165 results about "Vision algorithms" patented technology

The invention discloses a visual algorithm self-training method based on multi-agent collaborative optimization, and the method comprises the following steps: constructing a multi-agent system architecture which comprises a user interaction layer, an intelligent scheduling layer, an A2A protocol communication layer and a professional agent cluster layer; the user interaction layer analyzes a user task intention and generates an execution plan; the scheduling agent calls the professional agent to complete data processing, model construction, training, testing and deployment; a task process is coordinated through a standardized communication mechanism, and task execution is supported by combining an MCP tool set, a knowledge base module and a memory system; and when the task fails, automatically executing rescheduling operation, and finally outputting a self-training result. According to the method, the development efficiency, the self-adaptability and the intelligent level are remarkably improved, and the method is suitable for computer vision tasks such as industrial detection, intelligent security and protection and automatic driving.

The invention relates to the technical field of intelligent inspection of power equipment, and particularly discloses a power transmission line inspection system based on an autonomous correction long-endurance unmanned airship. The system aims to solve the problems of short endurance, small operation radius, single image angle, intermittent operation, leak detection, low precision, poor environmental adaptability and the like caused by dependence on manpower or GPS navigation in the conventional unmanned aerial vehicle inspection. The system comprises an energy supply module, a data acquisition module, an intelligent control module and a communication and data return module. Wherein the energy supply module is integrated with a solar energy collection, storage and intelligent scheduling unit to realize day-and-night continuous energy supply; the intelligent control module adopts an RT-DETR visual algorithm to identify a tower in real time, performs automatic deviation correction by fusing pose information, and accurately tracks a preset track; the data acquisition module supports multi-angle high-definition imaging, and the communication module guarantees real-time data return. And high-precision unmanned inspection with ultra-long endurance, full-automatic and wide-area continuous coverage is integrally realized.

The invention discloses a mechanical arm simulation system supporting industrial quality inspection, which relates to the technical field of industrial automation and comprises a multi-physics field simulation engine module, a digital twin interface module, a defect sample library module and a cooperative control module. According to the method, traditional entity mechanical arm debugging is replaced by rigid body dynamics simulation based on a physical engine, loss and production line shutdown caused by repeated trial and error of equipment are avoided, rapid construction of the digital twin environment of the production line is achieved by means of an efficient CAD model conversion process and a lightweight loading technology, and high-fidelity optical simulation is achieved through a parameterized defect model library and high-fidelity optical simulation. Various common industrial defects and complex optical effects are covered, the modeling bottleneck of a quality inspection scene is broken through, real-time closed-loop interaction of visual inspection and motion control is realized by using thread separation and an efficient communication technology, and the recognition robustness of a visual algorithm to tiny defects is guaranteed; the reliability of algorithm verification is improved through the quantitative output of algorithm performance indexes and the comprehensive test of extreme working conditions by the simulation system.

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.

The invention discloses an experiment teaching system based on augmented reality and machine vision, which relates to the field of laboratory teaching systems and comprises an augmented reality visualization module, a machine vision recognition module, a task flow control module, an error judgment and prompt feedback module, a teaching data acquisition and evaluation module and a teacher control terminal. The experiment steps and the three-dimensional model of the equipment are overlaid in an actual scene in real time through the augmented reality technology, student operation images are collected through a machine vision algorithm, a state vector is generated, matching degree calculation is conducted on the state vector and a standard vector, misoperation is recognized, and visual prompt is given; the recognition robustness is improved by adopting a moving average mechanism, and the teaching process is dynamically controlled through a task graph; and the scoring module comprehensively evaluates student performance based on indexes such as a misoperation rate, completion time and a prompt response rate, updates a personalized knowledge graph and realizes a teaching feedback closed loop. The visual, normative and intelligent level of experiment teaching is improved, and good practicability and popularization value are achieved.

