51 results about "Artificial vision" patented technology

The invention relates to the technical field of photovoltaic module detection, in particular to a TOPCon battery piece defect detection device and a TOPCon battery piece defect detection method. The optical imaging module comprises a 25,000,000-pixel camera, a three-color integral light source array and a multispectral optical filter group, and the light source array is composed of red, green and blue LEDs and supports pulse synchronous triggering; the traditional algorithm processing module is integrated with a Halcon image processing library and is configured with a VarThreshold operator dynamic threshold segmentation unit, an adjustable kernel size corrosion filtering unit and a light and shade separation unit, and an Erosion Rectangle1 operator is used for corrosion filtering; the AI defect detection module is internally provided with an improved YOLOv8 neural network model, and the model structure comprises a Backbone network embedded with a CBAM attention mechanism, a BiFPN network bidirectional feature pyramid and a dynamic loss function optimizer; according to the technical scheme, the limitation problems of low positioning precision, poor detection efficiency, single defect identification dimension and the like of an artificial vision or traditional probe contact detection means can be solved.

The invention provides a multiband photoresponse nerve synapse device based on a two-dimensional and organic heterostructure, an array and application, and belongs to the technical field of organic electric solid-state devices. The device provided by the invention adopts a three-end bottom gate top contact structure and comprises a gate substrate, an insulating layer, a channel layer and a source / drain electrode, and the channel layer forms an II-type heterojunction by a two-dimensional semiconductor material with a narrow-band gap characteristic and an organic semiconductor material. The heterostructure realizes ultraviolet to far infrared (365nm-10 [mu] m) multiband photoelectric response, and shows synaptic plasticity under light pulse modulation, including short-term enhancement, long-term enhancement and learning-forgetting-re-learning characteristics. Large-area array preparation of the device can be realized through methods such as chemical vapor deposition, solution spin coating, jet printing or layer-by-layer self-assembly. The technology not only can be used for brain-like learning and memory simulation, but also can be applied to the fields of sensing and memory integrated intelligent systems, optical neural networks, artificial vision, multi-mode intelligent sensing and the like.

The invention relates to the field of computer vision and deep learning, and particularly provides a cable low-voltage sheath surface small-size defect detection method and device.Firstly, in the target detection stage, the original structure of a YOLOv5 algorithm is three detection heads P3 / P4 / P5 which represent the detection capacity of the detection heads for small scales, medium scales and large scales from top to bottom; then, in an image classification stage, particle small-size pictures in target detection data are cut, an algorithm is input for training, and defect features are enhanced; compared with the prior art, machine vision can be used for replacing manual vision quality inspection; the omission ratio and the false detection ratio are reduced; the detection efficiency and the production safety are improved; and the production cost and the dependence on external resources are reduced.

The invention discloses a full-light-operated memristor based on an oxide / quantum dot composite structure and a method, and belongs to the technical field of optoelectronic devices. The invention discloses a structure of a full-light-operated memristor. The structure sequentially comprises a substrate, a lower electrode layer, an oxide / quantum dot composite structure layer and an upper electrode layer from bottom to top, the oxide / quantum dot composite structure layer is composed of an oxide layer and a quantum dot layer. According to the full-light-operated memristor disclosed by the invention, the oxide is combined with the high-light-sensitivity quantum dots to form II-type, Z-type and S-type heterojunctions, and photo-induced electrons in the heterojunctions can realize directional transfer from the quantum dots to the oxide under illumination, so that the effective working window of the device is improved. Compared with an existing full-light-control memristor, the composite structure shows a larger working window, the full-light-control performance of the device is remarkably improved, and practical application of the full-light-control memristor in the fields of neuromorphic calculation and artificial vision is facilitated.

A system for clinical analysis by an analytical chip with an actuator device which, in turn, includes a computer module integrating all the signals received by the system and transforming them into analytical results, connectable through a connectivity module with a data processing module connected to an external large-scale clinical data system to generate patterns of diagnoses. A fluid drive module for the impulsion of fluids inside the analytical chip. A thermal module distinguishes thermal areas of the analytical chip. An artificial vision module of the processes is inside the analytical chip. An electronic interface communicates the sensors and actuators integrated in the analytical chip and the actuators of the actuator device with the computer module. A power management module to autonomously power the portable system inside an external container that protects all the previous modules of the actuator device and where the analytical chip is inserted.

