272 results about "Diffusion network" patented technology

The invention relates to the technical field of new energy power generation prediction, in particular to a new energy power prediction method fusing typhoon meteorological information and a micrometeorological prediction result, which comprises the following steps: collecting multi-source observations such as a satellite scatterometer, a radar wind profile and laser wind measurement, and unifying coordinates; constructing a mesoscale wind field by adopting four-dimensional variational assimilation, inferring a micrometeorological field by utilizing a spectrum embedding diffusion network, and generating a multi-scale meteorological field by frequency domain phase consistency fusion; probabilistic wind speed is sampled in the countercurrent model meeting the condition of mass and momentum conservation, an energy conservation graph neural network is input, the wake effect is coupled, and a unit power quantile value is obtained; the power probability is sent to a risk weighting model, the weight is adjusted in real time according to the peak load, the reserve capacity and the climbing rate risk, a dispatching power curve and uncertainty are output, and grid-connected power errors are used for periodically updating the weight and the parameters of the last layer of the reversible model. According to the invention, the safety acceptance margin of the power grid to new energy is obviously improved.

An image-guided diffusion network has two Convolution Neural Networks (CNNs). A RGB image and an IR image are concatenated with a Gaussian noise image and input to a denoising neural network that merges information from the RGB and IR images as noise is removed over many iterations. Then an enhancement neural network up-samples for Super Resolution (SR) and convolutes to generate a condition vector that controls Global Feature Modulation (GFM) at three convolution layers to generate a SRGFM enhanced fusion image. Timesteps are embedded using adaptive group normalization blocks within Adaptive Bottleneck Residual (ABR) blocks in the denoising network, which is a UNet having many levels of ABRs, and in the enhancement network before feature modulation. Global image features are detected by triple convoluting the image input to the enhancement network to generate the condition vector that controls feature modulation blocks at three layers of convolution.

The invention discloses an industrial part defect sample accurate generation method based on a conditional diffusion model, and belongs to the field of image processing and artificial intelligence. The method forms a closed-loop cooperative system by constructing four deep coupling modules of physical constraint noise scheduling, multi-scale feature coupling, double-domain feedback optimization and adaptive weight adjustment; a defect physical forming mechanism is converted into a dynamic noise scheduling strategy, deep interaction between condition information and a feature map is established at multiple levels of a diffusion network, quality closed-loop optimization is achieved through dual evaluation of a pixel domain and a frequency domain, and training weight is dynamically adjusted according to defect scarcity. And multi-scale accurate control is realized, a quality guarantee closed loop is established, the problem of data imbalance is effectively solved, and the performance of an industrial defect detection model is remarkably improved.

The invention provides a brain tumor image accelerated diffusion network segmentation method based on fuzzy learning, and belongs to the technical field of medical image intelligent segmentation. The technical problem of low segmentation accuracy caused by fuzzy texture and boundary disappearance of the brain tumor image is solved. According to the technical scheme, the method comprises the following steps of S1, performing data preprocessing on a brain tumor image data set; s2, performing pre-segmentation on the brain tumor image by using a 3D U-Net network, and performing iterative noise addition on the obtained segmented image to enable the segmented image to become a pure noise image; s3, performing iterative denoising on the pure noise image, and learning a denoising process by using a fuzzy U-Net network; and S4, stopping the denoising process in advance, and segmenting the segmented image by using a 3D U-Net network to obtain a final segmentation result. The method has the advantages that the segmentation accuracy is high, decision support is provided for segmentation of the brain tumor image, and convenience is provided for diagnosis of doctors.

The invention discloses an overlapped cervical cytoplasm region segmentation method based on deep learning and a conditional diffusion model, and relates to the technical field of artificial intelligence analysis of medical images. According to the method, accurate segmentation of the overlapped cytoplasm region in the cervical cell image is realized through a morphological prior guided conditional diffusion process. The method comprises the following steps: constructing a multi-scale cervical cytoplasm mask pair image; designing a cytoplasm specific data enhancement and preprocessing process; building a multi-branch cervical cell morphology perception condition diffusion network; using a self-adaptive multi-scale combination loss function to optimize training; and a hierarchical classifier is adopted to freely guide sampling for reasoning. According to the method, the frequency domain and space domain features are fused, a cellular morphology and statistics priori knowledge base is established, and a strategy of generating complete cytoplasm by adopting non-overlapped parts is adopted, so that the problem that the traditional method is difficult to segment in complex backgrounds and overlapped regions is successfully solved, and reliable technical support is provided for early screening of cervical cancer.