The invention relates to an intelligent fruit tree planting monitoring method and system, and belongs to the technical field of intelligent agriculture. According to the scheme, a multi-source sensor network is deployed to collect soil, crop and environment data, a salinity-moisture collaborative stress decision-making model is constructed, and parameters such as soil salinity, canopy temperature and weather forecast are deeply fused; dynamically generating a non-fixed precise irrigation instruction; meanwhile, a low-visibility robust vision algorithm is integrated, and it is ensured that fruits and diseases and pests can still be reliably recognized in severe weather; by establishing a collaborative scheduling mechanism, efficient linkage of water and fertilizer management, plant protection operation and farming operation is realized. According to the method, the technical span from single threshold judgment to multi-factor intelligent decision is realized, the utilization efficiency of water resources and fertilizers is effectively improved, the production cost is remarkably reduced, and the mode of fruit tree planting is promoted to be converted from traditional experience management to data-driven intelligent management.

The invention provides an AI vision algorithm processing method based on a vehicle-mounted intelligent terminal and electronic equipment. The AI vision algorithm processing method comprises the steps of collecting driving scene image data in real time based on the vehicle-mounted intelligent terminal; performing multi-scale feature enhancement on the driving scene image data to generate an enhanced feature map with scene priority annotations; inputting the enhanced feature map into an AI vision algorithm based on a lightweight Transform inference model, dynamically adjusting the number of attention heads and a feature channel compression ratio, and outputting an identification result including a driving environment key target; and performing cross validation based on an identification result and heterogeneous data of the vehicle-mounted sensor, and performing real-time closed-loop correction on a feature extraction weight and a reasoning threshold value of an AI visual algorithm to complete dynamic visual perception of the driving scene. According to the method, the defects that the driving scene cannot be dynamically adapted, the identification precision cannot be balanced and the closed-loop optimization of the visual algorithm cannot be realized at present are overcome.

The invention relates to the technical field of vehicle-mounted electronic information and computer man-machine interaction, and discloses a vehicle camera APP display mode and system, and the mode comprises the steps: responding to a starting instruction, and recognizing a currently activated video input source as an in-vehicle or out-vehicle camera; dynamically loading a differentiated interaction strategy according to the type of the input source, if the mode is an in-vehicle mode, loading a man image beautifying control and a visual algorithm, and if the mode is an out-vehicle mode, shielding a beautifying function and releasing the computing power of the algorithm; a shooting trigger instruction is generated by monitoring a physical or visual logic signal, and an echo countdown state is entered after an image is collected; and if the user operation is not detected after the countdown is finished, automatically executing data archiving. The system comprises a display initialization module, a strategy configuration module, an acquisition control module and an interactive filing module. According to the invention, the potential safety hazard of manually operating the screen in the driving process is solved, and the non-inductive safe storage of the image data is realized.

The invention belongs to the technical field of industrial visual inspection, and discloses a PLC visual inspection linkage control method and system based on AI image recognition. The method comprises the steps of automatically adjusting exposure parameters of a camera based on an evaluation result of real-time brightness and contrast of an industrial product image to be detected, and obtaining a corrected industrial product image; extracting an illumination invariant feature from the corrected industrial product image, and generating an illumination invariant feature vector; inputting the original image of the same industrial product image to be detected and the corresponding illumination invariant feature vector into a pre-trained feature optimization model, and outputting an optimized illumination invariant feature vector; inputting the industrial product images shot under different exposure conditions into a pre-trained neural network model for fusion to obtain a shared feature vector; the problem of missing detection and false detection caused by illumination change in high-speed production of a traditional visual algorithm is solved.

The invention discloses a method for generating an intelligent autonomous navigation map of an agricultural robot, and relates to the technical field of agricultural robots, and the method comprises the steps: carrying out the image segmentation of a satellite map of a target farmland through an image segmentation model, so as to automatically recognize a farmland region, and rapidly extracting a farmland boundary contour through the edge curve fitting. Then a navigation path can be automatically generated based on the farmland boundary contour in combination with the basic attribute parameters of the target farmland, and an intelligent label is automatically added to obtain an intelligent autonomous navigation map for indicating the driving path and the working mode of the agricultural robot. The system realizes end-to-end automatic conversion from a satellite image to an intelligent autonomous navigation map, can adapt to diversified farmland environments with different topographic features, different crop types and different planting modes based on a deep learning computer vision algorithm, can meet the precision requirement of agricultural operation while ensuring the automation efficiency, and has a wide application prospect. And the method has strong generalization ability.

