204 results about "Image properties" patented technology

The invention provides a method for editing facial image attributes based on a text of a diffusion model. The method realizes flexible facial editing with high quality and identity consistency. The method comprises the following steps: constructing a description control face diffusion model comprising a noise prediction network and a variational auto-encoder, wherein the noise prediction network comprises a plurality of text alignment face Transform modules and an optimization residual feature module; the method comprises the following steps: inputting an original image and target text description based on a pre-training model, optimizing a target embedding generated by the input target text description by a text alignment face Transform module to obtain an optimized embedding, and performing model fine tuning after optimization is completed; and performing linear interpolation on the target embedding and the optimized embedding to obtain an initial editing result, introducing ArcFace-Loss as identity loss, extracting face features of an original image and the initial editing result through a pre-training model, calculating feature similarity and minimizing the loss, and ensuring consistency of face identities after editing.

A method for assessing the shape of an orthodontic aligner, said method comprising the following steps:a″) acquisition of at least one image at least partially representing the aligner in a service position in which it is worn by a patient, called “analysis image”;b″) analysis of the analysis image by means of a deep learning device, preferably a neural network, trained by means of a learning base, so as to determine a value for at least one tooth attribute of an “analysis tooth zone” representing, at least partially, a tooth on said analysis image, the tooth attribute relating to a separation between the tooth represented by the analysis tooth zone, and the aligner represented on the analysis image, and / orfor an image attribute of the analysis image, the image attribute relating to a separation between at least one tooth represented on the analysis image, and the aligner represented on said analysis image.

Image analysis is a vital field since images can provide contextual, environmental, and emotional factors. Conventional methods are facing challenges in analyzing an image accurately when the image is having lesser data or if the image is having less resolution. Conventional machine learning architectures are computationally intensive when run on high power computing devices for training and inference. The present disclosure provides a robust deep learning model to inference in any given environmental condition. Initially, image data is generated using a pre-trained Generative Adversarial Network (GAN). The GAN receives a plurality of images of varying domain and generates image data. The image data is annotated and segmented to obtain a contextual label map. The contextual label map is given as input to a pre-trained transfer learning model to obtain a plurality of image attributes including number of objects and activity performed by each object.

The invention discloses a fine-grained costume image retrieval method and a fine-grained costume image retrieval device based on large language model common knowledge injection. According to the method, firstly, fine-grained visual features of an input image are extracted through an image encoder, and optimization is carried out in combination with a low-rank adapter, so that the representation capability of an image patch level is improved; thirdly, generating attribute-enhanced common knowledge context through a pre-trained large language model, and enriching image attribute representation, thereby helping the model to understand and infer unknown attribute information in an open scene; according to the method, a switchable mode prompt and interpolation mechanism is introduced, and it is guaranteed that agent embedding can be dynamically supplemented when attributes or texts are missing. In the retrieval process, fine-grained image content matching is carried out based on the relation between image features and attribute enhancement contexts through an attribute-guided cross-modal attention mechanism. According to the method, through multi-modal feature alignment and optimization, the accuracy and robustness of clothing image retrieval in an open world scene are improved.

The invention provides an underwater multi-color space image enhancement method based on depth information guidance, which is characterized in that a model of the method is based on a training model framework capable of utilizing physical priori knowledge and multi-color information, and comprises a depth information generation module, a multi-color space feature fusion module and a depth enhancement module, the relation between a depth image and an underwater scene is enhanced through attention estimation, scene reconstruction training and fusion of multi-color information, so that insensitivity of a single RGB color space to brightness and saturation image attributes is made up; the feature representation capability is improved, and the scene adaptability of the model is enhanced.

A method, a system, and computer program product for optical systems including a metasurface that supports quantitative phase imaging are provided. Optical system characteristics of an optical system are received. The optical system characteristics include an optical path, the optical system comprising a light source and an imaging system. Optical parameters of at least one metasurface are received. A location of the metasurface is selected within the optical path of the optical system to modulate an incident wavefront generated by the light source. An image acquired by the imaging system is processed, based on the optical parameters of the at least one metasurface and based on the location of the metasurface within the optical path, to determine one or more image properties.

