756 results about "Text recognition" patented technology

The invention relates to the field of computer vision, and particularly provides a table recognition reconstruction method and system, a terminal and a medium, and the method comprises the steps: firstly decomposing a large-size table image into a plurality of overlapped sub-images, and carrying out the table structure detection and OCR character recognition of each sub-image through parallel recognition; then, sub-graph recognition results are integrated through a coordinate mapping and confidence coefficient weighted fusion algorithm, and boundary errors are eliminated; then, automatically distinguishing common cells based on an area clustering algorithm, merging the cells and a header region, and reconstructing a complete table logic structure; further understanding header semantics through a natural language model and repairing identification errors; and finally, realizing intelligent splicing and standardized output of the cross-page table. According to the method, the memory limitation of the traditional OCR technology is broken through, an oversized table can be processed, the recognition accuracy of a complex structure is improved, and the digitization efficiency of professional documents such as financial statements and engineering drawings is improved.

The invention discloses a method for identifying an engineering drawing detail table and generating a BOM table, and belongs to the crossing field of automation technology and image processing, and the method comprises the steps: obtaining a scanning or electronic image of an engineering drawing; through preset datum line positioning, recursively detecting a nested rectangular region conforming to an area difference threshold value, and determining a title bar, a detail list region coordinate and a table image; identifying lines and cross points, analyzing the line and column boundaries of the table, and constructing a topological structure; performing character recognition by adopting multi-engine OCR integration; the characters and the cells are associated, a header is recognized through semantics, and analysis data subjected to integrity verification are generated; and automatically generating a structured BOM table in a preset standard format based on the data, and outputting an editable file. According to the method, full-process automation is achieved, manual intervention is not needed, and manual input cost and personal errors are greatly reduced.

The invention discloses an industrial drawing analysis method and system combined with a multi-modal large model and OCR, and relates to the technical field of drawing analysis, the method comprises the following steps: carrying out layout area segmentation, text recognition and geometric element extraction on an industrial drawing to obtain a structured information set; constructing a multi-relation structure chart set of the industrial drawings; performing structure embedding and feature bias enhancement to obtain a structure feature set; performing semantic understanding on the industrial drawing to obtain semantic features, and performing multi-channel coding to obtain a multi-modal feature set; obtaining a preliminary analysis result set of the industrial drawing; and performing rule verification, and outputting an industrial drawing analysis result. The technical problems of low industrial drawing analysis efficiency, inaccurate automatic scheme analysis and limited processing capacity in the prior art are solved, and the technical effects of realizing full-process automatic analysis of the industrial drawing by combining the multi-modal large model and the OCR, improving the precision and efficiency of industrial drawing analysis and standardizing output information are achieved.

The invention relates to the field of drawing recognition, in particular to a PDF drawing data extraction method and system based on intelligent recognition. Comprising the following steps: reading an internal structure of a PDF engineering drawing to obtain a native text stream, a vector path and a grating image; identifying the native text flow through a shunt preprocessing framework to form structured text data; rendering the vector path and the grating image to obtain a background image; analyzing the structured text data by utilizing the intelligent recognition model through the character recognition and extraction sub-model, obtaining drawing metadata and recording the position, and obtaining a character recognition result; analyzing the background image through a graphic element recognition and classification sub-model, recognizing and classifying component elements, and obtaining a graphic recognition result; and performing fusion according to the visual space corresponding relation to form a drawing analysis result. According to the method, the adaptive capacity of engineering drawings with various sources and different qualities is improved through the shunting preprocessing framework and the intelligent identification model.

