966results about "Still image data querying" patented technology

A data processing system implements obtaining an image of a document and an indication of one or more content items to generate; analyzing the image of the document to generate a textual representation of contents of the document in the image; constructing a query based on the textual representation; analyzing the query using a second machine learning model to obtain embeddings representing one or more categories of information represented in the query; searching a knowledge graph based on the query embeddings to obtain results of the query; providing the query results to a content generation unit to generate the one or more content items based on the results of the query; and obtaining the one or more content items from the content generate unit.

A method and apparatus for generating photorealistic synthetic images by receiving multiple forms and multiple instances of user input corresponding to a user's visual idea, executing an iterative image search to identify pre-existing images semantically aligned with the user's visual idea, and using an image synthesis deep learning model to generate at least one synthetic image based on the multiple forms and instances of user inputs.

Embodiments described herein provide systems and techniques for training large language models. In one aspect, a process for performing in-context training of a language model is disclosed. This process may begin by receiving a language model that includes a context window of a predetermined size, as well as receiving a set of in-context prompt / completion pairs prepared for a target task. The process then constructs a first token sequence based on the set of in-context prompt / completion pairs. Next, the process fits the first token sequence into the context window. The process subsequently performs a first in-context training pass using the first token sequence to train the language model to generate a next token in accordance with the target task.

The invention relates to the technical field of image recognition, in particular to a harmful image detection and recognition system and method based on agent collaboration, and the system comprises a first agent module, a second agent module, a third agent module, a fourth agent module and a fifth agent module, the first agent module is used for analyzing image element semantic information, outputting structured data and constructing an image comprehensive semantic knowledge graph; and the second intelligent agent module is used for carrying out harmful feature detection and associated query on the image in combination with the structured data, and outputting a coarse-grained harmful semantic detection result and the confidence coefficient of the coarse-grained harmful semantic detection result. According to the harmful image detection and identification system based on intelligent agent cooperation, through hierarchical processing and cooperative work of the first to fifth intelligent agents, comprehensive analysis of the image is realized to obtain meta-semantic information, coarse-grained and fine-grained harmful semantic detection is performed on the meta-semantic information, and through multiple rounds of challenge and dynamic comprehensive analysis, the detection and identification efficiency of the harmful image is improved. Hidden harmful information can be accurately identified, so that the accuracy and interpretability of the harmful information are improved.

The invention provides a visualization system capable of synchronously displaying multiple types of data according to time. Comprising a distributed service cluster composed of a micro-service, a user management micro-service, a data transmission micro-service, a data storage management micro-service, a data extraction micro-service, a data visualization display micro-service, a data visualization template management micro-service and a resource monitoring micro-service; the micro-services are packaged in a containerized manner and interact by adopting a lightweight communication protocol, and automatic deployment and dynamic expansion are realized through a container arrangement tool; the system is provided with a basic parameter set with customizable micro-service configuration and a graphical configuration interface which are used for cooperatively generating a system operation environment configuration strategy. The user management micro-service establishes a multi-dimensional authority system based on a role access control model, and records an operation log to realize audit tracking; the data transmission micro-service integrates a plurality of heterogeneous data access modes, carries out transmission process verification and carries out secure channel transmission through the service gateway.

The invention discloses a laser point cloud shielded vehicle completion method and device based on a thunder-vision fusion deep learning framework in the technical field of automatic driving environment perception. The method comprises the following steps: acquiring original laser point cloud data and an aerial image; constructing a fan-shaped shielding area based on the original laser point cloud data; identifying two-dimensional bounding boxes, orientations and category labels of all vehicle targets in the image; inputting the fan-shaped occlusion area and the two-dimensional bounding boxes, orientation and category labels of all the vehicle targets into a pre-trained double-branch deep learning network, and predicting whether an occluded vehicle exists or not and the position, orientation and category information of the occluded vehicle; and selecting a vehicle point cloud template based on the category information of the shielded vehicle, and then generating a scene point cloud after vehicle point cloud completion as a final result of vehicle completion in the shielded region. According to the invention, an image-point cloud space mapping mechanism based on the Leiyu fusion deep learning framework is introduced, so that the capability of automatically identifying and positioning the shielded vehicle is remarkably improved.