The present invention provides a video generation method, comprising: encoding an image comprising a reference figure to obtain a feature representation of the reference figure; encoding a video comprising a reference action to obtain a feature representation of the reference action; injecting the feature representation of the reference figure and the feature representation of the reference action into a latent diffusion network in a video latent diffusion model by means of at least one timing hold module in a trained video latent diffusion model, wherein the latent diffusion network comprises a plurality of cross-attention and denoising units, one timing hold module is connected in series between at least two cross-attention and denoising units, and each timing hold module comprises a time converter and a space converter; using the latent diffusion network to denoise an inputted noise sequence to obtain a denoised latent vector sequence; and decoding the latent vector sequence to obtain an action video of a target object. The present invention further provides a video generation apparatus, an electronic device, a storage medium, and a program product.

The invention discloses an ultrasonic mammary gland image segmentation method based on fine-grained guidance, relates to the technical field of image segmentation, and attempts to use a fine-grained guidance attention decoding diffusion model. The defects of weak association of foreground and background semantics, unbalance of global and local features, poor fine-grained feature capture capability and the like in a traditional segmentation method are overcome. The core design is as follows: a clear de-noised image is obtained by using a de-noising diffusion probability model; based on a context cross-attention decoding diffusion network guided by fine granularity and a contained self-adaptive detail guide attention module, image prior information is learned and fused, and semantic correlation between a foreground and a background is enhanced; a context decoding cross attention layer is added, global information of input features and complex correlation between channels are effectively captured, and the image segmentation precision and efficiency are integrally improved. A contrast experiment result on a related data set shows that compared with a similar popular method, the method disclosed by the invention obtains a better image segmentation effect.

The invention discloses a power distribution terminal equipment fault intelligent diagnosis method and system based on the Internet of Things. The method comprises the following steps: S1, collecting multiple items of operation data of a power distribution terminal to construct a time sequence sample; s2, extracting power disturbance characteristics based on a sliding window, generating a behavior coupling matrix and a topological connection matrix, and calculating node redundancy; s3, fusing the two types of relationships to construct a dynamic graph; s4, inputting a graph diffusion network, fusing a structure and a behavior diffusion result, and generating a node state feature vector; s5, inputting a multi-label classification model to identify the fault type and probability; s6, generating a response instruction in combination with the fault type and the redundancy; s7, response is executed, the relation matrix and the classification model are updated, and the diagnosis process is optimized in a closed-loop mode. According to the invention, accurate fault identification and quick response of the power distribution terminal are realized, and the intelligent operation and maintenance level of a power supply system is improved.

The invention discloses a polarization image super-resolution method based on a conditional diffusion model, and the method comprises the following steps: 1, building an active imaging system, and shooting a clear high-resolution polarization image of a target object; step 2, degrading the high-resolution image of the target object into a low-resolution image by using a bicubic down-sampling interpolation method; according to the idea of generating an image by a diffusion model, a conditional diffusion network model is constructed by combining a low-frequency information extraction module, a high-frequency information extraction module, a residual error convolution module and a self-attention mechanism module, a data set is established, and polarization information of the image is fully utilized; the network is trained so that the network can output a high-resolution image according to an input low-resolution polarization image and a noise image obeying standard normal distribution, a good enhancement effect can be achieved by using the conditional diffusion model to carry out super-division on the low-resolution image, and stable performance can be kept in various environments.