The invention discloses a video character recognition and erasing method and system, a storage medium and an electronic device. The method comprises the following steps: acquiring a video; performing frame extraction processing to obtain a video frame picture; the method comprises the following steps: acquiring all text contents and coordinate data in a picture through OCR (Optical Character Recognition) identification, analyzing and identifying flower characters and subtitles on a video frame picture through a multi-modal large model, screening out a region of the flower characters and the subtitles needing to be erased through coordinate matching, and determining the region as an erased region; and intelligently erasing and repairing the flower and subtitle areas by adopting a video repairing technology, recovering the original state of the video, and generating an erased new video file. According to the method and the device, the full process of automation is realized, subtitles and flower characters in the video do not need to be manually marked, identified and erased, printed characters of commodities / packages are reserved, mistaken erasure is avoided, and the processing efficiency of video reediting is greatly improved.

The invention discloses a text recognition model training method and device, electronic equipment and a storage medium. The method comprises the steps of performing text prediction on a first training sample set of a current training round based on a first text recognition model of the current training round to determine a first prediction result and a first confidence coefficient of the first prediction result; selecting a first reference sample from each first training sample based on each first confidence coefficient; adjusting parameters of the first text recognition model based on a first loss value determined by the first prediction result to obtain a second text recognition model of the next training round; and determining a second reference sample based on the first recognition model and the first reference sample, adding the first reference sample and the second reference sample into the first training sample set to obtain a second training sample set of the next training round, and training a second text recognition model based on the second training sample set. According to the method, the problems of low text recognition accuracy and insufficient generalization ability of a text recognition model in a specific scene are solved.

The embodiment of the invention discloses an architectural drawing signature character recognition and control method and device, equipment and a medium. A specific embodiment of the method comprises the steps of obtaining an original architectural drawing image; performing full-graph character perception detection on the original building drawing image to obtain coordinate information of a full-graph detection textbox; generating an image label ROI image based on the coordinate information of the total image detection textbox; self-adaptive direction correction is carried out on the picture label ROI image, and a picture label ROI image after direction correction is obtained; performing scale space transformation enhancement on the ROI image of the picture label after the direction correction to obtain an enhanced ROI image of the picture label; performing structured semantic analysis extraction on the enhanced drawing tag ROI image to obtain structured drawing metadata; and based on the structured drawing metadata, downstream physical equipment is controlled to execute associated automation operation. The implementation mode provides key technical support for digital management of the constructional engineering drawings.

The invention provides a metal imprinting character recognition method and system fused with multi-direction double-exposure image deep learning. The method comprises the following steps: step 1, collecting double-exposure images in four directions on the surface of a metal imprinting character; 2, fusing the double exposure images based on a differentiated weight distribution strategy of the regional information amount; step 3, constructing a lightweight improved EAST imprint character target detection model, and realizing detection of an imprint character region through a trunk feature extraction network, a self-adaptive channel attention mechanism, a fusion network and an output layer; and 4, character recognition: adopting a text recognition algorithm SVTR based on a single vision model, and completing the recognition of an embossed character sequence through convolution down-sampling, three-stage feature extraction and linear layer prediction. According to the technical scheme, high-precision and high-efficiency identification of the metal imprinting characters is realized by optimizing image acquisition, fusing a strategy and a deep learning model.

The invention relates to the technical field of bill text recognition, and discloses a bill text recognition system and method based on deep learning. The method comprises the following steps: acquiring target bill original image data containing a multi-channel pixel matrix and spatial resolution information; based on a matching result of the bill edge features and a preset template, geometric distortion correction is carried out on the original image, and a corrected bill image is generated; inputting the corrected image into a pre-training text region detection network to obtain positioning information containing text line boundary coordinates and region confidence; text line image blocks are extracted according to the positioning information, character segmentation preprocessing is executed, and a character-level image sequence is generated; calling a deep character recognition model to classify the sequence character by character, and generating an initial text recognition result; semantic verification and error correction are performed on the initial result based on the bill type knowledge base, final structured text data are generated, and the processing requirements of bills of different types and qualities can be met.

