368 results about "Model image" patented technology

The invention discloses a modal language model image editing technology fusing non-perpetual culture elements. The modal language model image editing technology comprises the steps that a model is selected and trained, an LLaMA model is selected as a basis, LoRA is introduced for adaptive fine adjustment, and in this way, the model is subjected to adaptive adjustment under the condition that original parameter freezing is kept; according to the method, the understanding and reasoning capabilities in instruction editing are enhanced by combining MLLM (such as LLaVA), the MLLM can perform collaborative learning across text and image modalities and extract deep semantic information, so that the model can not only process basic instructions, but also understand complex non-perpetual culture elements, and the method has great significance in improving the understanding of the model on the non-perpetual culture elements. An enhanced bidirectional interaction mechanism is designed, and the mechanism realizes deep interaction between image and text features through a cross attention mechanism, so that the image features can be used as query and key value pairs to perform bidirectional communication with the text features, and the expression of a model in a complex non-perpetual scene is improved.

The invention discloses a diffusion model image generation method and device for structural disturbance guidance and consistent distillation fusion, and the method comprises the steps: carrying out the initialization of a first student network through a diffusion model, and initializing a first teacher network; generating an initial state of the submerged space according to the submerged space data set and a noise scheduling function of the diffusion model; introducing the classifier-free guidance after the structural disturbance into a first teacher network to obtain a second teacher network; according to the first teacher network, the second teacher network and the initial state of the submerged space, state prediction is carried out through a solver, and enhanced input of the first teacher network is generated; obtaining consistency loss according to the first student network, the first teacher network and the enhanced input; according to the consistency loss, performing parameter updating on the first student network to obtain a target model; the target model is used for generating a target image. The method can improve the quality of the generated image and increase the generation speed of the diffusion model, and can be widely applied to the technical field of image generation.

The invention discloses a foggy day perception-based feature refinement ship instance segmentation method and equipment, and belongs to the field of instance segmentation of ship images in a foggy day scene. According to the method, YOLOv8s is used as a basic model, a joint learning framework is designed, and defogging and feature refinement enhanced perception strategies are combined. Through a state space model and a multi-scale channel aggregation gating strategy, modeling image features are refined, and foggy day long-distance dependence feature information captured by the model is enhanced. And through the adaptive defogging module, fog background interference features are reduced. Through multi-window branch feature extraction and a multi-angle similarity measurement strategy, the ability of the model to extract local detail information and global information is further enhanced, the focusing ability of a ship target area and the adaptability to a complex haze scene are improved, and the ship instance segmentation performance and generalization ability in a foggy day scene are effectively improved.

A virtual try-on method and apparatus, and a storage medium. The method comprises: acquiring a clothing image to be subjected to try-on and a first model image; on the basis of the clothing image to be subjected to try-on, determining a target clothing image; on the basis of the first model image, determining a try-on area image and a first key point, wherein a try-on area comprises an area of model clothing, a body area which is in contact with the model clothing, and a background area which is in contact with the model clothing; and on the basis of the target clothing image, the try-on area image and the first key point, generating a try-on effect image by means of a trained diffusion model.

The invention discloses an overhead transmission line mechanical construction three-map one-table intelligent generation method, which comprises the following steps of: fusing a digital orthoimage model, elevation data and a power transmission and transformation project model through a coordinate transformation matrix, and establishing a three-dimensional scene with a terrain curvature attribute; based on the terrain curvature and equipment coordinates of the three-dimensional scene, a multi-objective optimization algorithm is adopted to synchronously plan a composite path of a cableway, a track and a road, and a road network list map containing an obstacle avoidance topology is generated; recognizing a tower footing projection polygon in the three-dimensional scene, segmenting the types of surface attachments in combination with a digital orthoimage model image, and generating a surface feature list map with a water and soil conservation identifier; and establishing a mapping function of the terrain curvature and the environment water conservation rule, outputting equipment types and spatial poses according to the hoisting safety margin and the earth surface disturbance threshold value, and generating an equipment list map. According to the method, a unified space reference and multi-system dynamic cooperation mechanism is constructed, and the planning precision and construction compliance of three graphs and one table are remarkably improved.

