36 results about "Vision processing unit" patented technology

The invention discloses an online monitoring system and method for ultrafast laser processing of micropores. The online monitoring system comprises a visual monitoring unit, a sound wave monitoring unit and a judgment control module which are arranged in an ultrafast laser processing unit. The visual monitoring unit is used for obtaining a visual monitoring result and comprises a visual acquisition unit and a visual processing unit; the sound wave monitoring unit is used for obtaining a sound wave monitoring result and comprises a sound wave acquisition unit and a sound wave processing unit; and the judgment control module is used for comprehensively judging the visual monitoring result and the sound wave monitoring result so as to control the ultrafast laser processing unit. According to the technical scheme, online sound-vision composite monitoring can be carried out on the process of ultrafast laser micro-hole machining, and therefore the micro-hole machining efficiency and the penetration rate are improved.

The invention discloses an intelligent grabbing system anomaly detection and processing method and device based on 3D vision, and belongs to the technical field of industrial robot 3D vision guide grabbing. The method comprises the steps of collecting 3D point cloud data of stacked flexible plates; carrying out point cloud preprocessing and template matching, and determining a plate pose and a grabbing point; generating a cushion block detection area and a deformation detection area based on the grabbed point pose; counting the number of point clouds in the detection area, comparing the number with a threshold value, and judging whether cushion block residues or plate excessive deformation exists or not; if yes, triggering processing, and if not, outputting a grabbing instruction. The device comprises a truss manipulator, an end effector, a 3D visual sensor, a visual processing unit and a robot control unit. According to the method, the detection area is dynamically generated, point cloud threshold judgment is fused, self-adaptive detection and processing of excessive deformation of the cushion block and the plate are achieved, the robustness, safety and automation level of a grabbing system are remarkably improved, and the method is suitable for automatic unstacking operation of stacked flexible plates.

The invention provides video processing equipment and a video processing method. The video processing device comprises a multi-core processor, a task scheduling module, a visual processing unit and a cooperative control module, and the multi-core processor is used for sending a task queue corresponding to a video stream to the task scheduling module; the task queue comprises a plurality of to-be-processed tasks and real-time levels corresponding to the tasks; the cooperative control module is used for monitoring the states of the visual processing unit and the multi-core processor to obtain state information and sending the state information to the task scheduling module; the task scheduling module is used for dynamically distributing each task to a target processing component corresponding to each task according to the real-time level and state information corresponding to each task, and each target processing component comprises a multi-core processor and a visual processing unit for processing; the target processing component is used for processing each task to obtain an output result corresponding to the video stream; and the multi-core processor is used for outputting an output result corresponding to the video stream.

The invention relates to the technical field of aquaculture, and discloses a marine product seedling raising accurate feeding control system which comprises an intelligent recognition module, a data analysis and decision module, an accurate feeding execution module, an environment monitoring module and a remote control and information management module. Through a high-definition camera and a computer vision processing unit of the intelligent identification module, key information such as the body type, the number and the activity state of the fry is accurately identified by adopting a deep learning algorithm. The big data analysis unit is combined with fry growth stages, types and density factors, a machine learning technology and a preset biological growth model are used for accurately calculating the real-time feed demand of each area, and the decision making unit generates an accurate feeding strategy according to the real-time feed demand. And a feeder of the precise feeding execution module realizes timed, quantitative and fixed-point feeding according to the strategy, so that the problem of inaccurate feeding amount caused by experience judgment in a traditional feeding mode is effectively avoided, the feed waste is greatly reduced, and the feed utilization rate is remarkably improved.

The invention discloses a double-station solder ball welding machine correction system based on machine vision and multi-axis cooperation, and belongs to the field of welding equipment correction. A double-station solder ball welding machine correction system based on machine vision and multi-axis cooperation comprises an image acquisition unit, a visual processing unit, a multi-axis control unit and a cooperative scheduling unit. The problems that in existing double-station welding equipment, due to the fact that a single-station vision scheme is simply copied, the equipment takt is slow, the productivity is limited, and comprehensive compensation for dynamic errors and optical environment differences under multi-station collaborative operation is lacked are solved. Compared with the prior art, the double-station equipment has the advantages that overlapped execution of visual correction and welding operation is achieved, and feed-forward and online compensation mechanisms for dynamic errors such as thermal deformation and vibration are integrated, so that high-precision positioning is guaranteed, and the production efficiency and the operation reliability of the double-station equipment are remarkably improved.