According to the title level identification large model training method, the title identification method and system and the program product provided by the invention, the first title information covering all texts of the original file is constructed through the multi-modal semantic model, and the limitation of plain text identification is made up in combination with the element matching page picture; second title information with semantic and visual features is generated through multi-modal fusion, so that the title judgment accuracy is improved; in the training process, effective title objects are screened in combination with original title objects to optimize pre-training data, and a large model which is high in precision and adapts to complex scenes is cultivated; in the identification process, the title information to be identified is constructed based on the valid title object. And performing dynamic branch processing according to a calling condition, if not, directly outputting an answer, if yes, generating accurate final title information by means of a trained large model, and finally constructing and outputting the answer, thereby realizing training and identification full-link coordination, considering complex scene adaptability and efficient and accurate identification, and comprehensively improving the structuralization and practicability of document title identification.

Apparatuses, systems, and techniques to identify text in a document obtained for text recognition are described. In at least one embodiment, text is identified based, at least in part, on comparing labeled text identified using one or more second neural networks with unlabeled text identified using one or more first neural networks.

The invention relates to the technical field of artificial intelligence, and discloses a paper text processing method based on intelligent glasses. The method comprises the steps that intelligent glasses are awakened through an awakening word, a user can conduct image recognition through a voice instruction, and whether a textbox is complete or not is judged. If the textbox is complete, character recognition is carried out; if not, the intelligent glasses help the user to adjust through voice guidance until the textbox is completely displayed. If the recognized characters are not the language set by the user, the intelligent glasses automatically translate the recognized characters into the user language, and voice synthesis is carried out to generate a voice file. And after generation, inquiring the user whether to play the voice, and if not, encrypting and storing the voice file. The problems that current text detection precision is not high, the reading process is not natural, user control is complex, and a feedback mechanism is single are solved.

The invention provides a medicine outer package label information intelligent checking method, system and device and a medium, and belongs to the technical field of information. The method comprises the following steps: collecting an original image of a medicine outer package through a handheld PDA terminal; performing label area positioning and geometric correction on the original image to obtain a regular label picture; inputting the regular label picture into a character recognition model, and outputting label text information; processing the label text information based on a large language model, and extracting key field information; verifying the key field information of each batch of medicines to generate standard label field information, constructing a template library containing the standard label field information of each batch of medicines, and binding and storing the standard label field information and the unique index code; and reading to-be-checked drug outer package label information, obtaining to-be-checked index codes and key fields in the to-be-checked drug outer package label information, calling corresponding standard label field information in the template library, comparing the standard label field information with the to-be-checked key fields, and outputting a comparison result.

The invention discloses a character recognition method and device for a handwritten image, terminal equipment and a storage medium, and the method comprises the steps: obtaining the handwritten image, and converting the handwritten image into a single-channel gray-scale map; wherein the single-channel grey-scale map is an image with a white background and black characters; performing reverse binarization processing on the single-channel grey-scale map to generate a reverse binarization image with black-matrix white characters; all contours in the reverse binarization image are recognized, and the contour with the largest area serves as a character core area; calculating an amplification coefficient according to the width and height of the reverse binarization image; performing boundary expansion on the character core region according to the amplification coefficient, and performing cutting based on the expanded boundary to obtain a cut image; and performing character recognition according to the clipped image. According to the invention, the character recognition accuracy can be improved.

The invention relates to the technical field of data processing, in particular to a data auditing method, system and device based on AI model assistance and a medium. The method comprises the following steps: firstly, performing text extraction on to-be-audited data through an OCR (Optical Character Recognition) module, then performing semantic analysis by utilizing an NLP (Network Length Polymorphism) module, and converting an unstructured text into a structured feature; processing the structured features based on a preset knowledge base to generate auditing feature vectors containing business semantics; then inputting the feature vector into a machine learning model composed of an anomaly detection model and a risk classification model, and generating a preliminary auditing conclusion from three dimensions of a basic rule, a business rule and a risk rule; and finally, carrying out confidence evaluation on the conclusion of the three dimensions, and triggering a corresponding processing flow according to an evaluation result. Through cooperation of a plurality of AI models, conversion from surface text recognition to deep semantic understanding is realized, a comprehensive evaluation mechanism based on multiple dimensions is established, and the accuracy and efficiency of auditing are improved.