The invention provides a remote sensing image scene classification method based on attribute guidance, and relates to the technical field of remote sensing image processing, and the method comprises the following steps: obtaining a to-be-classified remote sensing image and a corresponding scene class text; constructing and training a scene classification network; processing the to-be-classified remote sensing image and the scene category text by using the trained scene classification network to obtain a scene classification result corresponding to the to-be-classified remote sensing image; according to the method, the attribute features of the remote sensing image are utilized, so that the language information can contain each part of the image with differentiated content, and the consistency between the language information and the image content is improved.

Techniques for generating a novel image using tokenized image representations are disclosed. In some embodiments, a method of generating the novel image includes generating, via a first machine learning model, a first sequence of coded representations of a first image having one or more features; generating, via a second machine learning model, a second sequence of coded representations of a sketch image having one or more edge features associated with the one or more features; predicting, via a third machine learning model, one or more subsequent coded representations based on the first sequence of coded representations and the second sequence of coded representations; and based on the subsequent coded representations, generating, via the third machine learning model, a first portion of a reconstructed image having one or more image attributes of the first image, and a second portion of the reconstructed image associated with the one or more edge features.

The application discloses a fine-grained clothing image retrieval method and device for injecting common sense knowledge of a large language model. The method first extracts fine-grained visual features of an input image through an image encoder and optimizes them in combination with a low-rank adapter to improve the representation ability at the image patch level. Then, a pre-trained large language model generates attribute-enhanced common sense knowledge context to enrich the image attribute representation, thereby helping the model understand and reason about unknown attribute information in an open scenario. The application introduces a switchable modal prompt and interpolation mechanism to ensure that the agent embedding can be dynamically supplemented when attributes or text are missing. During the retrieval process, through an attribute-guided cross-modal attention mechanism, fine-grained image content matching is performed based on the relationship between image features and attribute-enhanced context. Through multi-modal feature alignment and optimization, the application improves the accuracy and robustness of clothing image retrieval in an open world scenario.

The present disclosure relates to systems, methods, and non-transitory computer-readable media that generates text and image attributions and provides for display in a digital document the image attribution of an image element and the text attribution of text in the digital document. In particular, the disclosed systems receive a prompt relative to a digital document, and in response, generates an answer to the prompt using a multimodal large language model. Furthermore, the disclosed systems generating an image attribution and a text attribution in response to a selection of at least a portion of the answer to the prompt. Specifically, the image attribution and the text attribution indicate portions of the digital document that provide support for the at least a portion of the answer. Moreover, the disclosed systems provide for display in the digital document of a client device, the image attribution and the text attribution.

A method for training an image generation model includes receiving training data having multiple training images, image captions corresponding to the training images, and corresponding image features, each training image associated with one image caption and one or more image features. The method further includes performing a training process to condition the image generation model using the training images, the image captions, and the image features, resulting in a trained model that generates images conditioned to the image features. The image features include image properties that are extracted from pixels or regions of each training image and camera properties that are associated with each training image.

The present application provides a method and device for thinning flowers and fruits based on image recognition. The method obtains a first image of a fruit tree to be processed captured by an image acquisition device. Based on the first image, the fruit tree type of the fruit tree to be processed is determined. According to the fruit tree type, a flower and fruit thinning model that matches the fruit tree type is determined. Based on the flower and fruit thinning model, a plurality of target information to be processed corresponding to the fruit tree to be processed and the attribute value of each target to be processed are determined. Among them, the target to be processed includes at least a flower bud image and a young fruit image. The attribute value represents the fruit setting rate of the target to be processed after processing. The first image, the second image of each target to be processed and the attribute value of each target to be processed are sent to a preset expert planting model to determine the processing risk value of the target to be processed. Based on the attribute value of each target to be processed and the corresponding processing risk value, the processing target is determined so that the flower and fruit thinning management terminal corresponding to the image acquisition device thins the flowers and fruits of the fruit tree to be processed according to the processing target.