Systems, devices, and methods for autonomous mowing in turf environments are described. The mowing vehicle integrates manual controls, a display, and a controller connected to a sensor system all-around coverage. Sensors feed a vision processing unit that detects, classifies, and maps transient and non-transient objects. Modules within the vision processing unit handle object detection, classification, boundary generation, path planning, actuation, user interaction, and job quality monitoring. Path planning adapts to dynamic conditions, other machines, and changing object sizes, while the actuation module executes adjusted paths in real time. The user interface enables operator oversight and settings adjustments, with cloud connectivity for data storage and reuse. Job quality is assessed through sensor and image data, enabling corrective actions.

The invention discloses an organoid ball transfer system, a transfer method and an automatic organoid culture platform, and relates to the field of organoid culture. The invention aims to solve the problems that the existing culture and transfer planting of the organoids mainly depend on manual operation, the transfer efficiency, the position precision and the culture consistency are difficult to guarantee, and the stability and the repeatability of experimental results are influenced by the manual operation. According to the organ-like ball transfer system, the transfer platform is adopted to execute the transfer work of the organ-like balls; and adopting a visual processing unit to track and position the organ-like ball in the transfer process of the organ-like ball. The organoid cell balls can be quickly and nondestructively transferred into the liquid matrigel, and the blank of an organoid automatic system of a small chamber culture system is filled.

The present invention provides an airport driverless vision detection system and a baggage tractor, belonging to the technical field of airport baggage tractors. The system includes a vision monitoring module and an active intervention module. The vision monitoring module and the active intervention module are communicatively connected. The vision monitoring module includes a vision monitoring unit and a vision processing unit. The vision monitoring unit is used to acquire image and video information of the baggage, and the vision processing unit is used to identify and generate status information of the baggage according to the image and video information of the baggage. By setting up the vision detection module and the active intervention module, the present invention realizes the real-time monitoring and protection of the baggage on the tractor. Through the vision detection module, the status of the baggage during transportation can be monitored at all times, specifically its shaking status and whether it has fallen. When shaking occurs, the intervention module can cooperate to provide a proper restraint and protection for the baggage in real time, preventing damage to the baggage caused by excessive shaking.

The invention relates to the technical field of intelligent manufacturing, and discloses an intelligent feeding system and method combining machine vision and path planning. The system comprises a visual acquisition unit, a visual processing unit, a path planning unit, a track monitoring and re-planning unit and a motion control unit. The visual processing unit is used for generating environmental parameters including material pose information, obstacle boundary information and identification confidence parameters; the path planning unit constructs a space cost field model based on the environmental parameters and generates an initial motion track; the track monitoring and re-planning unit triggers local track re-planning according to the obstacle position variable quantity and the accumulative change trend in the executing process of the mechanical arm; and the motion control unit executes hierarchical speed regulation according to the identification confidence coefficient parameter and dynamically adjusts the safety distance in combination with the running speed. By means of the closed-loop control mechanism, real-time response and self-adaptive motion control over the dynamic environment are achieved, and the stability and safety of feeding operation are improved.

An object of the system according to the embodiment is to integrally perform generation of an avatar, transition to a Metaverse, storage of a record of conversation with a user, visual processing of application information, and payment of an incentive to a voice actor.SOLUTION: A system includes a generation part, a recording part, a visual processing part, a transition part, and an incentive payment part. The creator immediately creates the Live2D based avatar. The recording unit stores the conversation with the user for a long time. The visual processing portion visually processes the active application information. The transitioner causes the avatar to transition to the Metaverse upon deactivation. The incentive payment unit pays an incentive to the voice actor.SELECTED DRAWING: Figure 1

An object of a system according to an embodiment is to stereoscopically visualize a product in a store so that a user can purchase the product while freely moving around.SOLUTION: A system includes an installation part, a visual processing part, a reproduction part, a navigation part, and a transition part. The installation part installs a camera for acquiring three dimensional data in a store. The visual processing part performs visual processing of displayed commodities and inventory commodities in the store on the basis of the data acquired by the installation part. The reproduction part reproduces the commodity on the three dimensional shop data generated by the visual processing part. The navigation section performs navigation for a user to freely move around the shop data reproduced by the reproduction section and purchase a commodity. The transition unit transitions from the electronic map service to the stereoscopic EC shop.SELECTED DRAWING: Figure 1