The invention relates to a riverway drowning monitoring method based on a lossless down-sampling network and multi-modal semantic disambiguation, and belongs to the field of computer vision, and the method comprises the steps: collecting heterogeneous data, carrying out the enhancement preprocessing, inputting the data into a target detection network, detecting the geometric detection confidence, and carrying out the initial screening of targets through the geometric detection confidence; for the suspected target subjected to preliminary screening, dense light smoothness of the suspected target in a continuous time window is calculated, an optical flow motion entropy is extracted according to a dense optical flow field, and a key frame image and a front-back time sequence feature map are optimized according to the motion entropy; the cloud end utilizes a visual large model to execute feature-semantic mapping based on a cross-model attention mechanism, and a textualized semantic result is generated; and carrying out Bayesian weighted fusion on the geometric detection confidence detected by the target detection network and a textualized semantic result output by the visual large model, and finally triggering a terminal sound-light alarm. According to the method, the bottleneck that a traditional vision algorithm lacks behavior semantic cognition is broken through, and the false alarm rate of false drowning is remarkably reduced.

The invention belongs to the technical field of unmanned aerial vehicle control, and particularly relates to an incremental Kd-Tree laser-inertial fusion positioning method based on improvement, the method provided by the invention adopts a practical filtering scheme, the operand is small, the real-time performance is high, and the requirement of a laser radar on an illumination condition is lower than that of a camera, so that the method is very practical. The noise is much lower than that of visual measurement, so that the precision of the method is higher than that of a visual algorithm; according to the method provided by the invention, two kinds of data are fused in a deep level, so that the method is not liable to fail during strenuous exercise, and a GPU is not needed during operation. The filtering algorithm adopts an iterative error state Kalman filtering algorithm, so that the precision of the system is not very low under nonlinear motion, and the map management efficiency is also improved by the improved incremental Kd-Tree algorithm in the aspect of map management. In conclusion, aiming at a special scene without a GNSS signal, the robot can efficiently position the robot and construct a surrounding environment map only by depending on the sensor of the robot.

The application relates to the field of intelligent path planning, and discloses an intelligent fertilizing vehicle path planning method suitable for a greenhouse environment, a road surface obstacle risk value evaluation mechanism based on an expert system and AHP (analytic hierarchy process) level analysis, road surface point cloud data established by combining binocular vision algorithm and deep learning technology, and comprehensive evaluation of road surface obstacle risks is realized. The obstacle avoidance mechanism of the traditional path planning method is effectively improved, so that the fertilizing vehicle can effectively deal with the path planning scene of a muddy path and a plant-shielded path in the greenhouse. The application improves the accessibility, timeliness and safety of the path planning of the greenhouse fertilizing vehicle.

This invention relates to the field of construction safety monitoring and intelligent visual analysis technology, specifically a wiring-free construction safety monitoring system based on visual algorithm analysis. The system includes: an independent power supply module, a visual perception module, a scene risk topology assessment module, a computing-risk coupling scheduling module, a multi-level wake-up execution module, and a context-related early warning output module. It collects initial visual flow data, remaining power, and predicted charging power of the target monitoring area; divides the topology regions into different risk levels through semantic segmentation and assigns risk weights; dynamically generates computing power scheduling instructions containing the target sampling frame rate and target model complexity; outputs the target violation event identification results; and generates the final early warning result by combining the spatial inclusion relationship between the occurrence location and the topology region boundary. This achieves priority protection for high-risk spatiotemporal nodes and reduces ineffective energy consumption and invalid alarms in low-risk scenarios.

Embodiments of the present application disclose a vehicle-mounted visual algorithm edge computing optimization method, device, equipment and storage medium, comprising: preprocessing image data collected from a road environment to extract scale features; receiving radio ranging data transmitted by a radio device, generating fusion data based on the image data and the radio ranging data; determining the spatial position information and threat level of a target based on the multi-scale features and the fusion data, calculating the computational complexity of an inference task, establishing a task priority queue, obtaining the state information of each hardware of a vehicle-mounted edge computing unit, and adjusting the task priority queue; and distributing the inference task to the corresponding computing unit through the adjusted task priority queue and the computational complexity. Through the above processing of the image data and the radio ranging data, the problem of being unable to complete the inference within a limited time, existing response delay or uneven distribution of computing power under resource limited conditions is improved.