The invention belongs to the technical field of infrared target detection, and particularly relates to an infrared small target detection method based on a diffusion model, which comprises the following steps: S1, constructing a conditional diffusion network model of a double-branch architecture, the model integrates a conditional network branch for fully extracting prior features from an input infrared image # imgabs0 # and a denoising network branch for recovering a clean mask # imgabs2 # from a noise mask # imgabs1 #; s2, constructing a two-dimensional Transform module, and remarkably enhancing the recognition capability of the model on the global structure and detail features of a target under a complex background through conjoint analysis of space and channel dimensions; and S3, designing a cross attention fusion module, and realizing accurate guidance of the condition prior branch information on the de-noising branch by calculating the cross attention between the condition prior branch and the de-noising branch. The technical problem of improving the accuracy and reliability of infrared small target recognition is solved. According to the method, an infrared small target detection task is converted into a generation normal form solution, and the accuracy and reliability of small target recognition are remarkably improved.

The invention discloses a metering processing method for an electric energy meter topology application and an electric energy meter of a transformer area topology. The metering processing method comprises the following steps: S1, constructing a measurement data set based on an equipment identifier, a power factor and a sampling timestamp of the electric energy meter; s2, generating an initial graph model containing user nodes and electric energy transmission edges based on the data set; s3, inputting a dynamic coupling backflow enhancement graph diffusion network to execute topology identification; s4, constructing a fusion path diagram and initializing an edge weight; s5, calculating an edge weight regulation factor, and updating the edge weight; s6, asynchronous propagation is carried out to output a transformer area power supply topological graph; s7, constructing a metering path based on the topological graph and identifying an abnormal path segment; and S8, correcting the abnormal path, and generating an optimization topology and evaluation report. According to the method, the precise identification of the power supply topology of the transformer area and the self-adaptive correction of the abnormal metering path are realized.

The invention discloses a polarization restoration imaging method based on a conditional diffusion model, and the method comprises the following steps: 1, shooting a clear polarization image of an underwater target object and polarization images corresponding to different turbid degrees, and 2, dividing the shot polarization images into a training set, a verification set and a test set according to the proportion of 7: 2: 1; a conditional diffusion network model is constructed by combining a polarization information preprocessing module, a residual error convolution module and a self-attention mechanism module according to the idea of generating an image by a diffusion model, a data set is established, and the polarization information of the image is fully utilized; the network is trained so that the network can output a clear image according to an input turbid polarization image and a noise image obeying standard normal distribution, a good enhancement effect can be achieved by using the conditional diffusion model to restore the turbid polarization image, and stable performance can be kept in various environments.

The invention provides a video deblurring method based on a memory diffusion network. The core of the method comprises a deblurring branch and a diffusion branch. The deblurring branch has a bidirectional circulation structure, maintains a simple and effective memory bank containing a specific updating mechanism, and can effectively utilize the space-time relationship between video frames to further ensure the time consistency between the frames in the final deblurred video. The diffusion branch comprises a lightweight diffusion network to generate multi-scale diffusion prior features. According to the method, the effect of an existing regression model acting on video deblurring is improved by combining the diffusion prior generated by the diffusion network. A simplified and effective memory bank is constructed to screen and store time frame information to accurately capture a complex space-time local relationship in a fuzzy video so as to realize deblurring with finer granularity, and a specific memory updating algorithm is designed to refresh the memory bank. The space-time information between continuous video frames is better utilized, and the time consistency between the frames is ensured.

The invention discloses a remote sensing image super-resolution reconstruction method and device, a storage medium and equipment, relates to the technical field of computer vision, and can solve the technical problem that the remote sensing image super-resolution reconstruction effect is poor. The method comprises the following steps: acquiring a sample remote sensing image and multi-modal auxiliary data, extracting spectral features, textural features and geometric features of the pre-processed multi-modal auxiliary data, and fusing to generate a condition feature tensor; performing forward diffusion on the sample remote sensing image to generate a noise-containing image sequence; in the process of training the UNet diffusion network by using the noisy image sequence, injecting the conditional feature tensor into preset levels of an encoder and a decoder of the UNet diffusion network, and optimizing network parameters through multi-objective loss to obtain a trained UNet diffusion network; and obtaining a to-be-reconstructed noise map, performing noise prediction on the noise map based on the trained UNet diffusion network to obtain target noise, performing reverse sampling on the noise map, and generating a target reconstructed image through iterative denoising.

