264 results about "Computer vision algorithms" patented technology

The embodiment of the invention provides a document generation method and device based on artificial intelligence, and the method comprises the steps: extracting multi-modal data features of a historical multi-modal file through natural language processing and a computer vision algorithm, and constructing a corresponding knowledge graph according to the multi-modal data features, converting the multi-modal data subjected to feature extraction into high-dimensional vectors by using a vectorization algorithm, and storing the high-dimensional vectors to form a vector library; splitting a current document generation task into a plurality of sub-tasks by adopting a task planning algorithm, and executing the sub-tasks at the same time to obtain a first document; performing iterative optimization of document content according to the document first draft, the cue word and the language model to obtain an optimized first draft, acquiring related information from the vector library by using an attention mechanism-based retrieval algorithm, and fusing the related information and the optimized first draft through a text generation algorithm to obtain a final document; according to the invention, the accuracy and intelligence in the document generation process can be effectively improved.

The invention discloses an electric power operation risk early warning method and system based on knowledge enhancement and multi-modal fusion. The method comprises the steps that video monitoring data, sensor monitoring data and service system data are collected in real time through multi-source sensing equipment deployed on an electric power operation site; the method comprises the following steps of: extracting entities and relationships from unstructured texts such as regulation documents and job logs by utilizing a natural language processing technology based on deep learning, extracting behavior characteristics from video streams by adopting a computer vision algorithm, and constructing an electric power security knowledge graph with dynamic updating capability; designing a multi-modal feature fusion algorithm based on an attention mechanism, and effectively integrating visual features, text features and sensor data; a graph neural network is adopted to train a dynamic risk prediction model to carry out risk prediction, intelligent research and judgment of electric power operation risks are realized, accurate management and control of the risks are realized through a grading early warning mechanism, and closed-loop management from risk perception to early warning treatment is formed.

The invention discloses a road surface and dirty state detection method based on a laser radar, and the method comprises the following steps: scanning a road surface through the laser radar, obtaining a visual image and environment data of the road surface through a sensor, carrying out the fusion processing of the point cloud data of the laser radar and the data of the sensor, and generating a three-dimensional data set; (2) filtering, denoising, aligning and normalizing the point cloud data by using a computer vision technology; performing feature extraction on the data by using a computer vision algorithm; (3) according to the identified surface features, identifying an abnormal region by using a deep learning algorithm to judge whether the road surface is dirty or not; (4) if smudginess exists, further analyzing the type and the distribution condition of the smudginess; and (5) sending a detection result to a central control system through a wireless communication module. The geometrical morphology of the pavement can be accurately measured, and meanwhile the cleanliness of the pavement is evaluated.

The invention belongs to the technical field of environmental monitoring, and particularly relates to a super-resolution fluorescence imaging analysis method based on hypersensitive fluorescent quantum dots. The quantum dots have high water solubility and stability, and can emit narrow-bandwidth fluorescence due to unique optical properties of the quantum dots. According to the invention, a quantum dot luminescent material is adopted to construct an ultra-high resolution microscope system which can be used for observing composite behaviors between micropollutants and quantum dots on a microscopic scale, and a three-dimensional fluorescence microspectrometer and a computer vision algorithm are integrated; the whole reaction process is deduced according to the evolution of the three-dimensional fluorescence spectrum between the micro-pollutants and the quantum dots in different space (concentration), time (reaction process) and chemical dimensions, and a multi-dimensional fingerprint spectrum is formed according to the whole reaction process and is used for identifying and tracing the micro-pollutants. The method has the advantages of being high in detection speed, high in fingerprint performance and less in interference of background factors.

The invention relates to the field of photovoltaic module detection, and discloses a module hot spot defect infrared and visible light fusion detection method, which comprises the following steps: multi-mode acquisition: multi-mode data of a photovoltaic module are synchronously acquired, and the multi-mode data comprise infrared temperature original data, visible light polarization images and environmental parameters; data preprocessing: preprocessing the multi-modal data based on a radiation transmission model and a polarization optical principle to obtain preprocessed data including an infrared temperature field and a visible light image; and dynamic feature extraction: extracting a space-time gradient tensor and a Riemannian curvature tensor of the infrared temperature field from the preprocessed data based on differential geometry and a computer vision algorithm. By synchronously acquiring multi-modal data, based on a radiation transmission model and a polarization optical principle, interference of atmospheric attenuation and environment temperature and humidity on infrared temperature measurement and influence of uneven illumination and fog effect on a visible light image are eliminated, and dynamic characteristics of an infrared temperature field and structural characteristics of visible light are mined.