The invention provides a method and a system for processing a cell image based on a DistSegNet model. The method for detecting the cells based on the DistSegNet model image processing technology comprises the following steps: preprocessing a collected cell image; performing cell segmentation on the cell image based on a DistSegNet model image processing technology to generate a cell nucleus region; based on the segmented cell nucleus region, generating a background region, a foreground region and an edge region by adopting morphological operation; calculating the shortest Euclidean distance from each background region pixel to the nearest foreground region pixel, and segmenting the cell image by using a watershed algorithm; performing result optimization on the segmentation map corresponding to each cell instance label to obtain a filtered segmentation label map; and extracting the outer contour of the intra-nucleus region and the outer contour of the surrounding region of each cell from the segmentation mark graph, and storing the outer contours to a corresponding result file. According to the invention, the accuracy and accuracy of a WSI-level cell detection result can be improved.

The invention relates to a lightweight multi-modal image fusion method based on a knowledge distillation technology. Compared with the prior art, the lightweight multi-modal image fusion method based on the knowledge distillation technology overcomes the defects that an image fusion method based on text guidance is large in calculation overhead and large in model scale. The method comprises the following steps: obtaining and preprocessing a source image; constructing a lightweight image fusion model; training a lightweight image fusion model; obtaining a to-be-fused image; and obtaining a multi-model image fusion result. According to the method, through a teacher-student network architecture and a customized prior distillation process, the semantic understanding ability of a large language model is successfully transferred to a lightweight student network, and under the condition that text guidance in a reasoning stage is not needed, relatively high fusion quality is still kept, and meanwhile, the calculation overhead is remarkably reduced.

The invention discloses an anti-corrosion control system and method for a pipeline operation robot, and relates to the technical field of robot control, and the system is composed of a plurality of functional modules, and comprises a multi-sensor fusion module which obtains a pipeline image in real time through a camera; meanwhile, a sensor is used for dynamically collecting internal environment parameters of the pipeline; the three-dimensional modeling and positioning module is used for constructing a pipeline three-dimensional model based on an SLAM algorithm and the pipeline image and positioning the position of the pipeline where the robot is located in real time; the intelligent decision-making module is used for identifying a pipeline corrosion area and position by utilizing YOLOv5 based on the pipeline image and performing corrosion analysis on the corrosion area, and the corrosion analysis comprises corrosion thickness and form contour; measuring the corrosion thickness of the corrosion area by utilizing electromagnetic eddy current detection, and extracting the shape contour of the corrosion area by utilizing deep learning model image recognition; and triggering a self-adaptive control strategy according to the internal environment parameters of the pipeline and the pipeline position of the robot.

The invention discloses a multi-modal remote sensing image fusion classification method under a modal missing condition, and the method comprises the steps: constructing a multi-modal remote sensing image model image fusion classification model, taking the data of three remote sensing image modals as input, carrying out the feature extraction of the data of each modal through a respective specific feature encoder and a shared feature encoder, and carrying out the feature extraction of the data of each modal; the method comprises the following steps: firstly, generating a specific feature, then fusing the specific feature and the shared feature, generating a single-mode fusion feature by adopting a jump link, and finally, fusing the fusion features of the three modes again, and generating a final fusion feature for model prediction. Two auxiliary tasks are introduced into the model, one is a domain classification task based on specific features, and the other is a comparative learning task for shared features. The model is not only suitable for full-modal training, but also effectively extracts and fuses the multi-modal features under the condition that partial modals are missing, and the classification performance of the model is improved.

Disclosed herein are methods, systems, and computer-readable media for generating image captions for training a machine learning model. Current image generation models are hindered by the prevalence of improper or inaccurate captions, which leads to suboptimal training data. This results in less effective image generation models. Disclosed systems and methods involve obtaining a text-to-image dataset including one or more digital image-caption pairs. Systems and methods involve generating a recaptioned dataset by applying an image captioner model to images in the text-to-image dataset. An image captioner model can be trained with an improved image dataset, a first tuning stage, and a second tuning stage, for improved performance.