The invention discloses a high-fidelity visual character image generation method, system and device and a storage medium, which are corresponding schemes, in the scheme, spatial domain and spectral domain feature mutual learning is realized based on a double-domain font coding mechanism, and the problem of character structure disorder is solved; based on a frequency domain perception refining mechanism, a frequency domain signal is dynamically modulated to improve stroke detail precision, collaborative optimization of character details and background quality is realized through an instance-level scaling coefficient, and the bottlenecks of character distortion and background instability in the prior art are broken through. Besides, the method supports multi-language accurate generation, can be widely applied to commercial scenes such as advertisement design, product packaging, film and television posters, teaching materials and the like, and can greatly improve creation efficiency and reduce manual correction cost; and meanwhile, the feasibility of cooperatively improving the multi-modal generation efficiency by the cross-domain features is verified, a high-quality weak supervision training sample in a scarce scene can be provided for a character recognition model, the blank of related data is effectively filled up, and the method has technical advancement and market landing value.

The invention discloses an end-to-end multi-task learning method and system based on Swin Transform, and belongs to the technical field of artificial intelligence and computer vision crossing, and the method comprises the steps: collecting low-quality paper document images which have physical degradation characteristics and cause substantial obstacles to information recognition, carrying out the corresponding text labeling of each image, and constructing a target data set; based on the target data set, adopting a backbone network capable of extracting multi-scale hierarchical features as a shared image encoder, and taking an image enhancement task and a text recognition task as two parallel downstream branches to construct an end-to-end multi-task learning network architecture; based on the output of the image enhancement task and the text recognition task, image enhancement loss and text recognition loss are calculated respectively, a joint loss function is constructed, and multi-task collaborative optimization is realized through end-to-end training; the robust recognition capability of fuzzy, continuous and low-quality handwritten characters is improved, and a recognition-friendly high-quality image is generated.

The invention discloses a handwritten text recognition method based on multi-stage enhancement. The handwritten text recognition method comprises the following steps: acquiring a handwritten text image; constructing a hierarchical dynamic multi-scale CNN backbone network to obtain a visual feature sequence; inputting the visual feature sequence into a time sequence multi-scale module to obtain a local enhanced feature sequence; performing global modeling on the local enhanced feature sequence by using a Transform encoder to obtain a global visual feature sequence; enhancing the global visual features to obtain a time sequence context feature sequence; dynamic weighted fusion is carried out on the global visual features and the time sequence context features through a gating fusion module; and sending the fused features into a linear classifier and a CTC decoder to obtain an identification result. According to the method, intelligent arbitration of global and time sequence features is realized through new technology application of a hierarchical multi-scale CNN trunk, time sequence context enhancement and a gating fusion mechanism, and the recognition accuracy and robustness are remarkably improved.

The embodiment of the invention discloses a drawing analysis method and device for an electric power tower, equipment and a storage medium, and the method comprises the steps: obtaining an original image of a two-dimensional engineering drawing of the tower, and carrying out the preprocessing of the original image to generate a standardized image; performing layout analysis on the standardized image, and identifying a plurality of functional areas; for a graphic region in the functional region, extracting graphic elements of the graphic region, and analyzing a topological relation between the graphic elements to construct a geometric model representing a tower structure; performing text recognition and semantic analysis on the text contained in the functional region to extract structured semantic attribute information; based on the geometric model and the semantic attribute information, establishing a corresponding relationship between the graphic elements and the semantic attribute information to generate structured data containing geometric and semantic information; and outputting the structured data as a data model in a preset format. In this way, high-precision information extraction is achieved, and the automatic analysis capacity and digital modeling efficiency of tower drawings are remarkably improved.