The invention relates to the technical field of concrete construction monitoring, and discloses a non-embedded concrete workability detection method based on deep learning, which comprises the following steps: shooting an image of concrete flowing in a mixer truck chute by using an image acquisition device, analyzing image data, and determining the workability of the concrete in combination with concrete mix proportion data, hydration reaction time and mixer truck chute angle. Therefore, the viscosity value and the slump of the concrete are output. According to the method, a computer vision algorithm, deep learning and machine learning are combined to finish capturing the characteristics of concrete flowing out of the mixer truck step by step, and machine learning is performed based on all captured characteristic values so as to output the concrete slump. An algorithm involved in the scheme supports an edge computing device carrying a network communication module to complete data acquisition and analysis, the communication module is utilized to transmit the acquired and analyzed data to a cloud server, and a commercial mixing station is accurately assisted to make a concrete mix proportion adjustment decision according to different engineering actual conditions.

The invention discloses a dance generation method and system based on a deep learning model, and belongs to the field of dance choreography, and the transmission method comprises the steps: collecting a dance motion video of a dancer, extracting skeleton data of each frame through a computer vision algorithm, and converting the skeleton data into a key point sequence; a convolutional neural network is used to extract spatial features of the dance movement video, a time sequence feature of the dance movement video is extracted through a graph convolutional network, the time sequence and rhythm of the movement are captured, and time coding is performed on each movement; constructing a multi-modal deep learning model, and training the model to generate a corresponding dance action sequence according to the input dance style and music rhythm; a style control unit is added, a dance style is converted into a style vector understood by a model through a style migration network, and music and dance actions are synchronously performed by using a music rhythm prediction model; and generating a new dance action through the deep learning model. According to the method, the dynamic change and the structural relationship of the dance movement are accurately captured, and the generated movement is smoother.

The invention discloses a multi-sensor fusion unmanned vehicle detection method suitable for a complex environment. The method comprises the following steps: obtaining road surface data through a laser radar and an infrared thermal imaging camera; detecting rain and snow weather conditions in real time, and calculating weather environment factors based on rainfall data and snowfall data; dynamically adjusting data weight coefficients of the laser radar and the infrared thermal imaging camera based on the weather environment factors; using a computer vision algorithm to identify an obstacle area, and extracting the distance and height of the obstacle according to the corresponding data weight coefficient; and the obstacle type is analyzed, the minimum safe distance or the minimum torque between the unmanned vehicle and the obstacle is calculated, and a driving decision is provided for the unmanned vehicle. According to the invention, the self-adaptive capability and safety of the unmanned vehicle in a complex environment are improved, the accident risk caused by sudden obstacles is effectively reduced through an intelligent early warning and intervention mechanism, and a solid safety foundation is laid for the wide application of an unmanned driving technology.

A method for reducing the risk of unexpected or unintended actions during a Remote Desktop Protocol (RDP) or SSH (Secure Shell) session by automatically detecting anomalies using vaulted credentials. The method includes initiating a RDP / SSH session on a server and taking screenshots of the session. The screenshots are sent to a system that uses a computer vision algorithm to transcribe the screen information into a text-based transcription of the session. The text-based transcription of the session recording is then provided to a pre-trained LLM which is prompted to look for anomalies using in-context learning. The detected anomalies are presented to a system administrator or sent as alerts to the system administrator.

The invention provides an indoor human body positioning and track generation method and device, equipment and a medium, and the method comprises the steps: obtaining target Bluetooth data and position data with timestamps, achieving the data fusion through a data fusion algorithm, and obtaining a first moving track point set; based on a preset data acquisition period and timestamp information of each track point of the first moving track point set, when the track points are missing, detecting by using a computer vision algorithm and determining a corresponding target by using a tracking detection algorithm based on video data acquired by monitoring equipment, and obtaining an image sequence containing the target; determining an actual physical coordinate of a target in each image based on camera calibration and perspective transformation, obtaining an image positioning result and a second moving track point set, and aligning with a timestamp of the first moving track point set and realizing data fusion to obtain a third moving track point; and according to the map information, matching each track point of the third moving track point set to a grid map to form a target track map.