The invention relates to a Chinese language and literature computer networking query reading system. According to the system, a query request processing module carries out mixed feature analysis processing according to a query request of a user, and a mixed query vector fusing radicals and semantics is generated; a distributed retrieval module performs distributed retrieval in a preset space-time fragment index structure based on the mixed query vector to obtain a candidate literature data set; the preliminary retrieval optimization module is used for carrying out self-adaptive conversion processing on abnormal characters and traditional and simple characters in the candidate literature data set according to a Chinese character evolution rule and a context to obtain standardized text data; and the visual query output module performs knowledge graph enhancement processing on the standardized text data to generate a visual query result. According to the system, efficient query and visual display of Chinese language and literature works are achieved through multi-modal processing, multi-dimensional indexing and knowledge graph technologies, and the problems that a traditional query mode is single in input, limited in retrieval dimension, insufficient in text processing and the like are effectively solved.

The invention provides an image data and pattern spot management system and method and electronic equipment, and relates to the technical field of geographic information data processing, and the image data and pattern spot management system comprises an image data indexing module which is used for carrying out three-dimensional indexing on a remote sensing image to obtain a space-time-data source association chain of the remote sensing image, the remote sensing image contains pattern spot data with an identifier; the pattern spot complete cycle construction module is used for performing multi-temporal analysis on the pattern spot data to obtain an analysis result, and constructing a pattern spot change life cycle map of the pattern spot data in combination with the association chain; and the tracing module is used for positioning the target pattern spot when the target pattern spot needs to be traced to obtain a change node of the target pattern spot and a space-time-data source index corresponding to the change node, and searching in the association chain according to the change node and the space-time-data source index to obtain a traceable data chain. According to the invention, the storage and management of the geographic image data and the pattern spot data are more systematized and refined.

Apparatus for visualization within a three-dimensional (3D) model and methods used therein are described, wherein the apparatus includes a processor and a memory communicatively connected to the processor, wherein the memory includes instructions configuring the processor to receive a query image, extract neural network encodings from the received query image, query a synthetic image repository for at least a matching synthetic image, and display an estimated position and orientation within the 3D model, wherein the synthetic image repository includes a plurality of synthetic images and their extracted neural network encodings, each synthetic image therein corresponds to a slice extracted at a specific position and orientation in the 3D model, and querying the synthetic image repository includes comparing the extracted neural network encodings between the query image and synthetic images.

Systems and methods for image search result filtering can include obtaining a search query, determining a plurality of candidate image search results, processing the search query with a generative model to determine a plurality of search result criteria, and refining the plurality of candidate image search results based on determining whether the candidate results satisfy the plurality of search results criteria. The systems and methods can perform a plurality of determinations based on the output of the generative model.

The invention provides an AI image detection method and system for an electronic circuit board, and relates to the technical field of circuit board production and detection. Compared with a traditional artificial vision detection method, the method has the advantages that preliminary detection and accurate detection are carried out through real-time images of all production procedures of the circuit board, automatic defect detection of the electronic circuit board is achieved, manual participation is not needed, and the defect detection efficiency is improved. Besides, preliminary detection and accurate detection are performed through the first AI image detection model and the second AI image detection model, so that the resource occupation amount in the defect detection process can be reduced, the real-time performance of defect detection of the electronic circuit board is ensured, and the defect detection efficiency of the electronic circuit board is further improved.

The invention relates to a task distribution method and system based on a zero-work market, and a terminal, and relates to the technical field of the zero-work market, and the method comprises the steps: obtaining a preset trigger signal of a mobile terminal; when the trigger signal is consistent with a preset search signal, controlling the mobile terminal to carry out article shooting prompt to obtain article image information; determining article features in the image according to the article image information; determining whether a preset domain database contains an article corresponding to the article feature in the graph or not; if the domain database contains an article corresponding to the article feature in the graph, determining a specific shot article according to the article feature in the graph; determining an appropriate working field and an appropriate field number according to the specific shot object; and when the proper field number does not exceed the preset field excessive number, controlling the mobile terminal to display the working content of the proper working field according to the proper working field. The task allocation method and device have the effect of improving the task allocation efficiency.