The invention discloses a multi-category material demand prediction system under an emergent public health event, and the system at least comprises a time sequence demand evolution coding module, a double-correlation learning module, and a regional joint prediction module. The time sequence demand evolution coding module is used for capturing a multi-time-scale demand mode from historical demand records and space-time emergent public health event characteristics; the double-association learning module is used for respectively learning experience association and adaptive association between categories; and the regional joint prediction module obtains fine-grained future demands by aggregating the demand mode output by the time sequence demand evolution coding module and the association mode output by the double-association learning module based on the graph diffusion network. According to the method, the regional characteristics and the dynamic characteristics of health event evolution are comprehensively considered, so that the daily demand quantity of each regional material category is accurately predicted.

Approaches presented herein provide for the generation of synthesized data from input noise using a denoising diffusion network. A higher order differential equation solver can be used for the denoising process, with one or more higher-order terms being distilled into one or more separate efficient neural networks. A separate, efficient neural network can be called together with a primary denoising model at inference time without significant loss in sampling efficiency. The separate neural network can provide information about the curvature (or other higher-order term) of the differential equation, representing a denoising trajectory, that can be used by the primary diffusion network to denoise the image using fewer denoising iterations.

The invention discloses a face image generation method and device, a storage medium and electronic equipment. The method comprises the following steps: determining k-th scene attribute description information from scene attribute description tags corresponding to a target scene; extracting a kth scene text feature from the kth scene attribute description information; performing denoising processing on the initialized noise image by using the kth scene text feature in the constraint diffusion network to generate a kth denoised face image; under the condition that k is not equal to P, de-noising processing is performed on the initialized noise image by using the (k + 1) th scene text feature in the constraint diffusion network, and a (k + 1) th de-noised face image is generated; under the condition that k is equal to P, P denoised face images are added to the face image set, and P is the maximum combination number of the scene attribute description labels. According to the method and the device, the technical problem that intra-class consistency of a generation result is difficult to ensure in a face image generation mode provided by a related technology is solved.

The invention provides an unbiased missing modal learning method based on a multi-stage double diffusion network, and belongs to the field of computer vision, natural language processing and multi-modal information fusion. The method comprises the following steps: a multi-modal feature extraction module maps text modal data, image modal data and audio modal data into a unified potential representation space; the potential space representation of the three modals is used as the original feature of the missing modality and the original feature of the available modality; based on the missing modal original features and the available modal original features, training a multi-stage double-diffusion module through forward diffusion and reverse diffusion to obtain a trained multi-stage double-diffusion module, and based on the trained multi-stage double-diffusion module, performing global structure generation and modal conversion through available modal data to obtain a multi-stage double-diffusion model; and performing local detail optimization reasoning to generate missing modal data. According to the method, the problem of modal generation deviation in the existing multi-modal learning is solved, and the effectiveness of the multi-modal learning effect in the missing scene is enhanced.

The invention provides a method for predicting a cabin material pile in a ship loading process of a ship loader based on a diffusion network, and relates to the technical field of bulk cargo loading and unloading automation. Three-dimensional point cloud data in a cabin in the ship loading process is acquired, the aligned three-dimensional point cloud data is projected to a two-dimensional plane parallel to the bottom of the cabin, and a material pile depth map is generated; constructing a training sample pair based on the historical ship loading operation data, and training a conditional diffusion network model to obtain a target conditional diffusion network model; inputting the material pile depth map of the first time period and planned blanking point information corresponding to the second time period into a target condition diffusion network model to obtain a material pile depth map corresponding to the second time period; and taking the material pile depth map corresponding to the second time period as an initial depth map for prediction in the next time period to carry out iterative prediction, and outputting a material pile depth prediction set after the material pile depth map prediction in all time periods in the ship loading process is completed. According to the invention, the fidelity and authenticity of the predicted material form can be improved.

The invention relates to the field of cross-modal information retrieval, in particular to a text video cross-modal retrieval method and system based on two-stage neighborhood diffusion, and the method comprises the steps: carrying out the feature extraction of a to-be-retrieved text and a to-be-retrieved video, and obtaining the global feature vectors of the to-be-retrieved text and the to-be-retrieved video; calculating random neighborhoods corresponding to the text and the video based on the cosine similarity between the to-be-retrieved text and the to-be-retrieved video; performing random sampling on the global feature vector of the text / video to be retrieved in the corresponding random neighborhood to obtain random neighborhood text embedding and random neighborhood video embedding so as to perform embedding space modal alignment of the text and the video; tensor splicing is carried out on random neighborhood embedding of the to-be-retrieved text and the to-be-retrieved video, and cross-modal data distribution of the text and the video is optimized based on a pre-constructed dynamic agency attention diffusion network. Through random neighborhood modeling and a dynamic agency attention diffusion network, the cross-modal alignment precision and the model generalization ability are improved.