The embodiment of the invention provides a visual editing system based on a LabVIEW platform, and the system comprises a project management module which is used for carrying out the creation, deletion, selection and loading operation of a visual algorithm project, and an image collection and display module which is used for displaying an intermediate result and a final result of an image in a processing process. The visual operator module is used for providing various visual operators for a user to call in the form of a graphical component, a bottom visual processing function of the LabVIEW platform is integrated in the graphical component, and the algorithm step module is used for receiving a selection instruction of the user for the visual operators from the visual operator module. According to the method, a user selects a plurality of visual operators, an editable visual algorithm process formed by the plurality of visual operators is recorded according to the order of the selection instruction, the system separates the visual algorithm process constructed by the user from a program source code and stores the visual algorithm process as a structured configuration file, and field personnel can directly debug, store and call an algorithm without modifying codes. And the maintainability and the flexibility of the system are obviously improved.

The invention discloses a method and system for monitoring the start-stop state of a conveying belt based on a traditional visual algorithm, and belongs to the technical field of traditional visual algorithms, and the method comprises the steps: installing an industrial camera at a proper position above the conveying belt; after receiving the original image acquired by the industrial camera, the monitoring host carries out graying processing on the image and carries out de-noising processing on the grayscale image; performing histogram equalization processing on the denoised image to enhance the contrast of the image; selecting a plurality of feature points in the belt area, and recording coordinates and feature descriptors of each feature point; for every two adjacent frames of images, matching the extracted feature points by using a feature point matching algorithm, and calculating an optical flow vector of each pixel point between the adjacent frames by using a feature descriptor-based matching algorithm and an optical flow method; and outputting a state and giving an alarm. Manual intervention is greatly reduced, accuracy is high, cost is low, adaptability is high, and implementation and maintenance are easy.

This invention relates to the fields of artificial intelligence and computer vision, and discloses a method for detecting residual manganese in manganese electrolysis plates based on multi-expert convolution. The method includes: constructing a first branch based on depthwise separable convolutional blocks combined with routing multilayer perception and convolutional layers with different kernels; constructing a second branch based on depthwise separable convolutional blocks and residual connections; constructing a third branch based on depthwise separable convolutional blocks; constructing a hybrid convolutional block by combining the first, second, and third branches with multi-head attention; constructing a detection backbone based on the hybrid convolutional block combined with multilayer convolution; constructing a multi-scale fusion structure based on multi-scale feature extraction and routing selection; constructing a single-input dual-output prediction head based on the hybrid convolutional block; and constructing a detection model based on the detection backbone, multi-scale fusion structure, and prediction head; acquiring an image of the manganese electrolysis plate and inputting it into the detection model to obtain the detection result of residual manganese. This method solves the problem that existing visual algorithms cannot accurately identify residual manganese in manganese electrolysis plates.

The present application belongs to the technical field of oil use peony phenotype parameter automatic extraction method, especially relates to a kind of oil use peony phenotype parameter automatic extraction method and system based on multi-organ image analysis, obtain the high-resolution original image of oil use peony multi-organ containing ColorChecker24 standard color card, the organ at least includes petal, blade, fruit pod;Color correction and geometric correction are executed to original image, 24 color blocks of color card are positioned using computer vision algorithm, color conversion matrix is constructed to map image from device-dependent RGB space to standard color space, while calculating pixel scale based on color block physical size;Four corner points of bearing tray are detected for fruit pod image, and perspective transformation matrix is constructed to complete perspective distortion correction.