The invention provides a molten steel transportation track playback method and device, electronic equipment and a medium, and the method comprises the steps: processing sample transportation data, extracting key transportation track information, and rapidly and accurately obtaining target track data through a preset query condition; the timestamp data provides a basis for the time sequence analysis of the track, each important node in the transportation process can be clearly reflected, the acquisition of the image data enables a user to check the state at a specific time point, the operability of the data is enhanced, track playback is performed by establishing a thread, the dynamic display of the transportation process can be realized, and the transportation efficiency is improved. And through graphical operation and abstract map display, the historical production condition can be backtracked more easily, and the user experience is better.

The invention provides a deep hash image retrieval method based on hierarchical multi-scale feature fusion. The method comprises the following steps: firstly, adopting a Nested Hierarchical Transform as a backbone feature extraction network, and synchronously capturing global context information and local detail features of an image through a nested local self-attention mechanism and a hierarchical feature aggregation module; then, constructing a multi-scale feature fusion module, extracting visual features of different receptive fields in parallel by using a multi-branch unequal-ratio expansion rate dilated convolution kernel, and adaptively weighting important features in combination with attention to generate discriminative multi-scale fusion features; and finally, designing a mixed loss function, optimizing intra-class feature compactness through center similarity loss, and realizing efficient and semantic similarity keeping Hash code generation in combination with a quantization loss constraint Hash code discretization process. According to the method, full expression of image features is realized through a collaborative mechanism of Transform and multi-scale feature fusion, and the precision and efficiency of large-scale image retrieval are improved through an end-to-end deep hash learning framework.

The invention relates to the technical field of intelligent information retrieval, in particular to an image-text retrieval generation method and device based on reference semantics and electronic equipment. The method comprises the steps of performing semantic segmentation on a document in a document knowledge base to obtain a plurality of paragraphs, performing semantic similarity judgment, and merging similar paragraphs; extracting semantic features of the merged paragraphs to form semantic features of lower-layer paragraphs; abstract description is carried out on the merged paragraphs, semantic features are extracted from abstract description, upper layer description semantic features are formed, and text level semantic features are constructed; image semantic features in the image knowledge base are extracted, normalization processing is carried out on the text level semantic features and the image semantic features, an inverted product quantitative index is constructed, and construction of an image-text reference semantic feature index is achieved; and processing the input text based on the image-text reference semantic feature index to obtain a retrieval result. According to the method, the problems of lack of previous knowledge, insufficient detail information and poor image-text relevance in image-text retrieval are solved.

The invention discloses a professional knowledge domain data processing method based on a large language model, and relates to the technical field of natural language processing, and the method comprises the following specific steps: knowledge graph-driven multi-modal preprocessing: collecting professional domain knowledge graph, text, image and audio data, according to the method, the data quality is effectively improved through multi-modal preprocessing driven by the knowledge graph, a solid foundation is laid for subsequent processing, the attention regulation and control algorithm of multi-modal perception is combined, a large language model can dynamically adjust attention according to modal features such as text context, images and audios, key information is accurately captured, and the method is suitable for popularization and application. The comprehensive understanding and processing capability of professional knowledge data is remarkably improved, and in addition, the generalization capability of the model is further enhanced through the modal self-adaptive data enhancement method, so that the model can better deal with diversity and complexity in the professional field.