A visual tracking system for tracking and identifying persons within a monitored location, comprising a plurality of cameras and a visual processing unit, each camera produces a sequence of video frames depicting one or more of the persons, the visual processing unit is adapted to maintain a coherent track identity for each person across the plurality of cameras using a combination of motion data and visual featurization data, and further determine demographic data and sentiment data using the visual featurization data, the visual tracking system further having a recommendation module adapted to identify a customer need for each person using the sentiment data of the person in addition to context data, and generate an action recommendation for addressing the customer need, the visual tracking system is operably connected to a customer-oriented device configured to perform a customer-oriented action in accordance with the action recommendation.

The invention discloses a multi-modal visual servo four-degree-of-freedom experimental recording holder device and a control method. The device comprises a holder end, a slide rail control end and an upper computer end, the holder end is provided with a first main control unit, a posture sensing unit, a motor driving unit, a visual processing unit and an execution unit; the sliding rail control end is provided with a second main control unit, a user interaction unit, a communication unit and a horizontal driving unit; and the upper computer end is connected with the visual processing unit through a network communication link. The method comprises the following steps: after a system is initialized, automatically triggering to enter different modes according to user selection or scenes; the visual processing unit collects an image, performs multi-modal target identification through a lightweight neural network, and processes shielding in combination with a trajectory prediction model; the first main control unit calculates attitude deviation according to the target position, and dynamically adjusts control parameters to drive the execution unit; and the second main control unit judges whether to start the cooperative movement of the slide rail or not according to the target horizontal position, and executes safe reset when reaching the limit.

The invention discloses a ground-imitating autonomous flight method and system of an unmanned aerial vehicle along a road, and belongs to the technical field of autonomous flight of unmanned aerial vehicles. The method comprises the steps that road images are collected in real time through a down-looking camera carried by an unmanned aerial vehicle, an edge enhanced double-branch road segmentation network is executed by a visual processing unit, and a road segmentation map is generated; adopting a random seed ray bidirectional boundary search algorithm to detect road boundaries from the segmentation map and fitting a center line; controlling the course of the unmanned aerial vehicle according to the deviation between the image center and the road center line; meanwhile, the relative height between the unmanned aerial vehicle and the road surface is measured through the laser ranging module, and precise terrain-imitating flight is achieved through a height control strategy. The system does not depend on satellite positioning and a digital elevation model (DEM), has the characteristics of road perception robustness, accurate boundary extraction, stable flight control and reliable height estimation, is suitable for a complex road environment without satellite signals and a prior map, and remarkably improves the automation level and adaptability of road inspection.

A visual tracking system for tracking and identifying persons within a monitored location, comprising a plurality of cameras and a visual processing unit, each camera produces a sequence of video frames depicting one or more of the persons, the visual processing unit is adapted to maintain a coherent track identity for each person across the plurality of cameras using a combination of motion data and visual featurization data, and further determine demographic data and sentiment data using the visual featurization data, the visual tracking system further having a recommendation module adapted to identify a customer need for each person using the sentiment data of the person in addition to context data, and generate an action recommendation for addressing the customer need, the visual tracking system is operably connected to a customer-oriented device configured to perform a customer-oriented action in accordance with the action recommendation.

The invention relates to the technical field of machine vision and automatic control, and discloses a product nondestructive testing and sorting system based on machine vision, comprising a visual processing unit used for collecting product images, operating an analysis algorithm to determine grade information of products, and sending the grade information to a main control unit; the main control unit is in communication connection with the visual processing unit and is used for receiving the grade information and forwarding the grade information to the sensing and control unit; and the sensing and control unit is in communication connection with the main control unit and comprises a sensing layer, a control layer, an execution layer, a display layer, a network layer and an application layer. According to the system and the method, a three-unit distributed architecture of visual processing, master control, perception and control is adopted, task decoupling is realized, the operation stability is improved, system integration defect quantification and maturity analysis of an LAB color space are comprehensively graded, the detection accuracy is improved, and environment monitoring and accurate sorting are realized.