The invention discloses a system and method for adaptively controlling the hedging function of a sanitation cleaning vehicle, and the method comprises the steps: a vehicle-mounted intelligent terminal senses whether there is a pedestrian or a target distance of the type of a pedestrian at two sides of the vehicle through a visual sensor disposed on the vehicle; calculating the transverse distance and the longitudinal distance between the vehicle and the traveler, the rider, the road edge and the guardrail according to a visual algorithm; according to the transverse distance of the pedestrian or the riding pedestrian and the opposite-punching distance table corresponding to the rotating speed, a control message is sent to the uploading control screen, the uploading control screen issues a control instruction to the uploading controller, the uploading controller executes the control instruction, and the opposite-punching operation coverage distance and the opposite-punching function are automatically adjusted in real time. The hedging operation process of the cleaning vehicle is optimized, the intelligent level is improved, the loading controller receives a hedging self-adaptive control command of the vehicle-mounted intelligent terminal through the loading control screen, the left and right hedging electric control ball valves are controlled to be opened or closed in real time, and finally real-time dynamic control over the hedging distance is achieved.

The invention discloses an intelligent detection method and system for garment quality inspection, and relates to the technical field of computer vision and intelligent manufacturing. The method comprises the following steps: carrying out pixel-level feature analysis on two-dimensional image data, identifying a sewing line and a printing area, and mapping the sewing line and the printing area to a unified space coordinate system; reconstructing the three-dimensional surface form of the garment based on a stereoscopic vision algorithm, and calculating the dimensional deviation of key parts; converting the spectral data into a CIELAB color space, and carrying out optical distortion correction on local chromatic aberration in combination with a three-dimensional curvature; and under a unified space coordinate system, performing correlation analysis on the stitches, the dimensional deviation and the corrected space distribution, identifying a composite defect which cannot be judged by a single defect, generating a correlation defect identifier, and finally forming a quality report. The technical problems that in the prior art, local chromatic aberration is easily influenced by three-dimensional curved surface deformation and optical distortion of ready-made clothes, information such as dimensional deviation, sewing and printing defects and the local chromatic aberration is isolated, comprehensive correlation analysis is lacked, and consequently the detection accuracy is insufficient are solved.

The invention discloses a soft air refueling algorithm training and verification method based on an unreal engine, and the method comprises the steps: constructing an air refueling scene according to a constructed refueling taper sleeve model and a terrain and weather scene, and simulating a refueling and receiving docking process; generating a target detection label according to the obtained scene mask image; wherein the target detection mark comprises a normalized central point two-dimensional coordinate of the refueling taper sleeve model and the width and height of a minimum enclosing rectangle; and taking the target detection label and the key point detection label corresponding to the scene image as labels of the scene image to construct a training data set so as to train and verify a visual algorithm. According to the method, high-fidelity simulation scene images and corresponding detection frames and key point detection labels can be generated in batches under various environmental conditions, the cost is low, and the constructed training data set can be used for training and verification of a visual algorithm, so that the algorithm is iterated and widely adapted.

1. Name of the design product: Edge vision analysis box (T-800). 2. Use of the design product: For protecting the integrated circuit related to executing vision algorithm. 3. Design points of the design product: In shape. 4. Picture or photograph best indicating the design points: Perspective view.

The application relates to the field of intelligent operation and maintenance control, in particular to an intelligent operation and maintenance method, device and system based on a visual algorithm and electronic equipment; the method comprises the following steps: if a first fault event is triggered again under the condition that a data completion event is triggered in advance in a digital twin model, generating an image acquisition instruction for an object device corresponding to the data completion event, collecting a live image of the object device through a target device, extracting running data of the object device from the live image through a visual algorithm, and further feeding back the running data to the digital twin model, so that the digital twin model can make fault prediction again; if the digital twin model triggers a second fault event which is the same as a fault device corresponding to the first fault event again, it is determined that the first event is valid, otherwise, the first event is invalid. Through the method, the operation and maintenance cost of the system can be reduced.

The invention discloses a rural sewage treatment station inspection management control system based on an intelligent wearable terminal. The system generates a first preset path based on inspection task requirements and environment distribution information through a path generation module, and guides an inspector to move according to the path through a sending module; the obtaining module collects position information and behavior characteristics of inspection personnel in real time. And the prompt module triggers a preset warning event according to the inspection data and sends prompt information. Meanwhile, the system optimizes path generation by adopting a preset multi-dimensional data analysis model, and realizes image information space positioning through a multi-view stereoscopic vision algorithm and the like. The system effectively solves the problems of difficulty in personnel monitoring, single equipment function and the like in rural sewage treatment station inspection, and can monitor the state of inspection personnel in real time, provide efficient communication and emergency support, improve the inspection efficiency and standardization, guarantee personnel safety and enhance the emergency response capability of the system in a complex environment.