The invention discloses a multi-mode convergence media content auxiliary creation method based on an AI technology, and the method comprises the steps: extracting key features according to a creation demand text inputted by a user, carrying out the retrieval in a knowledge base through employing a knowledge graph retrieval mode according to the key features, obtaining a creation material, and generating a first draft; carrying out cross-modal feature mapping on the first draft by adopting a multi-modal alignment model, aligning text-image-video embedded vectors through contrastive learning, and dynamically adjusting the correlation of multi-modal contents by utilizing an attention mechanism to obtain a multi-modal content packet; duplicate checking is carried out on the multi-modal content packet in multiple modes, and the multi-modal content packet after duplicate checking is sent to a user for manual editing. Through multi-mode processing methods such as AI auxiliary writing, AI illustration and video generation and whole-network duplicate checking and propagation value evaluation, the content production efficiency and quality are improved, the propagation effect is optimized, and the original content is protected.

The invention discloses a medical image report generation method based on multi-modal large model retrieval enhancement, which relates to the technical field of medical image processing and comprises an image vector model module, a vector database and similarity detection module, a multi-modal large model module, a dynamic retrieval module and a retrieval enhancement generation prompt project module. The multi-modal large model module comprises a multi-modal input splicing unit, a reasoning information processing unit and a diagnosis report set generation unit, the dynamic retrieval module comprises a similarity threshold value analysis unit and a retrieval quantity analysis unit, and the similarity threshold value analysis unit is used for analyzing a maximum similarity value between a new image and a database. According to the invention, an image vectorization processing technology based on a CLIP-ViT model is adopted, so that the system can accurately capture fine features of a focus, and the risk of key information omission is reduced; a dynamic threshold adjustment algorithm is introduced, and the rigidity defect of a traditional fixed retrieval strategy is overcome.

The invention provides an OCR (Optical Character Recognition) and network electronic identity authentication no-certificate comparison information system and device and an inspection method, and relates to the technical field of information security. Firstly, the OCR technology is adopted to recognize certificate characters and upload names and identity cards, face key point data are extracted through a deep learning model, deep feature vectors are output, a local end preferentially queries database electronic identity cards through the names and the identity card numbers, similarity is calculated through a cosine similarity formula, offline authentication is supported, and the face recognition is realized. And if not, network electronic identity authentication is realized through a cloud identity authentication system, and an electronic identity card is returned by calculating the similarity. Meanwhile, based on the track, communication and consumption data, risk indexes are calculated through normalization processing and a weighted summation formula, the risk indexes are compared with a preset risk threshold value to evaluate the risk level, when the audited personnel are evaluated to be high in risk, the track, communication and consumption data are synchronously returned, and identity authentication and risk prevention and control in the non-certified scene are achieved.

A cross-view-angle image retrieval method for unmanned aerial vehicle navigation comprises the following steps: step 1, enhancing an image, step 2, based on a ConvNeXt network for cross-view-angle image retrieval for unmanned aerial vehicle navigation, performing multi-scale feature extraction on the enhanced image, and step 3, performing multi-scale feature extraction on the enhanced image based on a ConvNeXt network for cross-view-angle image retrieval for unmanned aerial vehicle navigation. 3, performing multi-dimensional feature fusion based on an attention mechanism; 4, after multi-dimensional feature fusion, calculating a loss value, and based on a determined weight, obtaining a detection model for image retrieval; through the multi-scale feature convolution and multi-dimensional feature fusion technology, the technical problems of insufficient matching precision, poor retrieval robustness and the like caused by view angle, scale and structure differences between the satellite image and the unmanned aerial vehicle image are solved; the invention further comprises a system, equipment and a medium for implementing the method.

The invention discloses a multi-modal data retrieval method and device under dynamic collaboration and a medium, belongs to the technical field of information retrieval and natural language processing, and is used for solving the problems that unstructured data is difficult to process efficiently in traditional data retrieval, an information island phenomenon exists, and the data retrieval efficiency is low. And an effective connection and integration mechanism is lacked between data from different sources. The method comprises the following steps: performing task allocation processing on a corresponding Agent cluster on a question type to be queried to obtain an Agent task list; performing data query processing on the Agent task list under different data sources, and obtaining query result data; performing data primary processing on the query result data to determine initial retrieval data; performing multi-modal consistency check on the initial retrieval data, and integrating verified integrated retrieval data; and performing information strategy generation processing based on a fact verification mode and an explanation enhancement mode on the integrated retrieval data to obtain final answer information.