The application relates to the technical field of photoelectric tracking and confrontation, and discloses a high-precision laser dazzling closed-loop control system based on a deep learning vision algorithm, which comprises a laser dazzling rejection unit and a high-precision servo tracking device, adopts a double-rate asynchronous closed-loop architecture, a deep learning vision processing unit outputs target observation data with absolute time stamps in a slow loop, a high-precision servo controller executes high-frequency deduction based on inertial data in a fast loop, utilizes a historical state buffer queue and a forward reintegration algorithm to perform space-time alignment measurement update, and eliminates vision delay; the system further establishes an adaptive region of interest adjustment mechanism based on interaction of a prediction covariance matrix, and integrates a reverse PWM trigger exposure control, a dynamic geographic fence verification and a power nonlinear modulation function. The application effectively solves the tracking lag problem caused by vision delay under high dynamic conditions, and improves the tracking precision, robustness and operation safety of the system.

The invention provides a braille reading system based on phased array ultrasonic tactile feedback, and the system comprises a visual processing unit which is configured to collect the graphic and text information of a reading material, and transmits the graphic and text information to an upper computer; the upper computer is configured to distinguish a text region and a graphic region in the graphic and text information through a target detection algorithm; processing and generating a static focus coordinate sequence corresponding to the braille dot matrix; discretization processing is carried out on the graph area, feature points are extracted, and a dynamic focus track point sequence corresponding to the graph contour is generated; the tactile control driving unit is used for receiving the static focus coordinate sequence or the dynamic focus track point sequence, operating a phase optimization algorithm to calculate phase delay and modulation information, and generating multiple paths of driving signals after power amplification; and the ultrasonic transducer array is configured to respond to the driving signal to emit ultrasonic waves, and the ultrasonic waves are focused in the space through the phased array to form a tactile focus, so that non-contact braille and graph tactile perception is realized.

The application belongs to the cross field of computer vision and embedded system, and particularly relates to a monocular vision measurement system and method based on a dual-core heterogeneous architecture. The monocular vision measurement system based on the dual-core heterogeneous architecture comprises an image acquisition module, a vision processing unit, a current detection module and a control unit, and a man-machine interaction module. The image acquisition module acquires monocular images. The vision processing unit executes target detection, contour extraction, perspective transformation and size measurement algorithms. The vision processing unit receives the monocular images and outputs target measurement results. The current detection module is electrically connected with the control unit, and is used for monitoring system power consumption in real time and outputting a proportional voltage signal. The control unit is used for system scheduling, current signal acquisition and communication control. The control unit receives the target measurement results and the proportional voltage signal input by the current detection module, and outputs them to the man-machine interaction module. The man-machine interaction module is used for displaying measurement results and system states. Through the heterogeneous design of the control unit and the vision processing unit, effective cooperation of high-precision vision processing and low-power real-time control is realized.

The invention provides an unmanned aerial vehicle visual language motion navigation method and system based on sparse representation, and the method comprises the steps: converting an obtained image into a plurality of visual processing units, and carrying out the sparse processing of the plurality of visual processing units, so as to determine a key visual processing unit. And performing context merging processing on the remaining visual processing units to obtain a merged visual processing unit. And combining the key visual processing unit and the merged visual processing unit to obtain a visual processing unit set. And generating a predicted action sequence based on the obtained language instruction and the visual processing unit set, and performing navigation control of the unmanned aerial vehicle based on the predicted action sequence. In the scheme, through sparse processing of the visual processing unit, the processing amount can be reduced, navigation is carried out in combination with the language instruction of the user, and more accurate and specific navigation can be realized.

A vision system and method for a pine cone harvesting platform, belonging to the field of intelligent forestry equipment, addresses the problems of low target recognition accuracy, difficulty in maturity assessment, and insufficient harvesting positioning precision in existing drone harvesting systems operating in complex forest environments. The system includes a vision processing unit, a wide-angle gimbal camera, a LiDAR, and a hand-eye camera. The wide-angle gimbal camera acquires distant visual images. The vision processing unit uses a priori information-guided detection method to identify pine cone targets and combines LiDAR point cloud data to plan the platform's motion path. The hand-eye camera, mounted at the end of a robotic arm, acquires near-field images and depth information after the platform reaches the target area. The vision processing unit uses an attention-weighted mechanism to fuse color, size, and scale flipping angle features to determine maturity, locate the strike point, and control the harvesting actuator to complete the operation. This invention achieves high-precision identification and positioning of pine cone targets through two-stage visual collaboration and multi-dimensional feature fusion.