The invention discloses a method for automatically generating an AI model, which belongs to the technical field of image generation and comprises the following steps: acquiring basic characteristics, posture information and clothing description of the AI model; a model image is generated through a diffusion converter according to the basic features, the clothing description and the posture information, the model image is optimized through an optimizer, and the optimized model image is obtained; wherein the diffusion converter is optimized by combining LoRA; carrying out background fusion and adjustment on the optimized model image to obtain a final model image of the AI model; according to the technical scheme, the high-quality customized AI model image can be efficiently generated.

The invention provides an intelligent bill recognition method based on a deep neural network model, and relates to the technical field of intelligent bill recognition based on the deep neural network model.The intelligent bill recognition method comprises the steps that various bill data are collected, image features and text features are extracted respectively and preprocessed, and complete multi-modal data support is provided for subsequent modeling; secondly, modeling image features by using a convolutional neural network and modeling text features by using a deep neural network, splicing the image features and the text features into a joint feature vector, and deeply fusing information of an image and a text modal through a cross-modal attention mechanism to enhance multi-modal relevance; and finally, designing a dynamic weight adjustment mechanism, adaptively adjusting the optimization weights of the cross entropy loss of the classification task and the mean square error loss of the field analysis task based on the cross entropy loss of the classification task and the mean square error loss of the field analysis task, and balancing optimization conflicts in a multi-task scene.

In various examples, high-precision semantic image editing for machine learning systems and applications are described. For example, a generative adversarial network (GAN) may be used to jointly model images and their semantic segmentations based on a same underlying latent code. Image editing may be achieved by using segmentation mask modifications (e.g., provided by a user, or otherwise) to optimize the latent code to be consistent with the updated segmentation, thus effectively changing the original, e.g., RGB image. To improve efficiency of the system, and to not require optimizations for each edit on each image, editing vectors may be learned in latent space that realize the edits, and that can be directly applied on other images with or without additional optimizations. As a result, a GAN in combination with the optimization approaches described herein may simultaneously allow for high precision editing in real-time with straightforward compositionality of multiple edits.

The invention discloses a self-adaptive image fog removal method guided by a large-scale multi-modal model, belongs to the technical field of visual information processing, and is applicable to the field of foggy day scene image processing. The implementation method of the invention comprises the following steps: 1, obtaining degradation prior by using a pre-trained large-scale multi-modal model; 2, inputting the fog image and the degradation priori into a MoE-SSM model for training, so as to dynamically adjust model parameters and further perform dynamic perception defogging; 3, fusing the shallow feature Fs and the optimized feature # imgabs0 #, and reconstructing a clean and fogless image; 4, training a neural network by using a clean and fogless image and a truth value clear image by using a Charbonier loss function shown in a formula (8), so as to punish the deviation between the recovered image and the truth value clear image and encourage a consistent image gradient; compared with the prior art, the problem of compromise processing of receptive fields and calculation efficiency in existing defogging model image processing is solved, meanwhile, the requirement for a large amount of annotation data is reduced, and then the image processing accuracy and stability are improved.

The invention discloses an image processing method and device, a medium, electronic equipment and a program product. The method comprises the following steps: acquiring an article image comprising a first article and a model image comprising an interaction posture of a model and a second article; superposing the article image to the position of the first limb in the model image to replace the second article in the model image with the first article to obtain a first image; obtaining a first mask picture of the foreground area including the first article and the first limb; and according to the first mask image, performing local redrawing on the first limb in the first image to obtain a second image. Through superposition of the article image and the model image, the first article and the model are fused, and the real consistency of the generated content can be ensured. And according to the first mask image, local redrawing is carried out on the shielding conflict area in the first image, so that the natural interaction feeling of the model and the first article is improved.