The embodiment of the invention discloses a building engineering crack detection method and system based on image recognition, and the method comprises the steps: obtaining a building surface image, carrying out the preprocessing of the image, obtaining a standardized image, and carrying out the multi-scale decomposition extraction and integration of various features, and forming a multi-dimensional feature descriptor set; after feature importance is evaluated, a compact feature vector is generated through dimension reduction, quantization coding and compression, and then a multi-level feature index mechanism for optimized compression is constructed. A query feature vector is extracted from a newly collected image, searching and screening are completed by means of an index mechanism and a tolerance threshold, and a crack matching result is obtained; and based on the result, positioning cracks, classifying types, measuring parameters and evaluating severity, and generating a crack state report. The crack trend is analyzed in combination with the historical data time sequence, a multi-stage early warning mechanism is designed, maintenance suggestions are provided, and a real-time monitoring and early warning system is formed. According to the embodiment of the invention, the technical problems of high storage pressure and low real-time detection efficiency in the prior art can be effectively solved.

The invention belongs to the computer vision technology, and particularly relates to a small sample multi-modal prototype learning-oriented remote sensing scene classification method, which comprises the following steps of: acquiring RGB (Red, Green and Blue) images with category labels and text prompts of the RGB images as a support set; establishing a text prototype, an RGB prototype and a hyperspectral prototype of each category according to the support set; and extracting to-be-classified query set image features by using a pre-trained CLIP image encoder, calculating cosine similarities between the query set image features and the text prototype, the RGB prototype and the hyperspectral prototype of each category of the support set, taking the cosine similarities as input of a multi-layer perceptron, and obtaining the category of the to-be-classified RGB image through classification of the multi-layer perceptron. High-precision and high-robustness remote sensing scene classification is realized under the small sample condition, only prototype and similarity calculation is needed in the reasoning stage, and deployment and expansion are easy.

The invention relates to the technical field of data management, and discloses a database management system based on building remote sensing images. Comprising the following steps: identifying a target image from a building remote sensing image, dividing the target image into an image set, correspondingly importing the image set into a data layer of a library architecture, generating an image database, constructing an associated channel with an on-off point, configuring on-off conditions on the associated channel, analyzing a query demand from request data, and storing the query demand in the database. Selecting a corresponding query mode; compared with the prior art, the method has the advantages that the image association and query channels can be constructed among different data levels in the image database, and meanwhile, the on-off points configured with the on-off conditions are combined, so that the target images in the image database can be dynamically associated through the association channels on the basis of keeping independent, and the dynamic association efficiency is improved. Therefore, the dynamic storage association effect of the target image in the image database is realized, and the diversified query requirements of the image database are met.

The invention is suitable for the technical field of virtual fitting, and particularly relates to an e-commerce virtual fitting method and system based on multi-modal query, and the system comprises a graph-graph module which is used for starting from a blurred image / video and matching to a real commodity graph; the text-graph module is used for matching a real commodity graph starting from the query text; the virtual fitting module is used for generating a freely editable portrait with fixed clothes and optional conditions; and according to query modal information input by a user, automatically selecting a corresponding query module, and inputting the matched commodity clothing image into a virtual fitting module to generate a virtual fitting image. According to the method, large model assistance and a potential diffusion model are introduced, and a multifunctional plug-in is combined, so that one-stop operation is realized, free and editable virtual fitting images are completed, the shopping experience of a user can be remarkably improved, refunding and changing caused by improper goods are reduced, the operation cost of a merchant is reduced, and meanwhile, comprehensive and personalized clothing display of the merchant is supported.