The invention provides a lightweight multi-modal fusion document information structured extraction method and system, and relates to the technical field of data processing, and the method comprises the steps: obtaining a document image; preprocessing the document image to obtain an optimized image; through the MobileNetV3, text features of the optimized image are extracted; performing multi-scale feature fusion on the multiple text features to obtain a fused feature map; through a double-branch Tokenized MLP module, extracting a context relation feature between a position feature of the fusion feature map and a spatial position; detecting a text region of the optimized image according to the position features of the fused feature map and the context relationship features between the spatial positions; performing text recognition on the text region of the optimized image through LPRNet; through SLANetplus, a table area of the optimized image is detected; and extracting structured document information by combining the text region and the table region through a multi-modal encoder.

The invention discloses a handwritten medical record OCR recognition system and method based on deep learning. The system comprises an image acquisition and preprocessing module, a text detection and region positioning module, a handwritten character recognition module, a semantic correction module and a result storage and export module. The image acquisition and preprocessing module is used for receiving the original medical record image and performing image quality improvement processing to obtain optimized image data; the text detection and region positioning module is used for detecting a text region from the optimized image data and positioning the position information of the text region to obtain text region data; the handwritten character recognition module is used for performing recognition processing on a character sequence in the text region data to obtain an initial text recognition result; the semantic correction module is used for processing the initial text recognition result based on medical priori knowledge to obtain a corrected text recognition result; and the result storage and export module is used for outputting the corrected text recognition result to a target system in a structured format.

The invention relates to the technical field of insurance business processing, and particularly discloses an automatic car insurance business handling method based on an AI customer service robot. The method comprises the following steps: receiving and standardizing business related information through an information acquisition unit, screening effective information through an information quality evaluation model, extracting key fields by means of OCR text recognition, and generating structured insurance data in combination with multi-modal consistency comparison; a vehicle information system is called to verify data integrity and accuracy in real time and automatically complement and correct errors, and compliance verification is completed through an underwriting rule engine according to a preset rule of an insurance company to generate standardized insurance data; a rate engine is called based on the data to generate a quotation scheme, the quotation scheme is submitted to an underwriting system to generate an underwriting conclusion after being confirmed by the user, a payment two-dimensional code is automatically generated after underwriting is passed, an electronic insurance policy is pushed when the user pays successfully, and meanwhile, whole-process data is stored persistently, and a service ledger is generated.

Certain aspects of the disclosure provide a method for extracting attributes from documents with varying formats, layouts and complexities. The method displays a user interface (UI) that enables a user to obtain an unstructured document from a knowledge base. The method converts the unstructured document into a text document using a text recognition. The method obtains, as output from a large language model (LLM), an extracted page attribute from the text document. The extracted page attribute contains a first type of information recorded in text on a single page of the text document. The extracted document attribute contains a second type of information recorded in text on more than one page of the text document. The method obtains, as output from the LLM, an extracted document attribute from the text document. The extracted page attribute and the extracted document attribute are displayed in the UI.

The invention discloses a power distribution cabinet instrument intelligent identification method and system, and relates to the technical field of computer vision, and the method specifically comprises the steps: 1, generating a preprocessing image based on a to-be-identified image; step 2, detecting whether a pointer instrument exists in the preprocessed image, and if yes, executing step 3; if not, executing the step 4; 3, identifying a pointer instrument area based on the preprocessed data, extracting an image of the pointer instrument area, and performing angle identification based on the image of the pointer instrument area to obtain an initial angle; performing text recognition based on the pointer instrument area image to obtain a digital range and a unit; obtaining a physical quantity reading based on the digital measuring range, and then skipping to execute the step 5; 4, inputting the preprocessed image into the trained instrument detection auxiliary model, and outputting the category, position, physical quantity reading or state data of the instrument; and 5, mapping the related data to a visual interface. According to the invention, intelligent identification and real-time monitoring of various instruments can be realized.

The invention provides a video description information generation method and device, electronic equipment and a storage medium, and the method comprises the steps: obtaining a target video of to-be-generated video description information; segmenting the target video to obtain a plurality of video clips; extracting a key frame set corresponding to each video clip from each video clip; performing character recognition on the key frame set to obtain text information corresponding to each video clip; inputting the key frame set corresponding to each video clip and the text information into a preset target model to obtain clip description information corresponding to each video clip; and generating video description information of the target video based on all the fragment description information. By means of the information processing method and device, the technical problems of key information loss and irrelevant information redundancy in the related technology are solved.