The invention relates to the technical field of industrial detection, and provides a real-time monitoring method and device for slag in a boiler. A plurality of high-temperature-resistant cameras are installed around an observation opening in the bottom of a boiler in an annular array mode, the slag accumulation condition in the boiler is collected in real time through the high-temperature-resistant cameras, and a single-view-angle image is obtained; evaluating the visual field states of the high-temperature-resistant cameras according to the single-visual-angle images, and determining a plurality of to-be-modeled images from the plurality of high-temperature-resistant cameras; and generating slag point cloud data by using the at least one to-be-modeled image, and performing three-dimensional reconstruction on the slag point cloud data to obtain the slag pile volume. According to the method, the slag falling condition of the boiler bottom of the power plant boiler is monitored and recognized in real time, and the problems that the slag falling at the boiler bottom cannot be detected in real time depending on manpower, the detection accuracy is insufficient, the three-dimensional form of the slag falling at the boiler bottom cannot be synthesized by detecting the slag falling at the boiler bottom through sensor equipment, and practicability is poor are solved.

The invention relates to the technical field of image processing and machine learning, provides a method for improving the image generation quality of a multitask visual basis model based on a diffusion model, and provides a multitask visual basis model image generation method based on the diffusion model. Comprising a visual low-level semantic information learning adapter expanded according to the number of tasks and a multi-task visual basic model with a feedback auxiliary learning mechanism, and the visual low-level semantic information learning adapter expanded according to the number of tasks comprises a group of parallel small-weight convolutional layer modules specific to different tasks. A convolutional layer module shared among different tasks is connected in series behind the convolutional layer modules of the specific tasks; and the feedback auxiliary learning mechanism guides the multi-task visual basic model to better train through the minimum feedback loss so as to promote the improvement of the final image generation quality. The method is mainly applied to image processing occasions.

The invention relates to the technical field of computer vision and artificial intelligence, in particular to a visual state space model-based neural network optimization method for bridge diseases, which comprises the following steps of: constructing a neural network comprising a feature extraction backbone network and a task adaptation head network, modeling an image feature sequence by the backbone network based on a visual state space model, the task adaptive header network is used for capturing local details and long-distance global dependence and outputting disease categories, confidence coefficients and positions; a standardized data set is used for training a model, and dynamic optimization is completed by evaluating the importance of neurons and sorting and cutting redundant neurons; and preprocessing a to-be-detected image, inputting the preprocessed to-be-detected image into the optimization model, and outputting a precise disease identification result through feature extraction, multi-branch processing and non-maximum suppression. The method improves the disease feature extraction precision, achieves the light weight of the model, adapts to multi-scale disease detection, improves the practicability of an identification result, and is suitable for a bridge disease intelligent detection scene.

The invention provides a method and a system for processing a cell image based on a DistSeg model. The method for detecting the cells based on the DistSeg model image processing technology comprises the following steps: preprocessing a collected cell image; segmenting the cell image based on a DistSeg model image processing technology to generate a cell nucleus region; based on the segmented cell nucleus region, generating a background region, a foreground region and an edge region by adopting morphological operation; calculating the shortest Euclidean distance from each background region pixel to the nearest foreground region pixel, and segmenting the cell image by using a watershed algorithm; performing result optimization on the segmentation map corresponding to each cell instance label to obtain a filtered segmentation label map; and extracting the outer contour of the intra-nucleus region and the outer contour of the surrounding region of each cell from the segmentation mark graph, and storing the outer contours to a corresponding result file. According to the invention, the accuracy and accuracy of a WSI-level cell detection result can be improved.

Systems and methods for performing surface matching using three-dimensional (3D) and color information. An example method includes obtaining, by a 3D camera, a first 3D image of a field of view. The 3D image includes (i) a plurality of voxels, or 3D points, of the field of view and (ii) color information of the field of view. A processor normalizes the color information into a common color space, and determines scene histograms from the normalized color information. Each of the scene histograms is determined for a voxel of the plurality of voxels. The method further includes determining a color score from at least the scene histograms and model histograms, the model histograms being indicative of color information of a model image of a 3D object, and determining the presence of a surface of an object in the first 3D image from the color score.