The invention discloses a single-image three-dimensional head reconstruction method and system based on a guide diffusion model. The method comprises the following steps: obtaining a head portrait from a single image to be reconstructed; reconstructing a three-dimensional head model according to the head portrait; performing weak projection mapping on the three-dimensional head model under the predicted camera matrix to the head portrait, and obtaining a three-dimensional head model with front textures by utilizing pixel information of the head portrait; performing cylindrical surface UV expansion on the three-dimensional head model with the front texture to obtain a two-dimensional UV texture map and recording a UV mapping relation; inputting the two-dimensional UV texture map into the guided diffusion network model to predict the texture of the invisible area, and repairing to obtain a complete two-dimensional UV texture map; and then the complete two-dimensional UV texture image is reversely pasted to the three-dimensional head model. According to the method, the three-dimensional head model which is high in fidelity, free of artifacts and rich and vivid in texture details can be reconstructed from individual single portrait images of different ages, genders, races and facial expressions.

Systems and methods for generating 3D scenes include a masked red, green, blue, depth (RGBD) input, which is separated into a masked RGB input and a masked depth input. The masked depth input is compressed. The masked RGB input is compressed. A high definition (HD) map control signal is generated for a depth stream, and an HD map control signal is generated for an RGB stream. A depth output is generated based on inputs from the depth stream, the HD map control signal for the depth stream, text encoder, and random sampled noise. An RGB output is generated based on inputs from the RGB stream, the HD map control signal for an RGB stream, text encoder, and random sampled noise to train a dual stream diffusion network.

The invention provides a video generation method. The video generation method comprises the following steps: encoding an image containing a reference image to obtain a feature representation of the reference image; coding the video containing the reference action to obtain the feature representation of the reference action; injecting the feature representation of the reference image and the feature representation of the reference action into an implicit diffusion network in the video implicit diffusion model through at least one time sequence maintaining module in the trained video implicit diffusion model; wherein the implicit diffusion network comprises a plurality of cross attention and denoising units; a time sequence maintaining module is connected in series between the at least two cross attention and denoising units; the time sequence holding module comprises a time converter and a space converter; performing denoising processing on an input noise sequence by using the implicit diffusion network to obtain a denoised implicit vector sequence; and decoding the implicit vector sequence to obtain an action video of the target object. The invention further provides a video generation device, electronic equipment, a storage medium and a program product.

The embodiment of the invention provides an image restoration method and device based on a diffusion model and generative adversarial training, and the method comprises the steps: obtaining image data, inputting the image data into a degeneration encoder, and obtaining a first result; providing an interaction page, and obtaining image limitation information based on the interaction page; the first result and the image limiting information are input into an image restoration network to obtain a second result, and the image restoration network is generated based on a diffusion model in a pre-trained diffusion network; inputting the second result into an image decoder for processing to obtain a restoration result, and feeding back the restoration result; according to the scheme, the pre-trained encoder and the diffusion model can be finely adjusted so as to be suitable for restoration of the degraded image, and a more accurate restored image can be obtained.

The invention discloses a model training method and device, an image reconstruction method and device, electronic equipment and a medium, and relates to the technical field of neural networks. The method comprises the following steps: inputting an initial training image into an initial image reconstruction model; performing image segmentation processing through a segmentation network module which completes pre-training to obtain a segmentation mask probability training image, and inputting the segmentation mask probability training image and the initial training image into a noise adding unit of a diffusion network module to obtain a noise-added training image; inputting the noise-added training image into a denoising unit of a diffusion network module to obtain a predicted training noise and a reconstructed training image; and based on the reconstructed training image, the initial training image, the predicted training noise, the segmented mask probability training image and the real label image, calculating to obtain total loss, and based on the total loss, carrying out iterative updating on the initial image reconstruction model to obtain a trained target image reconstruction model. The method can improve the reconstruction quality of an image reconstruction model for an image with an occlusion area.

