39704 results about "Computer vision" patented technology

The invention discloses a multi-modal medical image data intelligent processing system, relates to the field of medical image analysis, and is applied to multi-modal medical image whole-process analysis of CT, MRI, PET, ultrasound and the like. According to the system, different modal image features are extracted and fused through a cross-modal manifold fusion network; a semantic guidance dynamic registration engine optimizes registration parameters to ensure that the registration error is less than or equal to 1.5 mm; the multi-task collaborative diagnosis network realizes multiple tasks such as disease classification; the clinical knowledge embedding and interpretable module generates a structured report and is in butt joint with an HIS system. Meanwhile, the model is optimized through a federated learning architecture, the adaptability of newly added data is improved by more than or equal to 20%, and intelligent processing and analysis of multi-modal medical images are realized.

The invention relates to the technical field of computer vision and multi-modal learning, in particular to a weak supervision target detection method guided by cross-modal pseudo labels. According to the method, a labeled source domain data set is constructed to train an image classification teacher model, and a teacher-student network structure is constructed; clustering the regional features of the target domain image, allocating pseudo tags to each cluster by optimizing the allocation cost between the source domain category and the target domain cluster, and constructing a pseudo tag pool; and training a student model on the pseudo label pool for region feature detection of the target domain image. According to the method, a cross-modal attention mechanism is introduced, so that more accurate semantic alignment between a source category label and a target domain feature is realized; the stability of label distribution is improved by a structure keeping regular term; the generalization ability of the model is further enhanced by multiple rounds of pseudo-label confidence learning. The method can be widely applied to tasks such as target detection, cross-domain transfer learning and open world recognition, and efficient and accurate weak supervision target detection is realized.

The invention relates to the field of image analysis, in particular to a temporal bone disease classification method and system based on a multi-modal medical image fusion technology. The method comprises the following steps: acquiring a multi-modal image of a patient, performing adaptive distortion correction, and generating a standardized image set; performing layer-by-layer anatomical structure semantic segmentation and multi-modal image fusion on the standardized image set to construct an image fusion framework; according to the image fusion framework, performing intelligent recognition on the fine structure of the temporal bone, and constructing a personalized temporal bone anatomical structure chart; performing tissue function state analysis and digital pathology dynamic simulation based on the personalized temporal bone anatomical structure chart, and constructing a digital pathology model; and performing intelligent pathological feature classification based on the digital pathological model to obtain an intelligent classification report. According to the method, rapid, efficient and accurate temporal bone disease classification is realized.

Some embodiments of a method may include: obtaining information for a three-dimensional (3D) Gaussian model corresponding to a 3D scene, wherein the information comprises a set of attributes of the 3D Gaussian model; parsing the information for a first attribute of the set of attributes, wherein the first attribute corresponds to a position of the 3D Gaussian model; parsing the information for a second attribute of the set of attributes, wherein the second attribute corresponds to a covariance of the 3D Gaussian model; parsing the information for third, fourth, and fifth attributes, wherein the third, fourth, and fifth attributes correspond to first, second, and third sets of spherical harmonics coefficients associated with the 3D Gaussian model; parsing the information for a sixth attribute of the set of attributes, wherein the sixth attribute is an alpha coefficient for the 3D Gaussian model; and rendering the 3D scene using the parsed attributes.

The invention discloses an AI-based animation split script automatic generation and visual preview method and system, and the method comprises the following steps: 1, receiving a natural language script text inputted by a user, the natural language script text comprising scene description, role action, dialogue and shot indication information; step 2, performing semantic analysis and structured analysis on the script text based on a natural language processing technology, and identifying and extracting key narrative elements; by introducing an artificial intelligence technology, end-to-end automatic generation and interactive optimization from a character script to a dynamic split rehearsal video are realized, the system can deeply understand scenes, actions, role emotions and shot languages in the script, corresponding visual elements are automatically matched and generated, and the dynamic split rehearsal effect is improved. And the timeline and the rhythm conforming to the film and television grammar are constructed, so that the efficiency and the consistency of the split creation are greatly improved, and the professional threshold and the manufacturing cost are reduced.