A method, apparatus, non-transitory computer readable medium, and system for image processing include obtaining a style kit including a first image generation input indicating a first image attribute, a second image generation input indicating a second image attribute, and a selectability parameter indicating that the second image generation input is selectable. A third image generation input is received from a user based on the selectability parameter, wherein the third image generation input indicates a third image attribute different from the second image attribute of the second image generation input. An image generation model generates a synthetic image based on the style kit, the first image generation input, and the third image generation input, wherein the synthetic image has the first image attribute and the third image attribute.

Described herein are systems, methods, and instrumentalities associated with estimating the motions of multiple 3D points in a scene and predicting a view of scene based on the estimated motions. The tasks may be accomplished using one or more machine-learning (ML) models. A first ML model may be used to predict motion-embedding features for a temporal state of a scene, based on motion-embedding features for previous states. A second ML model may be used to predict a motion field representing displacement or deformation of the multiple 3D points from a source time to a target time. Then, a third ML model may be used to predict respective image properties of the 3D points based on their updated locations at the target time and / or a viewing direction. An image of the scene at the target time may then be generated based on the predicted image properties of the 3D points.

An extended reality (XR) system is described herein. The XR system includes a display system mounted to the headset for displaying a display screen including computer-generated images thereon, and a controller that includes one or more processors programmed to execute an algorithm for operating in an optical character recognition (OCR) mode to display a text selection screen by generating an image distortion zone within the text selection screen associated with a non-viewable boundary and modifying received video images to adjust a viewable image attribute of a portion of the received video images displayed within the image distortion zone, and operating in the scrolling text mode to display a text display screen by generating machine-readable text OCR and modifying the generating machine-readable text using a lightweight language model for text correction.

The invention discloses an image acquisition method and electronic equipment, relates to the technical field of terminals, and can flexibly adjust the image attribute of an image so as to meet the requirements of flexible adjustment and personalized creation of a user. In the scheme, the electronic equipment can display the shooting interface, and the first preview image and the adjustment items of multiple image attributes such as saturation, color temperature or light and shadow are displayed to a user through the shooting interface, so that the user can adjust the image attributes through the adjustment operation on at least one adjustment item according to individual requirements. And adjusting the display of the preview image according to the corresponding image attribute.

An method of inspecting a tire carcass and either accepting the tire carcass as a candidate for retread or rejecting the tire carcass by identifying anomalies and or defects using an artificial intelligence deep learning image recognition software, whereby the system is improved by use of two dimensional image data augmented with depth data assigned to an image attribute of the a dimensional image for processing by, for example, a convolutional neural network.

A design device and a method for providing a 3D model for a design development of an object. The method includes specifying an object type to a trained artificial neural network. the artificial neural network is trained with a plurality of basic image data of objects of this object type; generating an image data set of the predetermined object type by the trained artificial neural network, the artificial neural network provides image data from a plurality of perspectives as an image data set; checking the generated image data set for the presence of a design selection criterion, if the design selection criterion is not met, image data are regenerated by a change in an image property criterion; generating a digital 3D model from the image data set.

The application discloses a garbage classification model training method, device and equipment. A set proportion of garbage images is selected from garbage images, and image properties of regions where each target is located in the selected garbage images are adjusted to obtain a first data set. The first data set includes garbage images whose image properties are adjusted and garbage images whose image properties are not adjusted in original garbage images. The garbage images have been labeled with categories of the targets and regions where the targets are located. Garbage images including multiple targets that overlap with each other in the first data set are copied, and the copied garbage images are added to the first data set to obtain a second data set. The copied garbage images are labeled with one of the multiple targets that overlap with each other. A target detection model is trained based on the second data set. The application not only improves the efficiency of garbage classification model training, but also further improves the accuracy and detection speed of garbage classification performed by the garbage classification model.