The invention relates to an artificial visual nervous system based on a bimodal neural device, which comprises a bimodal visual information calculation array composed of a physical convolution kernel array and a synaptic calculation array, and is characterized in that the physical convolution kernel array is used for receiving an ambient light signal and converting the ambient light signal into an electric signal for output; the synaptic calculation array is used for receiving the electric signals and converting the electric signals into visual signals, and the physical convolution kernel array and the synaptic calculation array are each composed of a plurality of bimodal transistors. The method has excellent bimodal cooperative computing capability, and realizes image feature extraction and preprocessing by simulating a receptive field mechanism (excitement / inhibitory response) of the retina bipolar cells through the physical convolution kernel array; the synaptic plasticity (LTP / LTD mechanism) of a central nervous system is simulated in combination with a synaptic calculation array, advanced calculation of visual signals is completed, and a complete optic nerve bionic system is formed by means of the array composed of single devices.

Method and integrated system for automation of discharge to a mining crusher that includes the analysis of oversize and crushable objects from the extraction truck, and that tracks by means of images associated with a trained convolutional neural network, the oversize and crushable objects as they fall and move through the crusher bowl, analyzing the material flow conditions in the crusher, ordering the start of the crusher hammer in the bowl when there is an oversized material, such as a rock, stopping the discharge from the truck automatically and when it is released and the flow is resumed in the crusher, it generates the order for a truck to start or continue the discharge to the crusher, achieving an auto- discharge system to the crusher. The analysis of the crushable objects from the extraction truck is carried out through artificial vision comprising a deep learning tool, for detection of non-crushable and oversize during the fall of the material into the crusher bowl (or discharge chute).

The invention relates to a method for optimizing carrier separation based on ion irradiation induced positive / negative photoconduction, which comprises the following steps: uniformly irradiating a single crystal sample by using an Xe35 + ion beam with the energy of 500-520MeV at room temperature, and controlling the dosage range of the ion beam to be 4.0 * 10 < 11 > cm <-2 >-3.0 * 10 < 12 > cm <-2 >. Defects with specific concentrations and types are accurately introduced into r-TiO2 crystal lattices, so that the on-demand design of the photoelectric property of the material is realized. A brand-new material preparation approach and a device prototype are provided for developing a new generation of sensing and computing integrated artificial vision sensor, a low-power-consumption neuromorphic computing chip and the like.

The object of the invention is a lens, in particular for artificial vision, comprising, along an optical axis (X):-a front optical unit having a positive power; -an intermediate focal plane on which the light from the front optical unit is focused; -an intermediate optical unit having a positive power, arranged behind the intermediate focal plane; and a rear optical unit having a positive power, including an aperture stop of the lens, and receiving the light from the intermediate optical unit. The entrance pupil of the lens is arranged in the front optical unit.

The utility model discloses a beef quality detection system using machine vision to replace artificial vision, and relates to the beef quality detection field, the beef quality detection system comprises a sampling unit, a controller, an execution unit and a conveying unit, the sampling unit comprises an image acquisition unit and a meat quality acquisition unit, the sampling unit is connected with the controller, and the controller is connected with the execution unit. The controller is connected with the execution unit, in the actual operation of the device, beef is collected and sampled through the sampling unit, a collected signal is transmitted to the controller, the controller transmits the signal to the analysis unit, the analysis unit compares the collected signal with data in a database, and the data in the database is analyzed. The method has more remarkable advantages from the algorithm to the detection precision, photographing and sampling detection are carried out at the same time, the quality of the beef is proved from the two aspects together, and the beef rating is more basis.