Embodiments of the present disclosure relate to rendering and inverse rendering using one or more generative models. “Rendering” refers to the process of generating a final visual image, video frame, or animation from a 2D or 3D model. “Inverse rendering” is a process that involves deducing or estimating the properties (e.g., material maps or other properties such as geometry, lighting, and textures) of a scene from observed images or visual data. Essentially, it aims to reverse the traditional rendering process. Various aspects of the present disclosure introduce editable light and material controls into generative models to allow for artistic creation. Various embodiments integrate generative models as a renderer for classic rendering pipelines to upcycle and enhance the style of rendered content.

The invention relates to a multi-mode large language model fine tuning method, system and device and a medium, and belongs to the technical field of artificial intelligence and computer vision crossing. The fine tuning method comprises the steps that an original business scene image is acquired and preprocessed, and a preprocessed image is obtained; performing bounding box coordinate labeling and semantic label definition on the entity target in the preprocessed image through a labeling tool, and outputting a structured labeling file; based on the preprocessed image and the structured annotation file, constructing a training sample set comprising multiple rounds of image-text dialogues; loading the pre-trained multi-modal large language model, configuring low-rank matrix decomposition parameters, and generating a fine tuning instruction set; and inputting the training sample set into a pre-trained multi-modal large language model, carrying out joint training operation based on the fine tuning instruction set, and outputting the fine-tuned multi-modal large language model. According to the method, the identification accuracy, the interaction capability and the system availability of the visual question-answering system in an actual application scene are improved.

The invention discloses an end-to-end automatic driving method based on dynamic multi-modal fusion in a complex scene, and belongs to the technical field of automatic driving. In order to solve the problems of sensor perception deficiency, cross-modal feature mismatching, unstable trajectory planning and the like easily occurring in night, low-illumination and complex dynamic environments in the existing end-to-end automatic driving method, texture details of a camera mode and geometric structure features of a laser radar mode are respectively enhanced through a double-flow feature refining mechanism; the characteristic difference between different modes is relieved; an information-driven dynamic fusion strategy is designed, the fusion weight is adaptively adjusted according to scene factors such as environment illumination and obstacle density, and the scene sensitivity and discrimination ability of the model are improved; asymmetric convolution and a low-rank-sparse decoupling technology are introduced, multi-order reconstruction of key channels is carried out on the multi-modal features, and the path change modeling capability is enhanced; and in combination with time sequence dependence of waypoints, outputting a future trajectory through an autoregression decoder to realize high-precision trajectory prediction and stable decision control.

The invention provides a medical image segmentation method based on geometry and topology cooperative guidance. The medical image segmentation method comprises the following steps of image preprocessing and data enhancement; a shared encoder; a dual-path cooperative decoder; carrying out multi-mode deformation iterative refining; and a multi-objective composite loss function and an optimization strategy. The method has the beneficial effects that the performance can be remarkably improved: through a unique geometry and topology collaborative refining mechanism, the segmentation precision and the boundary definition are far superior to those in the prior art, the topology correctness of an anatomical structure can be actively maintained and repaired, clinically unacceptable errors are remarkably reduced, and the reliability of a result is improved; in addition, operation can be simplified, stability and generalization are enhanced, and advanced application is promoted.

The invention discloses a brain tumor multi-mode large model construction method, device and equipment and a storage medium, and is applied to the technical field of brain tumor imagines.The method comprises the steps that pixel-concept level alignment is conducted on a multi-mode MRI image and a pathological text; constructing a multi-modal feature fusion network for fusing image features and text features by adopting an attention mechanism of pathology perception and combining medical semantic information; training the multi-modal feature fusion network to generate an analysis report and a segmentation result; according to the technical scheme of multi-task cooperation, cross-modal pathological semantic accurate alignment, pathological knowledge graph injection and lightweight and continuous optimization parallelization, full-process coverage of brain tumor accurate segmentation, analysis report generation and prognosis prediction is achieved, the problems that a traditional model lacks pathological semantic support and is insufficient in clinical adaptability are solved, and the clinical adaptability of the traditional model is improved. And the deployment feasibility and the dynamic optimization capability are also considered.