This disclosure presents a system and method for automating the monitoring, reporting, and verification (MRV) of sustainability projects. The system includes a mobile application for users to submit data and images, which are transmitted to a cloud-based storage system. Metadata from the images is extracted and validated against predefined coordinates. A computer vision algorithm evaluates the images, and results are stored in a cloud-based database. The system further includes a secondary verification module using remotely sensed imagery. This automated and distributed approach reduces the cost, complexity, and time of MRV processes, enhancing transparency, reliability, and accuracy. The method supports scalability across diverse sustainability projects, including regenerative agriculture, biodiversity, and greenhouse gas reduction, by leveraging integrated data collection, automated analysis, and distributed network validation.

The invention provides a point and line feature-based pose estimation method and electronic equipment. The method comprises the following steps of extracting point features and line features from an environment image; calculating one or more vanishing points in the environment image based on the extracted line features; according to the geometrical relationship between each line feature and the vanishing point, grouping the line features into different dominant direction sets, and removing the line features which do not belong to any dominant direction set to obtain an optimized line feature set; and fusing the point features and the optimized line feature set, taking the fused point features and the optimized line feature set as input together, performing pose estimation through a visual SLAM algorithm, and outputting pose information of the camera. According to the method, the weak texture environment is effectively dealt with, the point features and the line features are fused, and the line features are optimized by using the vanishing points, so that the key problems that the traditional visual SLAM which only depends on the point features is rare in features and easy to fail in tracking in the low-texture structured environment are solved, and stable and high-precision positioning in the severe environment is realized.

The invention relates to the field of integrated circuits, in particular to an intelligent visual SoC chip adaptive to an electric power scene and a development method of the intelligent visual SoC chip. The method comprises the following steps: constructing a chip design constraint model based on power industry environment parameters; according to the chip design constraint model, customizing a hardware architecture of the chip; integrating a visual algorithm module facing a power scene; performing deep coupling on the visual algorithm module and a hardware architecture through a special hardware acceleration unit to realize collaborative execution of an algorithm and hardware; constructing an environment-function collaborative verification platform, carrying out function and performance collaborative verification on the SoC chip under a multi-environment coupling condition, and establishing a linkage optimization mechanism of hardware parameters and algorithm parameters based on a verification result, so that the visual recognition accuracy rate of the chip under environment disturbance is controlled within a preset threshold value; and carrying out industrial packaging and reliability testing. The problem that a traditional visual chip is poor in stability in an electric power environment is solved, and the intelligent visual monitoring reliability of electric power equipment is improved.

The application discloses a vehicle identification and tracking method based on deep learning, which is characterized by the following steps: enabling a deep learning computer vision algorithm for a monitoring camera or a camera of a unmanned aerial vehicle, analyzing a real-time video stream, detecting and tracking a vehicle appearing in the video stream, filtering a specific vehicle by input information, capturing the specific vehicle, uploading a captured image to a background system for recording, and fully utilizing an existing hardware camera with high identification precision.

The invention relates to the technical field of computer vision and industrial automation, in particular to a hoisting equipment real-time positioning and trajectory tracking method based on a vision algorithm, which comprises the following steps of: S1, acquiring a video stream in real time through a camera deployed in a hoisting operation area, and preprocessing an image; according to the method, accurate mapping from two-dimensional image features to a global three-dimensional space in a complex hoisting scene is realized through a multi-source constraint fusion self-correction three-dimensional pose solving mechanism, and a trajectory prediction and'trajectory-load 'associated safety system is constructed based on a three-dimensional positioning result; the closed-loop control from passive trajectory tracking to active prediction planning, overload early warning and equipment linkage protection is realized, and the positioning precision, robustness and safety protection level of hoisting operation are greatly improved.