The invention discloses a method and system for removing bubbles in cement paste, electronic equipment and a storage medium, and relates to the technical field of cement paste bubble removal. Whether bubbles exist in the cement slurry in the slurry mixing tank or not is judged according to a cement slurry density measurement value obtained through measurement of the electronic densimeter, or whether bubbles exist in the cement slurry in the slurry mixing tank or not is determined in an artificial visual judgment mode, and a judgment result is obtained; when the judgment result is yes, an ultrasonic generator arranged on the slurry mixing tank is controlled to emit ultrasonic waves, and bubbles in the cement slurry are destroyed through the emitted ultrasonic waves. According to the invention, bubbles are removed through a physical means, namely ultrasonic waves, so that the purpose of replacing a defoaming agent is achieved, the bubble removal effect is good, and the method is suitable for cement paste for oil well cementation.

The invention discloses a neural retina synaptic device for artificial vision and a preparation method thereof. The neural retina synaptic device includes: a substrate; the electrode consists of a gate electrode, a source electrode and a drain electrode and is formed on the substrate; the channel layer is formed by a zero-dimensional quantum dot / two-dimensional material heterojunction thin film, has enhanced response to optical signals, is formed on the substrate, is used for connecting the source electrode and the drain electrode to form an electrical path, and is far away from the gate electrode; and the gate dielectric layer is made of an electrolyte gate dielectric material and is connected with the channel and the gate electrode to form a double-electrode layer. The gate electrode receives an electric signal to simulate a synaptic front end, the channel layer receives an optical signal to simulate a retinal cell light sensor, the source electrode, the drain electrode and the channel layer complete information collection, processing and storage by reading voltage, and the electric synaptic front end is adjusted to obtain positive and negative light current response. The method is used for constructing a sensing, storage and calculation integrated visual neuromorphic sensing system.

The invention discloses a cortex visual prosthesis control method and system based on reinforcement learning and behavior feedback, and the method comprises the steps: obtaining a visual image of an external environment, and extracting a feature vector of the visual image; inputting the feature vector into a reinforcement learning environment to generate an electrical stimulation pulse sequence corresponding to a plurality of microelectrode channels; applying the sequence of electrical stimulation pulses to a microelectrode array implanted in the visual cortex of a patient to induce light magical perception; behavior feedback generated when the patient executes a visual recognition task based on the light magical visual perception is obtained; and updating parameters of the reinforcement learning environment based on the behavior feedback to form closed-loop optimization. The stimulation strategy can be evolved on line, the cortical function recombination state of an individual caused by long-term blindness is automatically adapted, the morphological drift of light magical vision along with time can be effectively dealt with, and therefore the stability and long-term availability of artificial visual perception are remarkably improved.

The invention particularly relates to a carton printing color difference defect detection method based on artificial intelligence and image recognition, and relates to the technical field of industrial detection. Acquiring multi-mode active light field data; geometric and reflection decoupling calculation based on a physical model; and calculating the intrinsic chromatic aberration and judging the sensing fusion. According to the method, phase calculation and three-dimensional reconstruction are carried out to extract a surface normal vector, a mixed reflection model is substituted to solve an intrinsic diffuse reflection color, shadow stripping, highlight and other geometric interferences are solved, the extracted intrinsic color truly reflects printing ink characteristics, color difference calculation of a CIELAB color space is combined, high, medium and low saliency regions are divided according to region attention, and a differentiation threshold value is set. And a weighted chromatic aberration mean value is calculated for judgment, so that the deviation between traditional single threshold judgment and artificial visual evaluation is reduced, anthropomorphic quality judgment is realized, and the detection precision and the result reliability are greatly improved.

The invention discloses a pin winding mechanism of a pin winding machine, which comprises a pin winding vertical moving module, an upper pin winding plate and a lower pin winding plate, the driving end of the pin winding vertical moving module is connected with a pin winding mounting frame, the upper end of the pin winding mounting frame is connected with an upper driving assembly, the upper pin winding plate is mounted on the upper driving assembly, and the lower end of the pin winding mounting frame is connected with a lower driving assembly. The lower foot winding plate is installed on the lower driving assembly, an upper foot winding motor is fixed to the upper foot winding plate, an upper winding needle is rotationally connected to the upper foot winding plate, the upper foot winding motor drives the upper winding needle to rotate, a lower foot winding motor is fixed to the lower foot winding plate, a lower winding needle is rotationally connected to the lower foot winding plate, and the lower foot winding motor drives the lower winding needle to rotate. According to the pin winding mechanism of the pin winding machine, the movable pin winding plate drives one set of upper winding pins and the other set of lower winding pins to be matched with the position of a pin column on a current pin winding jig, the requirement for a combined structure of thick copper wire winding and thin copper wire winding is met, meanwhile, automatic high-low adjustment, left-right adjustment and front-back adjustment are achieved, and the precision is higher than that of artificial vision.

