, an identity authentication and key negotiation stage and a block chain synchronization stage. The method is oriented to an automobile domain architecture, direct low-cost identity authentication is carried out on the electronic control unit by utilizing the PUF (namely a physical unclonable function), the necessity of installing safety hardware on the electronic control unit can be eliminated, and secret key leakage is prevented; meanwhile, symmetric cryptography is adopted, a hash function, a pseudo-random function and a Lagrange interpolation formula are combined to balance the speed and safety of the key negotiation process of the electronic control unit, and compared with an existing vehicle intranet identity authentication and key negotiation method, the method has remarkable performance advantages.

The invention discloses a GPGPU (General Purpose Graphics Processing Unit)-based singular value decomposition acceleration method and equipment, belongs to the technical field of computer technologies, and is used for solving the problems that when an SVD (Singular Value Decomposition) technology is applied to hardware, a relatively large calculation amount and a large amount of memory occupation can be caused, and different index strategies influence the parallel operation efficiency of the hardware, so that the hardware performance is poor. And the overall performance of hardware is reduced. The method comprises the steps of obtaining an index sequence of a to-be-decomposed matrix; generating a plurality of groups of index sequence pairs based on the index sequence; sending a kernel function kernel to the GPGPU, and traversing each group of index sequence pairs through each block in the kernel function kernel; decomposing matrix data corresponding to the first group of index sequences in the to-be-decomposed matrix through the queue in each block to obtain a first decomposed matrix; and decomposing matrix data corresponding to the second group of index sequences in the first decomposition matrix to obtain a second decomposition matrix, and so on until the last group of decomposition matrix is obtained, thereby obtaining a singular value decomposition result.

The invention relates to a medical image processing technology, and discloses a pan cancer species histopathologic full-slice classification method and device, which is a trained multi-modal pan cancer species classification model based on histopathologic full-slice image data and full-slice diagnosis report pairing data of various tissue parts and various cancer species, and has the advantages of wide application range and high efficiency. And full slices of various pathological tissues can be classified. For classification tasks of different tissue parts, only text prompts (prompts) are used, i.e., a candidate cancer species list is set, and zero-sample full-slice classification can be realized without training. And for rare cancer species, full-slice classification can be carried out. Besides, the model has high expansibility, in order to better adapt to downstream tasks, better classification performance can be achieved by combining slice embedding with a linear classifier for use, freezing a slice embedding part and only finely adjusting the linear classifier, and the requirement for the training data size is low.