The invention relates to the field of commodity visualization, in particular to a commodity display interaction visualization method and device. The method comprises the following steps: collecting a multi-azimuth image of a commodity, carrying out three-dimensional texture modeling, and constructing a three-dimensional texture mapping model; performing material light rendering on the three-dimensional texture mapping model to generate a material rendering result; collecting an environment detection image of a commodity display environment, and performing environment illumination adaptation compensation on a material rendering result to obtain an illumination compensation rendering commodity; carrying out attribute information visual layout on the illumination compensation rendering commodity to obtain a commodity visual space; and carrying out interaction response animation analysis according to the commodity visualization space, carrying out multi-target parallel rendering, and executing commodity interaction visualization operation. The form and surface details of the commodity in the real world are accurately restored, the visual reality sense is improved, and the interactive experience feeling of browsing the commodity by a user is enhanced.

The present invention sets forth techniques for performing style transfer from multiple supplied style images to a supplied content image to generate novel images that include style elements from the multiple supplied style images and content elements from the supplied content image. The techniques include guiding one or more self-attention and cross-attention layers included in a machine learning model based on the multiple supplied style images, such that content elements and style elements included in the style images are not entangled when generating the novel images. The techniques also distill a small subset of representative attention map values from multiple style images, improving performance while reducing computational costs compared to processing all attention map values from the multiple style images.

The invention discloses an industrial image anomaly detection method based on deep learning, and particularly relates to the technical field of industrial visual detection. The problems of high false alarm rate, fuzzy fine defect positioning, insufficient real-time response capability, difficulty in model increment updating and the like caused by data distribution drift in an industrial scene are solved. According to the method, robust features are extracted through a multi-scale feature fusion auto-encoder, and a dynamic memory bank is constructed to update a normal sample prototype online; a dual-path detection mechanism is adopted to cooperate with a pixel-level reconstruction error and attention weighted feature matching deviation; efficient edge reasoning is realized in combination with block parallel processing and model compiling optimization; and designing an elastic incremental learning framework to prevent disastrous forgetting. And finally, false alarms caused by environmental changes are reduced, accurate positioning of pixel-level defects is realized, millisecond-level detection requirements of high-resolution images are met, safe and efficient model online evolution is supported, and adaptability and reliability of an industrial quality inspection system are comprehensively improved.

The invention discloses an illness state tracking management system and method based on a lingual face diagnosis instrument, and belongs to the technical field of traditional Chinese medicine tongue diagnosis and modern information technology fusion. The system comprises a multi-modal data acquisition module, a data processing and analysis module and an augmented reality visualization module; according to the method, illness state tracking is achieved through the steps of multi-modal data acquisition, data preprocessing and feature fusion, personalized digital twinborn model establishment, augmented reality visualization presentation and the like, multi-dimensional data such as tongue picture macroscopic features and tongue surface microorganism distribution can be integrated, dynamic association between the data is revealed, the health state and the intervention effect are visually displayed, and the method is suitable for being popularized and applied. The method is suitable for the fields of traditional Chinese medicine health management and chronic disease monitoring.

The invention discloses a three-dimensional attitude estimation method combining global modeling and local refinement, which comprises the following steps of: firstly, extracting a two-dimensional attitude sequence by using a human body video data set; secondly, inputting the two-dimensional attitude sequence into a structural modeling main branch, modeling a spatial topological relation and a time sequence dynamic state between joints, and outputting a global three-dimensional attitude sequence; and inputting the two-dimensional attitude sequence into a local refining branch, modeling dynamic change and detail information of a local area, and outputting a local three-dimensional attitude sequence. And finally, fusing the global three-dimensional attitude sequence and the local three-dimensional attitude sequence, generating a three-dimensional attitude sequence output, and completing three-dimensional attitude estimation. According to the method, the problem of insufficient cross-frame information transmission in a traditional method is relieved, and the accuracy and robustness of attitude estimation in a dynamic complex scene are remarkably improved.