A simple and efficient PCB surface crack detection device comprises a feeding assembly, a roller coating assembly, a detection assembly and an intelligent detection system. The feeding assembly comprises a machining table and a bearing tray. The roller coating assembly comprises a roller coating frame, a coating roller and a fluorescent ink box; the detection assembly comprises a detection frame, a camera obscura and a visual detection camera. The simple and efficient PCB surface crack detection device disclosed by the utility model has the beneficial effects that the fluorescent roller coating structure is designed to be matched with the feeding structure and the visual detection structure, so that the detection of the PCB surface crack is simple and efficient, and compared with artificial visual detection, fluorescent marking and machine visual detection are more accurate and reliable; a cleaning structure and a drying structure are designed, so that the PCB coated with fluorescent ink can be quickly cleaned and dried, and the detected PCB can better flow into a subsequent processing flow; the utility model has the advantages of strong practicability and strong popularization significance.

The invention belongs to the field of semiconductor photoelectronic devices, discloses a two-dimensional niobium oxyhalide intrinsic type artificial visual synapse device and a preparation method thereof, and particularly relates to application of an intrinsic type two-dimensional NbOX2 material as a channel material in the artificial visual synapse device, the intrinsic type two-dimensional NbOX2 material is arranged along the surface of a substrate, and the intrinsic type two-dimensional NbOX2 material is arranged along the surface of the substrate. One end of the electrode covering the area between the source electrode and the drain electrode covers the source electrode, and the other end of the electrode covering the drain electrode; the source electrode and the drain electrode are made of two-dimensional Van der Waals semimetal; the edge, close to the drain electrode, of the source electrode and the edge, close to the source electrode, of the drain electrode are parallel to each other and are both perpendicular to the crystallographic c-axis of the intrinsic two-dimensional NbOX2 material. According to the invention, the independent intrinsic two-dimensional NbOX2 material is selected as the channel material, destructive functionalization processing does not need to be carried out on the channel material, intervention of an additional charge trapping layer is not needed, and the artificial visual synapse device constructed based on the intrinsic two-dimensional NbOX2 material can well simulate synapse behaviors.

A method for moving workpieces (50, 51, 52) to be machined and / or machined pieces (60, 61, 62) in a robotized work island (300), comprising an operating machine (250) for working sheet metals and a robot (100) for moving the workpieces (50, 51, 52) and machined pieces (60, 61, 62) and having an operating end (101) provided with gripping means (110) and an artificial vision apparatus (1), comprises the steps of positioning the operating end (101) above a first supporting plane (261) to acquire by means of a camera (2) of the vision apparatus (1) a first three-dimensional image of the first supporting plane (261) and of workpieces (50, 51, 52) and / or stacks of workpieces (55); processing the first three-dimensional image to determine position and orientation of the first supporting plane (261), number, geometry, position and orientation of each workpiece (50, 51, 52) and / or of each stack of workpieces (55, 56); identifying one or more workpieces (50, 51, 52) to be picked up and comparing its geometry with respective reference geometry to ascertain whether the workpieces (50, 51, 52) are compliant; picking up the workpiece (50, 51, 52) if compliant and moving it towards and a worktable (260) of the operating machine; releasing the workpiece (50, 51, 52) onto the worktable (260) so as to be machined and obtaining a corresponding machined piece (60, 61, 62); picking up the machined piece (60, 61, 62) from the worktable (260); arranging the operating end (101) with the vision apparatus (1) on the second supporting plane (262) to acquire a second three-dimensional image of the second supporting plane (261) and any machined pieces (60, 61, 62) and / or stacks of machined pieces (65, 66); processing the second image so as to determine position and orientation of the second supporting plane (262), number, position and orientation of any machined pieces (60, 61, 62) and / or stacks of machined pieces (65, 66); positioning and releasing the machined piece (60, 61, 62) on the second supporting plane (262).