The invention discloses a building construction drawing key information intelligent extraction method and system. The system comprises a text recognition and error correction module, a CAD structure analysis module, a multi-modal information fusion and understanding module and a conflict resolution and structured output module. According to the method, text recognition robustness is improved by fusing an OCR technology and a large language model error correction mechanism, a spatial relation graph is constructed in combination with CAD vector analysis, building field knowledge is generated and injected through retrieval enhancement, and finally structured information is output through multi-modal collaboration and conflict arbitration. According to the method, the problems of high text misrecognition rate, incomplete component information extraction and weak semantic relationship reasoning in the prior art are solved, the accuracy and integrity of key information extraction are remarkably improved, and digital and efficient processing of architectural drawings is supported.

The invention discloses a scoring method and device of a recognition engine, a storage medium and an electronic device, and relates to the technical field of smart home, and the method comprises the steps: inputting a target audio into a trained voice large model to obtain a first recognition text, and obtaining a second recognition text output by processing the target audio through a third-party text recognition engine; calculating a plurality of index parameters between a first phoneme sequence corresponding to the first recognition text and a second phoneme sequence corresponding to the second recognition text, the plurality of index parameters comprise at least one of the following parameters: similarity between different phoneme sequences, common phoneme proportion between different phoneme sequences, phoneme error rate between different phoneme sequences, and phoneme length matching degree between different phoneme sequences; summarizing the index parameters according to preset parameter weights to obtain a target score of the third-party text recognition engine for recognizing the target audio, the target score being used for indicating the quality of the recognition text output by the third-party text recognition engine.

The application provides a character recognition method and system. The character recognition method comprises the following steps: determining a picture to be recognized; previewing the outline of the picture to be recognized, and setting the position of each vertex on the outline; generating a closed area according to the position of each vertex; judging whether the generated closed area covers the area to be recognized; if the closed area does not effectively cover the area to be recognized, resetting the vertex on the edge of the area not completely covered until an effective area that effectively covers the area to be recognized is generated; cutting the content outside the effective area and recognizing the character in the effective area. The application solves the problem of quickly and accurately positioning the recognition area by calculating the two points or multiple points to form a closed area in the photographed picture, marking and recognizing the selected character content in the closed area, and improving the recognition accuracy.

The application provides a table recognition method, device and equipment and a storage medium. The method comprises: acquiring a track point sequence collected when a user writes on a screen, and preprocessing the track point sequence; determining table track points and text track points from the preprocessed track point sequence; rendering the table track points into an image to obtain a pure table image, and rendering the text track points into an image to obtain a text image; performing table recognition on the pure table image to obtain a table recognition result, and performing text recognition on the text image to obtain a text recognition result; matching the table recognition result and the text recognition result to obtain a matching result; and generating a table containing text content based on recognition information containing the table recognition result, the text recognition result and the matching result. The table recognition method provided by the application can restore a table containing text content written by a user.

The invention provides a text recognition method and device, equipment and a storage medium, which are used for solving the problem of low flexibility of text recognition. The method at least comprises the following steps: acquiring a text area where a plurality of characters contained in a to-be-recognized image are located; performing character feature extraction on the text region to obtain a character feature sequence; wherein the character feature sequence comprises respective character features of a plurality of characters; performing language prediction on each character feature contained in the character feature sequence to obtain respective language prediction categories of the plurality of characters; and when a target language category belonging to the small language category exists in the obtained prediction categories of the languages, performing text recognition on the text region by adopting a text recognition strategy matched with the target language category to obtain text content formed by a plurality of characters. Based on respective language prediction categories of a plurality of characters contained in the text region, the process of automatically selecting the text recognition strategy is realized, and the flexibility of text recognition is improved.