The invention relates to the technical field of precise geometric quantity measurement, and discloses an electronic component size and contour automatic metering method, which comprises the following steps: calculating a nominal truncation phase distribution diagram of a to-be-measured electronic component by using a virtual projection imaging model, controlling a projection unit to project a single-frequency sine stripe grating to the electronic component and collecting a deformed stripe image, calculating the gradient of the modulation degree distribution map and extracting edge contour features to solve the actual pose of the assembly; performing modulo 2pi differential operation on the nominal and actually measured truncated phase distribution maps, and eliminating high-gradient artifacts by using an edge mask to generate a phase residual map; and converting the phase residual image into a height deviation distribution diagram according to the phase height mapping relation, and by constructing a phase domain differential metering channel, the phase unwrapping ambiguity in discontinuous surface measurement is avoided.

A dynamic vision system includes a variable focus liquid lens optical assembly. The dynamic vision system includes a variable lighting assembly. The dynamic vision system includes a control system configured to adjust one or more optical parameters and data collected from the variable focus liquid lens optical assembly in real time. The dynamic vision system includes a control system configured to adjust the variable lighting assembly. The dynamic vision system includes a processing system that dynamically learns on a training set of outcomes, parameters, and data collected from the variable focus liquid lens optical assembly to train a set of machine learning models to control the variable focus liquid lens optical assembly to optimize collection of data for processing by the set of machine learning models.

The invention discloses a small target detection and state perception method based on multi-scale feature fusion, and belongs to the field of computer vision and deep learning. Multi-scale semantic features are extracted through a backbone network; two uplink fusion paths and two cascaded downlink enhancement paths are constructed, and multi-scale feature fusion is performed, so that the perception capability of targets with different sizes is enhanced, and the accuracy and robustness of detection are improved; and meanwhile, a regional state sensing mechanism is constructed based on a detection result, continuous monitoring and intelligent analysis of target space distribution, behavior trend and dynamic change are realized, and the adaptability and response speed of the system in a complex environment are improved. The method gives consideration to the detection precision and the calculation efficiency, and is suitable for real-time application scenes with limited resources.

The invention discloses a light-weight instrument small target detection model and method for a complex industrial scene, belongs to the crossing field of deep learning and edge calculation, and aims to solve the problem that an existing method cannot meet the real-time detection requirements of edge equipment such as an inspection robot in the aspects of precision, efficiency and small target detection capability. The model comprises a lightweight backbone network used for extracting multi-scale features from an input image; the cross-scale feature fusion network is used for bidirectionally fusing the multi-scale features, retaining shallow space details and deep semantic information and outputting fused features; the deformable large-kernel target sensing module is deployed at a specified position of the cross-scale feature fusion network so as to better capture feature information related to a target area and improve the feature expression capability of a small target; and the multi-task decoupling prediction network performs classification, positioning regression and confidence prediction on the input image in parallel, and a positioning regression branch adopts an EIOU loss function of decoupling width and height optimization.

The invention discloses an ocean multi-mode environment perception and intelligent ship navigation decision-making method based on a double-branch vision-semantic encoder. The method comprises the following steps: S1, acquiring a multi-source data image containing a ship and a surrounding environment thereof from an existing public maritime data set or platform; s2, training a double-branch vision-semantic encoder by using the multi-source data image, and inputting a to-be-processed image extracted in real time into a multi-modal feature matrix in the trained double-branch vision-semantic encoder; s3, based on the multi-modal feature matrix, obtaining positioning information of the ship and surrounding environment elements, and constructing a dynamic security domain model; and S4, in combination with the dynamic security domain model and the multi-ship relative position relationship, carrying out quantitative evaluation on the navigation risk, and generating a self-adaptive navigation strategy based on an evaluation result. According to the invention, high-precision ship positioning and environment element identification under complex weather and illumination conditions are realized by using all-weather characteristics and multi-scale visual feature coding of SAR imaging.

The invention discloses a lightweight-based satellite landslide image intelligent detection method, device and equipment and a medium, and relates to the technical field of disaster detection, and the method comprises the steps: obtaining a whole-scene optical satellite image containing a landslide and a non-landslide region and landform auxiliary data; a dynamic segmentation strategy is adopted to carry out differential segmentation and standardized preprocessing on an image based on topographic data, and a standardized image is obtained. A target landslide image is screened through a double-layer machine learning model, the target image is input into an improved lightweight convolutional neural network for processing, an initial detection result is obtained, finally edge optimization and coordinate calibration are performed on the result, and an accurate landslide area detection result is output. The identification precision of the landslide image is effectively improved through dynamic segmentation and double-layer screening, the improved lightweight convolutional neural network realizes efficient detection in a low-resource environment, and the accuracy of a landslide detection result is further improved.