The invention relates to the technical field of visual identification, in particular to a warehouse control management system and method based on computer vision and mobile robot technologies, and the method comprises the steps: training a target detection model for identifying a line bucket based on a computer vision algorithm, and deploying the target detection model to an AI control box; a camera, an AI control box, a warehouse management system, a robot management system and a robot are integrated through communication; the AI control box obtains a storage location image captured by a camera, inputs the storage location image into a target detection model, and identifies the line barrel and the state thereof to obtain identification data; and after the warehouse management system stores and updates the data, tasks are sent to the robot management system according to requirements, and the robot management system schedules one or more robots to carry the wire buckets. According to the method, automatic identification of the wire barrel and the state of the wire barrel is achieved, the robot can be accurately dispatched to move and place the wire barrel, automatic management of the wire barrel is achieved, workshop management efficiency is greatly improved, and manual errors and safety risks are reduced.

The invention relates to the technical field of geotechnical engineering, and discloses a wall-attaching type microwave rock crushing method and device suitable for shield tunneling machine loading, and the method comprises the steps: obtaining an image of a surrounding rock tunnel face in real time through image collection equipment, and carrying out the preprocessing; analyzing the preprocessed image of the tunnel face of the surrounding rock, and analyzing the surface state of the surrounding rock by using a computer vision algorithm to obtain the current hardness of the surrounding rock; judging whether a microwave system needs to be started according to the current hardness of the surrounding rock; if the current hardness of the surrounding rock is smaller than the preset threshold value, the pushing system and the microwave system are not started; if the current hardness of the surrounding rock is larger than or equal to the preset threshold value, the control module starts the pushing system to push the microwave system to make contact with the surrounding rock tunnel face; meanwhile, a microwave system is started to heat the surrounding rock; in the working process of the microwave system, the images of the tunnel face of the surrounding rock are continuously obtained and analyzed, the hardness of the surrounding rock is judged according to the image analysis result, and the precision and adaptability of the microwave rock crushing technology can be improved.

The invention relates to the field of computer vision algorithms, in particular to a defect detection data screening method based on a Pontryagin maximum principle, which comprises the following steps of: training a training data set and testing an evaluation data set to obtain a model reference index; based on a preset proxy data set, respectively calculating definition, labeling integrity and distribution deviation degree indexes to obtain an initial quality weight vector; iteratively updating the basic model parameters and reversely iteratively updating the target vector to obtain a sample quality score; a training scoring device scores and sorts full samples of the training data set; and dividing the sorted full samples into a plurality of candidate screening intervals, and screening high-quality data to train a final model. According to the method, the entropy weight method is adopted to weight the three dimensions to obtain the initial quality weight, so that the initial weight of the sample can reflect the own basic quality difference, the problem that the traditional uniform weight ignores the sample quality difference is avoided, and the accuracy of the sample quality score is further improved.

The invention relates to a two-way early warning method and system for work depot operation, processing equipment and a storage medium, and the method comprises the steps: arranging a vibration sensor on each track, and enabling the vibration sensors of all tracks in a certain region to form a corresponding vibration sensor array; the signals collected by the vibration sensors are processed, the vibration signals, collected by the vibration sensors, of the train passing through the track are extracted, and the track of the coming train is determined; a computer vision algorithm is adopted, based on the video monitoring data of the track of the coming train, the behavior state of a working worker in operation in the track is detected, and the behavior state is sent to a train center controller of the coming train; the train center controller of the coming train judges whether the working personnel in operation have potential threats or not according to the detected behavior state of the working personnel in operation in the track, if yes, early warning is carried out, and the method can be widely applied to the field of operation scenes of the railway service depot.

The invention relates to the technical field of three-dimensional measurement, in particular to a linear structured light bearing bush and gear three-dimensional measurement method and system, and the method comprises the following steps: 1, data collection: employing a CCD camera arranged on a measurement platform to shoot laser images containing laser line patterns from different angles; step 2, image processing: extracting feature points in the laser image by using a computer vision algorithm; and 3, three-dimensional reconstruction: based on the known CCD camera parameters and the position change of the feature points on the to-be-measured object, obtaining three-dimensional data of the to-be-measured object by using a space coordinate conversion technology. And step 4, error correction: carrying out error analysis on the obtained three-dimensional data. And step 5, result output: outputting final three-dimensional measurement data of the to-be-measured object based on the corrected result. Through the automatic and intelligent measurement means, the measurement period can be further shortened, and the production efficiency is improved.