The invention discloses a visual reconstruction method and a related product. The method comprises the following steps: acquiring image data of a target image; mapping the coordinate of each visual pixel in the image data and a visual cortex to obtain an array matrix; the array matrix comprises a mapping relation between each visual pixel and a visual cortex; based on the sound focus position corresponding to each visual pixel in the image data, obtaining a driving parameter of each array element in the stimulation equipment; processing the array matrix based on the driving parameter of each array element to obtain a driving sequence; and controlling stimulation equipment based on the driving sequence, and stimulating the visual cortex to present the target image. The defect that traditional single-point or uniform scanning stimulation lacks space coding capacity is effectively overcome, the spatial resolution and perception consistency of artificial vision are remarkably improved through topological mapping from an image to a cortex layer and accurate regulation and control of a sound field, and meanwhile long-term foreign matter reaction is avoided.

Device and method to determine the quality of a meat product comprising an identifying element that identifies said meat product by means of the emission and / or reception of an identifying signal of the meat product, said identifying signal comprises at least predetermined puncture requirements and / or a puncture depth (zi) in the meat product, positioning means configured to displace a head, where said head is disposed at one end of said positioning means, and comprises an insertion element configured to be inserted inside said meat product at a predetermined puncture depth (zi), and measure at least a pH signal in a predetermined measurement point (xi, yi, zi), the puncture point (xi, yi) being predetermined in accordance with a prior analysis of the anatomy of said meat product by means of an artificial vision device that determines the puncture point (xi, yi) in accordance with the predetermined puncture requirements and anatomy of the meat product.

Embodiments herein provide a method and system for a hyperspectral artificial vision for machines. The system receives a hyperspectral signal of a target material as an input to a neural network model. The system initializes by selecting the number of primitive layers to be used. The system iteratively cycles through all training data (pixels) and updating weights for each unsuccessful material class prediction. Model with two primitives serves as baseline, after which the system adds another primitive layer and repeats the training procedure. The system keeps repeating these processes until obtains convergence. Where the system come to a halt, the system obtains the optimal number of primitives for the given materials. The generated new color pixel is used as a discriminator to aid in locating the target material. The new artificial color is a mixture of weighted chromatic primitives which are optimized for sensitivity / (Spectral Response Functions) SRFs.

A driving shaft three-station assembly detection device comprises a workbench, the workbench is provided with a first workbench and a second workbench which are symmetrically arranged front and back, the first workbench and the second workbench are each provided with a driving shaft assembly device, each driving shaft assembly device comprises a driving shaft supporting and rotating device, and the periphery of a driving shaft is connected with a rubber protection sleeve. The rubber protective sleeve is matched with the hoop locking device; a T-shaped support is arranged between the first workbench and the second workbench, and the two sides of the T-shaped support are connected with hoop locking devices through first guide rails. A third working table is arranged on the side edge of the first working table and the side edge of the second working table, and a driving shaft laser marking device is installed on the third working table. According to the utility model, the hoop locking function of the rubber protective sleeve and the manual visual inspection of flaws in the circumferential direction of the driving shaft are completed through the two driving shaft assembling devices, the marking function is completed in combination with the driving shaft laser marking device, and the driving shaft assembling and marking efficiency is improved under reasonable budget through the semi-automatic process design.

The utility model relates to test screen detection technical field, concretely is test screen full -automatic calibration device, including detection bin, the inside installation of detection bin has operating table, the both sides installation of operating table has clamping subassembly, the inside installation of detection bin has detection subassembly, the bottom installation of operating table has adjusting unit, the outside of detection bin is provided with mechanical arm unit, one side of mechanical arm unit is provided with storage bin, the clamping subassembly includes clamping seat, and clamping seat fixedly connected in the both sides of operating table, the one end of clamping seat is installed with clamping motor, the one end drive coupling of clamping motor has two -way screw shaft. The test screen full -automatic calibration device of improvement, the mode of artificial vision and contact type detection calibration is set to mechanical arm and is grabbed test screen to detection station, and after detection is finished, the mode of taking and placing test screen is carried out through mechanical arm, avoids the damage of artificial contact to screen net, and then improves the precision of test screen detection.