The invention relates to an image enhancement device with gaze tracking. An electronic device may have a display and a camera. Control circuitry in the device may collect information about a user's gaze point using a gaze tracking system and other sensors, may collect information about real-world images by analyzing the real-world images, such as information about content, motion, and other image attributes, and may also collect information about the user's gaze point by analyzing the real-world images. User visual information such as user visual acuity, contrast sensitivity, field of view, and geometric distortion may be acquired, user inputs such as user preferences and user mode selection commands may be acquired, and other inputs may be acquired. Based on the gaze point information and / or other acquired information, the control circuit may display real-world images and supplemental information on the display. The supplemental information may include enhancements such as icons, text tags, and other computer-generated text and graphics overlaid on the real-world image, and may include augmented image content such as an amplified portion of the real-world image.

The present application relates to the technical field of image processing, and particularly relates to a scenic spot image processing method and device, equipment and a storage medium, the method comprising obtaining a to-be-processed scenic spot image, determining a target region in the to-be-processed scenic spot image, removing the target region in the to-be-processed scenic spot image to obtain a removed image; obtaining image attribute information of the to-be-processed scenic spot image, obtaining a similar image corresponding to the removed image according to the image attribute information; determining a target region in the similar image, completing the removed image according to the target region in the similar image to obtain a completed image; inputting the completed image and the image attribute information into a preset image inpainting neural network model to perform image inpainting on the completed image, and obtaining an inpainted image of the to-be-processed scenic spot image, which realizes the elimination of sundries and inpainting of the to-be-processed scenic spot image, and meets the image processing requirements of users.

A system and method performed by an electronic device for generating a high-resolution image from an input low-resolution image are provided. The method includes determining one or more feature vectors for each of attributes associated with an input low-resolution image based on one or more image attributes, determining a degradation representation map of the input low-resolution image based on at least one of the input low-resolution image and the one or more feature vectors of the one or more image attributes, generating one or more personalization parameters based on at least one of the input low-resolution image and a user input, obtaining a first output image based on degradation information, a first process, and the one or more personalization parameters, obtaining a second output image based on the degradation information, a second process, and the one or more personalization parameters, and generating the high-resolution image by upsampling at least one of the first output image and the second output image.

The application provides an image encryption method and device, equipment, storage medium and program product, and relates to the technical field of privacy calculation. The method obtains an original image, generates a chaotic sequence corresponding to the original image, generates a secure thumbnail image and a chaotic compressed ciphertext image that reserve visual features of the original image based on the chaotic sequence and the original image, and embeds the chaotic compressed ciphertext image in the least significant bit of the secure thumbnail image to obtain a visually perceptible image. The method solves the contradiction between security and recognizability of the traditional encryption method through the fusion of multi-level encryption and visual perception technology, significantly improves the security and user recognizability of image transmission, reduces the possibility of identification by attackers, and ensures that users can quickly identify image attributes through visual preview.

An image processing method and apparatus, an electronic device, and a storage medium are provided, the image processing method includes: processing an image to be processed to obtain a panoramic complementary image of a target pixel ratio according to an image attribute of the image to be processed; determining a plurality of target patch maps on a bounding box according to the panoramic complementary image; wherein a displayed content on the bounding box corresponds to the panoramic complementary image; and determining a panoramic surround image based on the plurality of target patch maps.

Optical image capture unit (10) for inspecting sewer pipes and which can be coupled to a sewer inspection system, comprising: a housing (12, 14) which can be coupled to the sewer inspection system; an image sensor (20) arranged in the housing for capturing image information of an image of an object displayed on the image sensor, a lens (16) arranged in or on the housing for generating the image of the object on the image sensor (20), a lens (18) arranged between the objective (16) and the image sensor (20), the imaging properties of which can be changed by means of an input signal (19) in order to focus the image of the object produced by the objective onto the image sensor (20); and a control unit (22) arranged in the housing (12, 14) which is designed to generate the input signal (19) based on the image information.