This invention relates to the field of measurement and testing technology, and discloses a comprehensive valve inspection device and method, comprising: a loading module and a rotating support module fixed on a frame; the rotating support module is equipped with an intermittent rotating turntable, and a multi-station vision inspection module is arranged circumferentially along the intermittent rotating turntable; the multi-station vision inspection module includes a disc surface inspection station, a rod diameter inspection station, a rod length inspection station, a neck inspection station, and a rod end face inspection station arranged sequentially, each station being equipped with a corresponding vision inspection mechanism; the control module includes a vision processing unit and a control unit, which are electrically connected to the loading module, the rotating support module, and the multi-station vision inspection module, respectively. This invention solves the problems of single inspection function, reliance on manual labor, low efficiency, and lack of data management, achieving the goals of automatic multi-parameter inspection, high precision, high efficiency, and intelligent recognition.

The invention relates to a small vehicle for field pesticide-free weeding, and solves the technical problems that a traditional weeding machine cannot enter the field in the growing period of long-stalked crops, and weeding is not conducted without removing roots. The vehicle body unit achieves all-terrain self-adaption through an adjustable auxiliary wheel assembly, a telescopic motor of an inner main frame telescopic motor box controls the ground clearance of an auxiliary wheel, an arc-shaped gear motor of an outer main frame motor box drives a wheel set to be transversely unfolded along an arc-shaped groove, and the vehicle can walk in a balanced mode between long-stalk crop field ridges of 20 cm to 1 m. The mechanical claw unit adopts a three-head mechanical claw single body moving in a two-dimensional plane, and the roots of weeds are instantly clamped and pulled out by an electromagnetic driving shaft; and the visual processing unit executes a localization weed recognition algorithm through the binocular camera and the core controller, precisely positions weed coordinates and cooperatively controls the vehicle body and the mechanical claw to act. The height limitation of a traditional agricultural machine is broken through, the weeding eradication rate is larger than 95%, the weight is reduced by 60% or above, solar autonomous power supply and obstacle crossing capacity are achieved, and the device is suitable for full-growth-cycle operation of long-stalk crops such as corn.

The invention discloses an automatic sorting system and method for yellow leaves and green leaves of sugarcane tails. The automatic sorting system comprises a feeding conveying device, an image collecting device, a visual processing unit and a sorting execution device which are sequentially arranged. The system obtains a sugarcane leaf material image through an image acquisition device, and a visual processing unit comprehensively analyzes color, shape and texture features based on a multi-feature fusion recognition algorithm, accurately distinguishes sugarcane tips, green leaves and yellow leaves, and generates a sorting instruction; the sorting execution device sorts different types of materials to corresponding discharging channels at the accurate moment through the angle and air pressure adjustable high-pressure air flow nozzle array according to the instruction, and finally the materials enter the independent smashing and packaging unit. According to the sugarcane leaf sorting device, high-precision and high-efficiency automatic sorting of sugarcane leaves is achieved, the problem of value loss caused by mixing treatment is effectively solved, and the economic value and the resource utilization level of sugarcane by-products are remarkably improved.

The application provides a kind of unmanned aerial vehicle visual language action navigation method and system based on sparsification representation, by converting the obtained image into multiple visual processing units, and the key visual processing unit in multiple visual processing units is determined by sparsification processing.Remaining other visual processing units are subjected to context merging processing, to obtain merged visual processing units.Combining key visual processing units and merged visual processing units obtains visual processing unit set.Based on the obtained language instruction and visual processing unit set, a predicted action sequence is generated, and the navigation control of the unmanned aerial vehicle is executed based on the predicted action sequence.In this scheme, by sparsification processing of visual processing unit, the processing amount can be reduced, and navigation can be more accurate and specific by combining the language instruction of the user.