The object of the present invention is a lens, in particular for artificial vision, comprising, along an optical axis (X) : - a front optical unit with positive optical power; - an intermediate focal plane whereupon optical rays from the front optical unit are focused; - an intermediate optical unit with positive optical power, arranged behind the intermediate focal plane; - a rear optical unit with positive optical power that contains the stop aperture of the lens and receives optical rays from the intermediate optical unit. The entrance pupil of the lens is arranged in the front optical unit.

An artificial vision device, according to one embodiment of the present invention, comprises: an image generation device configured to generate an image corresponding to an object; a stimulation control device configured to divide the image into a plurality of regions and generate stimulation current information on the basis of the plurality of regions; and a stimulation generation device comprising a plurality of electrodes and configured to apply a current to the plurality of electrodes on the basis of the stimulation current information. The stimulation control device comprises: an object type classification unit configured to generate object type information by classifying the type of the object of the image; a feature extraction unit configured to generate a feature extraction image by dividing the image into the plurality of regions on the basis of the image and the object type information; and a stimulation pattern generation unit configured to generate the stimulation current information which includes information on the magnitudes of the current corresponding to the respective plurality of regions, wherein the magnitudes of the current corresponding to the respective plurality of regions are different from each other.

The invention relates to the technical field of artificial vision and neuromorphic devices, and particularly discloses a retina-like adaptive vision device based on a perovskite and hydrogen bond organic framework material heterostructure and a preparation method of the retina-like adaptive vision device. According to the device, a hydrogen bond organic framework is adopted to regulate and control perovskite crystal lattices, carboxyl and lead ion coordination are utilized to induce crystal lattice strain, defect energy barriers are effectively improved, non-radiative recombination of carriers is inhibited, and dynamic regulation and control of photon-generated carriers are achieved. The device can realize the functions of light adaptation and dark adaptation driven by light intensity without voltage modulation, and shows the Weber's law response characteristic similar to the retina of human eyes. The device shows ultra-low monopulse power consumption, wide spectral response range, response speed close to human eyes and high dynamic range, is obviously superior to a traditional CMOS sensor, and provides a new way for high dynamic range imaging and on-chip neuromorphic calculation.

The invention provides a tellurium oxide memristor and preparation and application thereof, and belongs to the technical field of semiconductor device application, a thermal evaporation method is used for preparing a TeOx film, and a memristor of an A / TeOx / A structure is constructed by a magnetron sputtering method for depositing an electrode; according to the memristor disclosed by the invention, synaptic plasticity and OR and AND logic gates are realized under light and electrical stimulation; under the modulation of light pulses with different wavelengths, the memristor can be used for biological synaptic plasticity simulation, such as associative learning behaviors, Pavlovian conditioned reflex and the process of'melatonin secretion and suppression 'in the brain; besides, the short-term plasticity and the long-term plasticity displayed by the device can simulate the function of the visual neural network in the aspect of identifying and memorizing noise images, the synaptic plasticity function is further combined with the artificial neural network (ANN), the device can achieve accurate image identification and classification, the accuracy rate reaches 92.4%, and the device is suitable for popularization and application. Therefore, the photoelectric device based on TeOx has huge potential in the field of artificial vision application.

In artificial vision parameter automating method, the system includes a sensory module, a communication module, a database, an image processing module, a learning module and a automating and storing module. The sensory module is configured to take images. The communication module is configured to be send the data obtained from the sensory module to the database. The database is configured to store the images received from the communication module. The image processing module is configured to convert a selected image from the database into a low-resolution phosphene image, to generate test images with various artificial vision parameters. The learning module is configured to extract important features from high-resolution training images, and to build various machine learning models competent at identifying visual stimuli from low-resolution phosphene images. The automating and storing module is configured to automatically select and store primary test data, based on the learning in the learning module.