Embodiments of the present application disclose an image generation method and device, computer equipment and a storage medium; an image generation instruction can be received, the image generation instruction including image attribute information of a target image to be generated, environment information is obtained, the environment information is collected through a display device displaying the target image, data mapping is performed on the environment information through a preset image generation model, a content image corresponding to the environment information is generated, style features are extracted from at least one preset style reference image, content features are extracted from the content image, and the target image is obtained based on the fusion of the content features and the style features; in the embodiments of the present application, the image generation model can map the environment information to obtain the content image, the content features in the content image can be fused with the style features in the style reference image, and finally the target image is generated, so that a wallpaper image with rich content and various styles can be generated, and the visual experience of a user is improved.

The embodiment of the present disclosure discloses an object information determination method, device, equipment and storage medium, the method comprises: acquiring a panoramic image and image attribute information of the panoramic image; determining pixel point matrix information of the panoramic image according to the image attribute information of the panoramic image, the pixel point matrix information includes depth direction information of each pixel point in the panoramic image relative to the image capture position point; identifying the panoramic image, determining a target object in the panoramic image and object region information of the target object; and determining region attribute information of the target object according to the pixel point matrix information and the object region information. By using the method, the region attribute information of the object in the real scene can be effectively determined through the analysis of the panoramic image, the determined region attribute information of the scene object can be used as data support to participate in the actual business operation, and the cost investment of introducing the region attribute information from the data information provider is reduced.

The invention discloses a multi-level dynamic layer management method and system, and belongs to the technical field of computer graphic processing and data management, and the method comprises the following steps: obtaining layer data from a plurality of external data sources, converting the layer data into a uniform format in the system, and importing the layer data into the system; constructing a multi-level tree structure system for the imported layers according to a preset rule or a user-defined mode so as to present a hierarchical relationship between the layers; marking the basic graphic image attribute of each layer, the position of each layer in the hierarchical system, the association relationship between each layer and other layers, and the business-related attribute; responding to various dynamic adjustment operations in real time, and updating the layer state according to the layer attribute and the association relationship; a visual and convenient operation interface is provided, and a multi-level graph layer structure tree is displayed; and different layer operation authorities are allocated for different user roles. The method is widely applicable to layer management systems in multiple fields, and the overall efficiency and practical value of layer management can be improved.

The invention relates to the technical field of gastric cancer image recognition, and discloses a gastric cancer early lesion image recognition method based on deep learning. The method comprises the steps of obtaining a to-be-analyzed gastric cancer medical image set, and screening an adaptive deep learning model group according to image attributes and recognition requirements; efficiency indexes such as model memory occupancy, single-frame processing delay and throughput rate are continuously tracked, priorities are evaluated for candidate models according to tracking data, and a model activation sequence list is formed. Selecting models one by one according to the list, and checking whether all the assigned images can be completed within a specified time limit or not: if the checking is passed, allocating computing resources to initialize a single model copy, and if the checking is not passed, deducing the number of the required model copies. Checking available computing resources, if the available computing resources are sufficient, starting corresponding copies, and if the available computing resources are insufficient, recovering low-priority activated model resources and then checking again; finally, whether all the active models are located at the top of the list or not is verified, if yes, scheduling is terminated, and if not, remaining models of the list continue to be processed.

The invention discloses a face image attribute editing method based on a generative adversarial network, and the method comprises the steps: firstly obtaining a target image with a target attribute and a source image without the target attribute, and obtaining a neutral diagram of the source image through a pre-trained generative adversarial network; secondly, mapping the source image, the neutral diagram of the source image and the target image into a multi-layer implicit representation through an encoder; and then extracting a to-be-edited hidden unit from the hidden representation, mutually exchanging the to-be-edited hidden units of different levels in the source image hidden representation and the target image hidden representation according to the target attribute, and inputting the final level source image hidden representation into a decoder to generate a new image containing the target attribute. Finally, a face image attribute editing model is constructed and comprises a generator and a discriminator, the generator is composed of the encoder and the decoder, and training is carried out. The method can ensure that the generated image is closer to the target image in local details, and realizes accurate attribute editing.