The invention discloses a power grid GIS data editing method and system, electronic equipment and a storage medium. The method comprises the steps that a server constructs power grid vector slice data based on power grid GIS data and converts power grid equipment data into GeoJSON data; the client obtains and loads the vector slices, picks up a power grid graph from the vector slices according to the position of the edited equipment, adsorbs the power grid graph to the edited equipment, and requests the server for GeoJSON data of the corresponding equipment; the server performs query based on the edited device to obtain a device associated with the edited device, and sends GeoJSON data of the edited device and the associated device to the client; and the client establishes a relation with the GeoJSON data according to a pre-configured vector slice layer, captures an editing operation, superposes the vector slice and the GeoJSON data and then displays the superposed vector slice and GeoJSON data in real time. The problems of graph rendering and large data volume online editing are solved by using a data layering mechanism combining vector slicing and GeoJSON, and the rendering efficiency is improved.

The invention relates to a crystal pixel lookup table generation method and device, computer equipment and a storage medium. Determining a preset number of first pixel peak points from the first crystal pixel lookup table, performing data enhancement processing on the first pixel peak points in the first crystal pixel lookup table to obtain a second crystal pixel lookup table, and performing data enhancement processing on the first pixel peak points in the first crystal pixel lookup table to obtain a second crystal pixel lookup table, and obtaining a target crystal pixel lookup table based on the first crystal pixel lookup table and the second crystal pixel lookup table. According to the embodiment of the invention, the second crystal pixel lookup table is obtained by performing data enhancement processing on the first pixel peak point, and the phenomenon that the crystal of the medical equipment disappears along with the use time is simulated, so that the target crystal pixel lookup table is obtained based on the first crystal pixel lookup table and the second crystal pixel lookup table; the number and diversity of crystal pixel lookup table data sets are improved, and the complexity of crystal pixel lookup table generation is reduced.

The present disclosure provides a character recognition model training method and apparatus, a character recognition method and apparatus, a device and a medium, relating to the technical field of artificial intelligence, and specifically to the technical fields of deep learning, image processing and computer vision, which can be applied to scenarios such as character detection and recognition technology. The specific implementing solution is: partitioning an untagged training sample into at least two sub-sample images; dividing the at least two sub-sample images into a first training set and a second training set; where the first training set includes a first sub-sample image with a visible attribute, and the second training set includes a second sub-sample image with an invisible attribute; performing self-supervised training on a to-be-trained encoder by taking the second training set as a tag of the first training set, to obtain a target encoder.

The invention relates to an unmanned aerial vehicle scheduling method and system for emergency rescue, a terminal and a storage medium, and belongs to the technical field of emergency rescue, the unmanned aerial vehicle scheduling method comprises the following steps: sending an inspection instruction to a front-end unmanned aerial vehicle cluster, and the front-end unmanned aerial vehicle cluster performing inspection according to a preset inspection path; after the front-end unmanned aerial vehicle cluster finds the disaster site, receiving a preliminary assessment report of the disaster site sent by the front-end unmanned aerial vehicle cluster; the front-end unmanned aerial vehicle cluster generates and uploads a preliminary evaluation report after discovering a disaster site; inputting the preliminary evaluation report into a pre-constructed scoring model to obtain a score value of the preliminary evaluation report; calling a current dispatching scheme matched with the score value from a historical dispatching database; and sending a rescue instruction to the standby unmanned aerial vehicle cluster and the rescue terminal according to the current dispatch scheme. The method has the beneficial effects that the response time of the rescue unmanned aerial vehicle is shortened, so that the search and rescue efficiency is improved.

The invention provides a promotion image-text generation method and device and a model training method and device, and relates to the technical field of computers, in particular to the technical fields of large language models, data processing, intelligent promotion and the like. According to the specific implementation scheme, user information and product information are obtained; based on the user information and the product information, generating an article abstract by using a first model, the article abstract including event information related to the user information and the product information; based on the article abstract and the product information, utilizing a second model to generate a promotion article; and determining a promotion picture corresponding to the promotion article according to the promotion article, and fusing the promotion article and the promotion picture to determine a promotion image-text. According to the invention, the generation efficiency and quality of the promoted image-text can be improved.