The invention relates to the technical field of medical information systems, in particular to an intelligent classification treatment system and method for medical waste in an operating room, and the system comprises a waste recognition module which is used for recognizing the medical waste through a computer vision algorithm; determining a waste category based on the recognition result; the electronic guiding module is in communication connection with the waste identification module and is used for receiving the waste category information sent by the waste identification module; based on the waste category information, generating an electronic guide for guiding the medical staff to store the waste; the waste weighing and recording module is in communication connection with the electronic guiding module and is used for calculating and recording the weight and the capacity of the waste; waste weight and volume data are generated; the RFID reading module is in communication connection with the waste weighing and recording module and is used for automatically reading RFID electronic tag information; the source tracing of the waste is realized based on the RFID electronic tag information; the system can automatically complete the whole process from waste identification, classification guidance, weight recording to information tracking.

This disclosure describes an interactive augmented reality system for improving surgeon's view and context awareness during laparoscopic and video assisted surgeries. Instead of purely relying on computer vision algorithms for image registration between pre-operation (or intra-operation) images / models and later intra-operation scope images, the system can implement an interactive mechanism where surgeons may provide supervised information in initial calibration phase of the augmented reality function, thus achieving high accuracy in image registration. Besides the initialization phase before operation starts, interaction between surgeon and the system can also happens during the surgery. Specifically, patient tissue might move or deform during surgery, caused by for example cutting. The augmented reality system can re-calibrate during surgery when image registration accuracy deteriorates, by seeking additional supervised labeling from surgeons. The augmented reality system can improve surgeon's view during surgery, by utilizing surgeon's guidance sporadically to achieve high image registration accuracy.

Described here are systems and methods that utilize visual imagery and an optical flow-based computer vision algorithm to measure river velocity in streams or other flowing bodies of water. The systems and methods described in the present disclosure overcome the barriers of conventional flow measurement techniques by providing a fast, non-intrusive, remote method to measure peak flows.

The invention relates to the technical field of invisible orthodontic appliance wearing monitoring and adjusting systems, in particular to an invisible orthodontic appliance wearing monitoring and adjusting system and method based on computer vision, and the system comprises a cloud storage module which is used for storing initial oral data and re-consultation oral data of a patient; the cloud processing module is used for analyzing the oral cavity picture uploaded by the patient and executing analysis software; the client module is used for managing an electronic medical record of a patient, providing treatment information, obtaining an analysis result and generating personalized feedback; according to the invention, a computer vision algorithm is combined with oral medical knowledge, so that not only can the position change of teeth be accurately identified and tracked, but also the judgment ability of the device can be continuously optimized through a deep learning algorithm. Meanwhile, the cloud computing architecture adopted by the invention enables the system to process mass data, and powerful data support is provided for formulating a personalized treatment scheme.

The invention discloses a construction personnel state monitoring and early warning system and method based on attitude tracking, and relates to the technical field of monitoring and early warning, the system comprises a data acquisition module, an attitude tracking module, a behavior analysis module and an alarm module which are connected in sequence; the data acquisition module is used for capturing a real-time video stream of a constructor; the attitude tracking module is used for extracting skeleton key points of the constructor from each frame of image in the real-time video stream by using a computer vision algorithm to obtain coordinates of the skeleton key points of the constructor; a posture recognition model is deployed in the behavior analysis module, coordinates of the skeleton key points are input into the posture recognition model, actions of construction personnel are determined, unsafe behavior detection is carried out, and an unsafe behavior recognition result is obtained; and the alarm module is used for triggering an alarm when the unsafe behavior identification result is an unsafe behavior. According to the invention, the real-time performance, precision and automation degree of monitoring and safety alarm of constructors are improved.