The invention relates to a multi-modal medical image intracranial aneurysm segmentation method based on geometric deep learning, and the method comprises the steps: 1, designing a medical diffusion network based on a deep neural network of frequency domain convolution; 2, preprocessing the three-dimensional grid data to realize the manifold of the model and calculate the characteristics of the grid model; and step 3, using a multiple loss function to reduce the influence of the unbalanced data set, and the like. According to the method, the tedious process of manual feature design and multi-stage processing in a traditional segmentation method is avoided; the MedDiffusion Net designed by the method solves the problem of intracranial aneurysm segmentation of medical images of various modalities, and experimental results on MRA, CTA and DSA data show that the method can obtain accurate intracranial aneurysm segmentation without training data of different modalities respectively; and adaptively fusing the features in the neighborhood by learning a diffusion parameter t.

The invention discloses a background coherent story picture book generation method based on a diffusion model, and belongs to the field of computer vision and generative artificial intelligence. The method comprises a training stage and a testing stage: in the training stage, bidirectional cross attention fusion and modal soft selection are carried out on text, background and role multi-modal conditions through a feature enhancement fusion module, model optimization is carried out by utilizing joint alignment loss, and efficient parameter fine tuning is carried out on a diffusion model by adopting an efficient parameter fine tuning method; in the test stage, a reference image input by a user and a text sequence are processed into a fusion condition, a large language model is driven to generate an image mark, the image mark is converted into a diffusion condition through a mapper, and each frame of image is generated step by step by combining an autoregression mode with a multi-condition injection diffusion network. According to the method, the problem of inconsistency of cross-frame backgrounds, styles and roles is effectively solved, and high-quality and coherent generation of the long-sequence story picture book is realized.

A computer-implemented system for foreign object detection in a scene. The system includes a memory-suppress diffusion network module adapted to reconstruct a reconstructed image from an encoded image, and a contrastive dissimilarity network adapted to combine the input image and the reconstructed image to predict an anomaly map for the input image. The encoded image is based on an input image, and the memory-suppress diffusion network module and the contrastive dissimilarity network are trained using only normal, real images. The system leverages only normal images in training and does not compromise the detection performance at the inference stage.

The invention relates to an innovative garment multi-view-angle generation method and device. Multi-view-angle images with high quality and consistency are generated through a self-adaptive local condition multi-view diffusion model. Screening and acquiring a large number of model clothing image data sets from the clothing database, and combining the data sets into clothing image pairs after screening; global and local features are dynamically selected by using a self-adaptive condition feature selection module so as to adapt to requirements of different visual angles; a united attention-based diffusion UNet network is constructed, and a cross-view interaction self-attention module and a cascade cross attention module are adopted to enhance the consistency and quality of multi-view images; the model is optimized and trained through a joint loss function, and the high quality of the generated image at the pixel and visual perception level is ensured. According to the method, the limitation of the traditional technology is broken through, the accuracy and consistency of garment view generation are remarkably improved, the innovative application in the field of garment design is promoted, and the satisfaction degree of consumers on garment display is improved.

The invention relates to the technical field of video super-resolution reconstruction, in particular to an unmanned aerial vehicle video super-resolution reconstruction method and system based on an implicit diffusion model, and the method comprises the steps: encoding a low-resolution frame into a low-dimensional latent feature through an implicit neural expression auto-encoder; denoising the latent features of the reference frame through an implicit neural U-shaped diffusion network to obtain clean latent features, generating a high-resolution reference image through an implicit neural decoder, and performing optical flow alignment to obtain a time consistency control signal; and finally, inputting the current frame latent features and the control signal into an implicit neural U-shaped diffusion network with a historical sampling correction module, and outputting a super-resolution image of any scale by an implicit neural decoder after iterative denoising. According to the method, in the unmanned aerial vehicle video super-resolution reconstruction task, any-scale reconstruction is realized, the problems of space-time inconsistency and sampling error accumulation are effectively relieved, and the method has better performance in the aspects of perception quality, time sequence consistency and the like.