To realize a service for easily generating an apparel image.SOLUTION: According to an embodiment of the present invention, an apparel image generating system has: a display unit for displaying a selected screen for which a user can select at least one of a model image, an item image, a background image, a face image and a pose image; and an image generating unit for using an image generation model to generate an apparel image from the image selected by the user.SELECTED DRAWING: Figure 9

The present teachings relate to a method comprising: receiving at a user container runtime environment, via a container registry, an image of an obfuscated containerized model of an asset; wherein the containerized model comprises an API, and wherein the containerized model has been generated in a supplier runtime environment located within a supplier network environment, interfacing, with a model execution logic, the API of the containerized model image, wherein the interface comprises a process data signal that is transmitted to the API of the model image, and a model result data signal that is transmitted from the API of the model image; wherein the process data signal comprises data related to the usage conditions of the asset, and the model result signal is indicative of the response of the asset to at least some of the usage conditions, wherein the user container runtime environment is located within a user network environment which is isolated from the supplier network environment. The present teachings also relate to a method for transmitting a model of an asset, a method for transmitting an augmented model, a computer software product and a computer network arrangement.

The invention relates to the technical field of data processing, in particular to a BIM live-action model-based bridge construction progress monitoring method, which comprises the following steps of: extracting feature points of a bridge region in a construction bridge grayscale image and a bridge BIM model image, and matching to obtain a plurality of groups of feature matching points; obtaining the matching accuracy of each group of feature matching points according to the position difference of the bridge local area to which each group of feature matching points belong and the gray difference between the surrounding pixel points; screening all groups of feature matching points according to the matching accuracy to obtain a plurality of groups of successfully matched feature points; according to the number of the successfully matched feature points and the difference between the actual consumed material amount of the bridge and the planned consumed material amount of the bridge, the bridge construction progress degree is obtained; and performing construction monitoring on the bridge based on the bridge construction progress degree. The accuracy of the monitoring result of the bridge construction progress is improved.

Disclosed in the embodiments of the present application are a model rendering method and a related apparatus. The method comprises: for an initial three-dimensional model to be processed, simplifying the model structure of the initial three-dimensional model on the basis of a topological structure feature of the initial three-dimensional model, so as to obtain a simplified three-dimensional model, wherein the number of meshes comprised in the simplified three-dimensional model is less than the number of meshes comprised in the initial three-dimensional model, and the difference between the simplified three-dimensional model and the initial three-dimensional model meets a preset similarity requirement; by means of lighting simulation processing, determining a lighting intensity corresponding to a vertex in the simplified three-dimensional model when the simplified three-dimensional model is in a virtual environment; and on the basis of the simplified three-dimensional model and the lighting intensity corresponding to the vertex in the simplified three-dimensional model, rendering the simplified three-dimensional model, so as to obtain a model image for representing the simplified three-dimensional model. The method can automatically reconstruct and render a three-dimensional model, thereby improving the reconstruction efficiency of the three-dimensional model and reducing labor costs.

Disclosed are example embodiments of systems and methods for virtual try-on of articles of clothing. An example method of virtual try-on of articles of clothing includes selecting a garment from a pre-existing database. The method also includes loading a photo of a source model wearing the selected garment. Additionally, the method includes generating a semantic segmentation of the model image. The method also includes extracting the selected garment from the photo of the model. Additionally, the method includes determining a correspondence between a target model and the source model by performing a feature point detection and description of the target model and the source model, and performing feature matching and correspondence validation. The method also includes performing garment warping and alignment of the extracted garment. Additionally, the method includes overlaying and rendering the garment.