The invention discloses a face quality label generation method based on cooperation of a multi-modal large model and a computer vision algorithm, and relates to the technical field of artificial intelligence and computer vision crossing, in particular to a face quality label generation method based on collaborative decision making of the multi-modal large model and the computer vision algorithm. The method is especially suitable for a high-security-level identity verification system, an automatic identification photo auditing platform, a large-scale face registration pre-screening process, a living body detection front level and other scenes, and is used for carrying out fine-grained, interpretable and high-robustness quality evaluation and defect positioning on an input face image. The core problems of weak semantic understanding ability, unstable discrimination precision, rigid fusion strategy, poor scene adaptability and the like in the prior art are solved by constructing a four-level technical architecture of dimension decoupling, double-channel parallelism, dynamic confidence fusion and scene adaptive evaluation.

The invention discloses an intelligent monitoring and data analysis management platform for a municipal pipe network. The system comprises a plurality of monitoring areas, an environment sensor group, an environment data processing unit, a corrosion risk assessment module, an intelligent inspection unit, a data transmission unit and a cloud platform. The environment sensor group collects environment data in a monitoring area in real time, and the environment data processing unit analyzes the data in real time through an edge computing technology and generates historical data. And the corrosion risk assessment module assesses the corrosion risk according to the real-time data and the historical data, determines a high-risk area and generates a detection task. The intelligent inspection unit automatically navigates to a target area for light spot projection detection, and the deformation analysis module analyzes image data through a computer vision algorithm and identifies a corrosion area and information thereof. And the data transmission unit uploads the detection information to the cloud platform. Through automatic inspection and light spot projection, the corrosion detection efficiency and precision are improved, and safe operation of pipe gallery equipment is ensured.

A system for monitoring shopping baskets (e.g., baskets on human-propelled carts, motorized carts, or hand-carried baskets) can include a computer vision unit that can image a surveillance region (e.g., an exit to a store), determine whether a basket is empty or loaded with merchandise, and assess a potential for theft of the merchandise. The computer vision unit can include a camera and an image processor programmed to execute a computer vision algorithm to identify shopping baskets and determine a load status of the basket. The computer vision algorithm can comprise a neural network. The system can identify an at least partially loaded shopping basket that is exiting the store, without indicia of having paid for the merchandise, and execute an anti-theft action, e.g., actuating an alarm, notifying store personnel, activating a store surveillance system, activating an anti-theft device associated with the basket (e.g., a locking shopping cart wheel), etc.

The invention discloses a metal internal characteristic selective extraction method, which comprises the following steps of: (1) scanning a block-shaped metal sample by adopting high-resolution X-ray CT (Computed Tomography), reconstructing characteristics including defect distribution, microstructure, holes and cracks in metal through a three-dimensional reconstruction algorithm, and generating a high-precision three-dimensional model; (2) selecting a required area, calculating voxel resolution according to the distance SDD between a radiation source and a detector and the distance SOD between a sample and the radiation source, displaying a reconstructed three-dimensional image in three-dimensional visual analysis software, and performing threshold segmentation through a computer vision algorithm, and automatically identifying and extracting required features for observation; the whole sample is placed in a square frame in an externally tangent form, and a Cartesian rectangular coordinate system is established. According to the invention, after the internal structure is observed by using the X-ray microscopic imaging technology, an interested area is selected for extraction by using linear cutting according to the three-dimensional structure information, and the internal defect of the sample can be efficiently and accurately positioned and extracted.

The invention provides a method for detecting the area of a defrosting and demisting area of an automobile windscreen, and the method comprises the following steps: S1, carrying out the automatic collection of the defrosting and demisting area of the windscreen through a fixed camera, obtaining an image of the windscreen in front of an automobile, and carrying out the preprocessing of the collected image; s2, performing contour recognition on the preprocessed image by using a computer vision algorithm to obtain a contour image of a defrosting and demisting area of the windshield glass; s3, performing camera calibration on the identified contour image of the defrosting and demisting area of the windshield glass by using a Tsai two-step calibration method, including preliminarily estimating internal parameters and external parameters of a camera, establishing a pinhole camera model according to the preliminarily estimated internal parameters and external parameters of the camera, and correcting radial distortion in the imaging process of the camera; and S4, carrying out area calculation according to the calibrated and corrected contour image of the defrosting and demisting area of the windshield glass.