The invention relates to the technical field of intelligent fitness, in particular to an AI-based annular exercise fitness planning interaction method and system.The method comprises the steps that a fitness planning interaction instruction is received, an intelligent fitness system is confirmed based on the fitness planning interaction instruction, a static parameter value set is obtained based on a data acquisition unit, and the static parameter value set is calculated; based on the static parameter value sets, a pre-constructed reinforcement learning model and a training scheme generation unit, generating a personalized annular training scheme, and if it is confirmed that the multiple dynamic parameter value sets meet a preset first warning condition, obtaining an action matching coefficient, and acquiring a correction scheme based on a pre-constructed visual processing unit, an action matching coefficient and a real-time interaction and correction unit, acquiring an adjusted training posture based on the correction scheme, and realizing AI-based annular exercise fitness planning interaction after confirming that the evaluation quantity and the adjusted total quantity meet a preset evaluation condition. According to the invention, the accuracy and intelligence of fitness planning interaction can be improved.

The invention belongs to the technical field of automobile detection, and particularly relates to a non-contact four-wheel aligner which comprises an integral steel structure, a steering wheel gradienter, an electrical signal processing module and a PLC (programmable logic controller) control system. Centering and positioning mechanisms are installed at the positions, corresponding to the front roller sets and the rear roller sets, of the top of the whole steel structure, XDP visual processing units are arranged on the centering and positioning mechanisms, the installation positions of the two front roller sets and the two rear roller sets on the whole steel structure are different, and other structures are the same. The front roller set comprises a floating mechanism and a roller mechanism installed on the floating mechanism, and the floating mechanism and the roller mechanism provide supporting and floating for four tires of the vehicle. The detection accuracy can be improved, the defects in the prior art are effectively overcome, and the performance of the four-wheel aligner is greatly improved.

The invention discloses a PE film defect detection method and detection equipment based on industrial vision. The device comprises a transmission and reflection imaging channel, a coding synchronization mechanism and an industrial visual processing unit. The method comprises the following steps: synchronously acquiring transmission / reflection banded images, constructing a thickness-defect dual-task deep learning model taking a dual-channel image as input, and carrying out joint training on thickness regression, defect segmentation and thickness equivalent region background consistency by using joint loss; in the online reasoning stage, a film thickness distribution and defect probability graph is output, reflection brightness thickness compensation is carried out based on thickness prediction, and a defect mask and a defect list are generated by combining an adaptive threshold value and connected domain analysis. Compared with an existing single-channel threshold method, the method has the advantages that thickness moire and real defects can be effectively distinguished, the false alarm rate is remarkably reduced, the detection rate of small-size and low-contrast defects is increased, and the method is suitable for real-time quality monitoring of a multi-formula and multi-working-condition PE film production line.

The invention discloses an interaction system and method for visually impaired people to interact with an environment, relates to the technical field of image recognition, and aims to solve the technical problems that the visually impaired people are difficult to acquire and process information in an office administrative scene and are difficult to accurately associate the information, and the interaction system comprises an intelligent glasses terminal and a background service system; the intelligent glasses terminal comprises a multi-mode sensing module, an equipment space positioning module, a processing and communication module, an audio output module and a user interface, and is used for realizing user instruction input; the background service system is deployed in a local server and comprises an equipment connection and management unit, a computer vision processing unit, an interaction intention analysis and decision unit, a voice synthesis unit, a prediction model and a safety protection unit. The intelligent glasses terminal is combined with the background service system, so that the problem of information association between cross-equipment and a paper carrier in the prior art is thoroughly solved, and the autonomy of the visually impaired people for handling the office administrative affairs is remarkably improved.

The invention discloses a solar intelligent strawberry picking robot which comprises a moving platform, and an operation table, a solar cell panel, a plurality of picking mechanical arms and a storage basket are arranged on the top of the moving platform. A control bin is arranged at the bottom of the mobile platform; walking systems are arranged on two sides of the control bin; walking identification cameras are arranged at the front end and the rear end of the mobile platform; a picking recognition camera is arranged at the end of the picking mechanical arm. Compared with the prior art, the device has the advantages that the picking efficiency is improved, the manual labor intensity is reduced through automatic picking, and the picking speed and efficiency are improved; the picking cost is reduced, solar energy is used for supplying power, energy consumption is reduced, and meanwhile labor cost is reduced; the picking precision is improved, accurate positioning and maturity judgment of the strawberries are achieved through the visual processing unit, and mistaken picking and missing picking are reduced; and the walking recognition camera and the picking recognition camera are provided with infrared light supplementing modules, so that the device adapts to a weak light environment, and all-weather operation is ensured.