The invention discloses a diffusion model image completion method based on boundary preserving perception, and belongs to the technical field of artificial intelligence and image generation. The method comprises the following steps: firstly, preprocessing an input image and a corresponding defect mask, and extracting gradient attenuation boundary features and a structure edge graph; a boundary sensing convolution module is introduced into the completion model and is used for boundary information modeling of a coding end; meanwhile, a gating mechanism is adopted in U-Net jump connection, the weight of a gradient attenuation mask is combined, and the boundary information transmission intensity is dynamically adjusted; furthermore, in the diffusion generation process, a region self-adaptive scheduling strategy is adopted, only high-step diffusion iteration is carried out on a defect region and a boundary neighborhood, and low-step or freezing operation is carried out on a non-defect region, so that balance between boundary consistency and calculation efficiency is realized. Experimental results show that the method is superior to an existing diffusion model image completion method in the aspects of structure retentivity, edge continuity and visual quality, and has wide application prospects.

The invention discloses a modal language model image editing technology fusing non-perpetual culture elements. The modal language model image editing technology comprises the steps that a model is selected and trained, an LLaMA model is selected as a basis, LoRA is introduced for adaptive fine adjustment, and in this way, the model is subjected to adaptive adjustment under the condition that original parameter freezing is kept; according to the method, the understanding and reasoning capabilities in instruction editing are enhanced by combining MLLM (such as LLaVA), the MLLM can perform collaborative learning across text and image modalities and extract deep semantic information, so that the model can not only process basic instructions, but also understand complex non-perpetual culture elements, and the method has great significance in improving the understanding of the model on the non-perpetual culture elements. An enhanced bidirectional interaction mechanism is designed, and the mechanism realizes deep interaction between image and text features through a cross attention mechanism, so that the image features can be used as query and key value pairs to perform bidirectional communication with the text features, and the expression of a model in a complex non-perpetual scene is improved.

Systems and techniques are described herein for segmenting images. For instance, a method for segmenting images is provided. The method may include encoding, using a machine-learning-model image encoder, an image to generate a plurality of image features; selecting a first image feature from among the plurality of image features; determining a first image point related to the first image feature; encoding, using a machine-learning-model prompt encoder, the first image point as a first encoded prompt; generating, using a machine-learning-model image decoder, a first mask based on the plurality of image features and the first encoded prompt, wherein the first mask is indicative of pixels of the image that are semantically similar to the first image point; and storing the first mask.

The invention discloses a model scheduling method, a storage cluster, a device and a medium, and relates to the technical field of computers. The method comprises the following steps: performing hierarchical processing on a model mirror image file in advance to obtain a base layer mirror image file and a model layer mirror image file, storing the base layer mirror image file in a cache space corresponding to a cloud platform, and storing the model layer mirror image file in a network sharing space. The mixed storage of the cache space and the network sharing space of the mirror image file is realized, and the redundant transmission and storage overhead is also reduced. In response to the current scheduling request, after the scheduling of the base layer mirror image file is completed for the first time, the base layer mirror image file does not need to be mounted for the non-first scheduling, so that the multiplexing of the base layer mirror image file is realized. And only the incremental model layer mirror image file is mounted for model iteration, so that the transmission overhead is reduced, the iteration requirement of each model is met, the network bandwidth consumption is reduced, and the deployment efficiency is improved.

The invention relates to the technical field of three-dimensional model image data processing, and discloses a three-dimensional model particularization splitting deduction and multi-mode feedback method, which comprises the following steps: carrying out lightweight three-dimensional model reconstruction according to an equipment process structure diagram, and generating a glTF format data packet with a process information embedding characteristic; establishing a virtual three-dimensional scene through a three.js engine, and instantiating a three-dimensional model object in a glTF format; non-uniform granulation processing is carried out on the equipment model by adopting a splitting and deducting mechanism, discrete motion tracks of parts are solved through a response algorithm, and an equipment dynamic decomposition path display effect is constructed; a calibration sensing link framework is established, an interaction response mechanism based on user behaviors is constructed, and a multi-mode visual feedback system of a large screen is achieved; according to the method, the technical bottlenecks of slow model loading, high interaction delay and the like in traditional industrial large screen display are effectively solved, and the dynamic interaction display effect and user perception of an industrial equipment structure are improved.