A computer implemented method for training a whole medical image foundation model, including: receiving a plurality of medical image datasets; extracting local sections of image data from the plurality of medical image datasets; obtaining one or more causal variables associated with the local sections and / or patient; training one or more self-supervised learning models based on the local sections of image data and the causal variables; combining the one or more trained self-supervised learning models with a deep learning network configured to combine a latent representation of the local sections of image data from the one or more trained self-supervised learning models into a patient-level representation; and combining, with the one or more trained self-supervised learning models and the deep learning network, at least one further network or function configured to accept the patient-level representation as input, the at least one further network or function operable to perform one or more patient-specific prediction tasks.

The invention relates to a multi-scale remote sensing image spatial-temporal difference cooperative water quality dynamic monitoring method, and belongs to the technical field of water quality monitoring. The method comprises the steps that a multi-scale time sequence remote sensing image is acquired, a key monitoring domain is delimited through pollution risk and function partition coupling, and time-space registration and spectrum calibration are completed in combination with hydrological parameters; the method comprises the following steps: directionally extracting water quality parameter correlation difference characteristics, quantifying hierarchical characteristic correlation intensity and filtering non-pollution interference signals; a feature-oriented inversion framework is built, a pollution diffusion boundary is delimited, a pollution source is traced, and a water quality parameter space-time dynamic distribution diagram is generated through multi-feature adaptation fusion; verifying the adaptive deviation of the region and the local dimension, dynamically correcting the weight of the model, and constructing a two-factor early warning rule to form a whole-process monitoring link. According to the method, multi-scale image space-time difference characteristics are fully utilized, the accuracy and dynamic response capability of water quality monitoring are improved, and reliable support is provided for pollution source tracing and risk early warning.

The invention discloses an automobile central control screen small target detection method based on YOLOv11 improvement, and the method specifically comprises the steps: S1, generating an image data set, carrying out the preprocessing and enhancement, and dividing the data set; s2, an improved C3k2GCConv module, a WFU module and a CGAFusion module are introduced, and a YOLOv11 network model is constructed; s3, training the model by adopting a cosine annealing learning rate and a mixed precision training strategy; s4, inputting a to-be-detected central control screen image into the improved YOLOv11 detection model, and outputting the category and bounding box coordinates of a target; and S5, performing screening and optimization through a post-processing module, and finally outputting a detection result in the form of a bounding box and a category label. According to the method, by improving the YOLOv11 model, a display target can be effectively recognized in complex environments such as strong light direct incidence, screen reflection and dim light, and the stability and robustness of the model in a complex illumination scene are improved.

The invention discloses a remote sensing-based canopy scale urban green land vegetation classification method. The method comprises the following steps of: obtaining and fusing multi-source remote sensing data; constructing a canopy scale multi-dimensional feature set fused with multi-source remote sensing data; constructing an urban green land vegetation enhancement sample set; constructing a low-dimensional feature set after optimization; the invention further discloses an urban green land automatic classification method based on the canopy scale of the improved random forest. According to the method, a multi-dimensional feature set is constructed through multi-source high-resolution remote sensing data; an automatic dynamic feature optimization and weight distribution mechanism based on mRMR is introduced, redundancy is effectively reduced, and discriminative features are highlighted; constructing a heterogeneity-driven adaptive sample set in combination with multi-source prior knowledge; and based on an improved random forest algorithm which introduces dynamic weighted node splitting, neighborhood constraint and adaptive category balance, high-precision, canopy-scale and adaptive classification of urban green land vegetation is realized.