The invention belongs to the technical field of visual inspection, and particularly relates to an industrial defective product detection method and defective product detection device.The industrial defective product detection method comprises the steps of S1, image collection, S2, image preprocessing, after image collection, image preprocessing to improve image quality, S3, key features in images are extracted through a computer vision algorithm, and S4, defect recognition and classification. S5, decision making and judgment: classifying the products through a processing system according to the identified defects or abnormal characteristics, and judging whether the products are defective products, S6, outputting results, judging whether the products are qualified or unqualified, and finally outputting qualified or unqualified judgment results, and the system is connected with a production line, automatically sorts the defective products or prompts an operator to carry out manual intervention. According to an image sensor, a light source, image processing software and an analysis algorithm, the product appearance and defects are recognized and classified, and through the image obtaining and processing technology, whether the product meets the design standard or not is detected in real time, and the problems of surface defects, shape inconformity, size out-of-tolerance and the like are recognized.

The invention relates to the technical field of domestic pig selling systems, and provides a remote domestic pig selling system and method.The remote domestic pig selling system comprises the steps that a target pig house is divided into a plurality of breeding areas and at least one functional area, and based on interaction information input actions of buyers, the pig output action that pigs of a target batch are driven to the functional areas from the breeding areas is executed; the image acquisition device arranged in the functional area is used for acquiring pig images, and pig information is synchronized to remote buyers in a video synchronization mode by using various computer vision algorithms, so that the buyers can know detailed conditions of each batch of pigs more clearly, the transparency of remote purchase is improved, and the communication efficiency is improved. Meanwhile, the robust correction problem of using the fisheye camera in different hog houses is considered, a self-supervision mode is innovatively used for correcting the camera image, and the interaction accuracy of hog selling information between the buyer and the seller is improved.

The invention discloses a calligraphy image recognition method and system based on a computer vision algorithm, and relates to the technical field of calligraphy image recognition, and the method comprises the steps: collecting dynamic trajectory data through a high-speed camera device and a pressure-sensitive pen, and generating a dynamic trajectory matrix; a multispectral imager is used for obtaining a material characteristic pattern of paper fiber deformation and ink mark penetration, and fusion data is generated through affine transformation alignment; inputting the fusion data into an LSTM-CNN network to extract spatio-temporal features, outputting a dynamic stroke segmentation map, and separating material codes, technique features and style latent variables from the fusion data through a three-channel decoupling network; carrying out preliminary classification, carrying out secondary classification on characters with similar shapes in combination with an attention mechanism, outputting a fine-grained classification result, carrying out weighted fusion on the classification result, carrying out style verification and track logicality, and outputting identification tags and style attributes; and misjudgment samples trigger incremental learning of the laminated classification model, and the weight of a secondary classifier is optimized.

The invention discloses an earthquake large model construction method, device and equipment based on a computer vision algorithm. The method comprises the following steps: acquiring label seismic data; extracting profile data in the tag seismic data to obtain multiple groups of profile data; randomly selecting at least one preset mode to mask the plurality of groups of profile data, and constructing a plurality of groups of mask profile samples; based on self-supervised learning, respectively inputting the plurality of groups of mask profile samples into a plurality of pre-constructed computer vision algorithm models for training to obtain a plurality of base models; and fusing the plurality of base models based on an attention mechanism, and constructing a large earthquake model. According to the method, features of different dimensions in seismic data are extracted by using different computer vision algorithms, and the features are fused based on an attention mechanism, so that the constructed large seismic model can obtain more comprehensive and accurate feature representation, and the large seismic model with high calculation precision and strong generalization ability is constructed. And the accuracy and timeliness of intelligent identification of the seismic section are improved.

The invention discloses a cognitive load assessment method based on fusion of behavior characteristics and heart rate variability in a classroom video. According to the method, face and behavior videos of students in a real classroom are collected, behavior characteristics such as eye movement tracks, sitting postures and facial expressions of the students are extracted through a computer vision algorithm, heart rate variability indexes are predicted in combination with a remote photoplethysmography technology, and a time sequence characteristic sequence is formed. And then, a time sequence deep learning model is adopted to carry out joint modeling on the multi-modal time sequence characteristics, and the cognitive load scale level is taken as a supervision signal to construct a classification model to realize cognitive load level prediction. The method has the advantages of non-contact, automation, high adaptability and the like, and can be applied to personalized teaching monitoring and intelligent teaching feedback.