The invention discloses a glioma T cell prediction and prognosis evaluation method based on pathological images, particularly relates to the field of patient prognosis health evaluation, and aims to solve the problems that existing pathological evaluation is difficult to combine with tumor structure heterogeneity and immune infiltration distribution, and the future progress risk of a patient cannot be predicted based on follow-up visit pathological data. A spatial heterogeneity map of a tumor core area and an invasion edge is constructed in a digital pathological section, density gradients of T cells in different areas are calculated to generate a distribution heterogeneity coefficient, interaction processing is performed on the two types of characteristics in combination with historical follow-up visit records, and a time sequence neural network constructed based on a gating structure is combined to obtain a high-efficiency characteristic of the tumor. And outputting disease progress probabilities of a plurality of follow-up visit time points in the future to form a prognosis trajectory prediction curve, calculating a quantitative recurrence risk score according to curve slope change and an immune fluctuation mode, and finally generating an individualized management scheme, thereby realizing quantitative prediction evaluation and risk management of the prognosis trend of the glioma patient.

The invention discloses a spraying robot trajectory planning method based on depth camera scanning. The method comprises the steps that wall surface boundary polygon data, an initial spraying stroke set and a spraying dosage model parameter set are obtained; calculating a predicted coating thickness field, and performing difference calculation on the predicted coating thickness field and a preset target thickness to generate a thickness error field; performing connected domain clustering on the thickness error field to generate topological thickness error regions, and determining region type labels for the topological thickness error regions one by one; calling a matched editing operator from a discrete stroke editing operator library, performing geometric constraint verification, and generating a candidate editing scheme; and evaluating the candidate editing schemes based on a preset comprehensive scoring function, updating the initial spraying stroke set by using the scheme with the optimal score, obtaining an optimized spraying stroke set, and converting the optimized spraying stroke set into a robot control instruction. According to the method, the problem that global coverage and local thickness uniformity cannot be considered in traditional geometric planning is solved, and high-quality full-automatic spraying is achieved.

The invention relates to the technical field of micro-nano motion control, and discloses a self-adaptive control method of a micro-nano high-precision motion platform. The method comprises the following steps: acquiring real-time pose feedback data and target trajectory data of a motion platform, and extracting dynamic response features; calling the trained motion feature analysis network to perform multi-modal feature separation, and generating a platform pose feature set; based on the set, performing space-time coupling analysis on the working environment parameters through an environment disturbance perception model to obtain a fusion result containing mechanical deformation characteristics and environment disturbance characteristics; inputting a fusion result into a dynamic compensation model to calculate a track correction amount, and outputting a driving compensation instruction; and calibrating the target trajectory data in real time according to the compensation instruction, and generating an actual control signal. According to the method, through multi-modal feature analysis, space-time coupling perception and dynamic compensation, the control precision and stability of the motion platform in a complex environment are improved, and the method is suitable for a micro-nano high-precision motion control scene.

The invention provides a medical image segmentation method based on an AFMHiFormer. The method comprises the steps that firstly, a multiple data enhancement module is provided, and the data distribution diversity is improved while the enhancement stability is guaranteed; secondly, a segmentation model AFHiMFormer is constructed, and the model architecture adopts a double-branch encoder and a multi-scale decoder; thirdly, a feature enhancement module is provided to construct a dynamic complementation mechanism of semantic enhancement and boundary modeling; fourthly, a multi-scale feature fusion module is introduced, multi-scale context information is captured through parallel hole convolution with different expansion rates, and self-adaptive fusion of global and local features is achieved; and fifth, a cross-scale fusion module is designed in the multi-scale decoder, so that the deep layer branch and the shallow layer branch are efficiently fused in a multi-level feature space. According to the method, the advantages of CNN and Transform are combined, dynamic fusion of local and global features is realized by providing a new module, and a remarkable performance advantage is shown in a medical image segmentation task.

The invention discloses a medical image segmentation method based on TransUNet, and belongs to the technical field of medical image segmentation. The method comprises the steps of firstly performing data preprocessing on an original image to obtain preprocessed data; and a DCA attention module is used at a jump joint, so that the problem that a semantic gap exists between characteristics of an encoder and a decoder due to the fact that a simple jump connection scheme is difficult to capture a multi-scale context is solved. The semantic difference leads to redundancy between low-level and high-level features, and finally the segmentation performance is limited. Secondly, a multi-scale boundary sensing module is added to the top layer of the encoder, so that the neural network can better segment the boundary of the target image in the training process; and inputting the preprocessed data into the improved TransUNet model to train the medical image, and outputting an image segmentation result.