Abstract: A heat exchanger processing method, in which a heat exchanger includes a plurality of micro-channel flat tubes, and the heat exchanger processing method includes: providing a plurality of micro-channel flat tubes, in which each micro-channel flat tube includes a plurality of channels extending in a length direction of the micro-channel flat tube, and the plurality of micro-channel flat tubes are spaced in a first direction; moving a collection assembly starting from an initial position along the first direction to collect specific information of one or more micro-channel flat tubes one by one, and transmitting the specific information to a control assembly; comparing, by the control assembly, the specific information with pre-stored information and generating, based on a comparison result, feedback information; generating, by an instruction assembly based on the feedback information, instruction information; and adjusting a position of a corresponding micro-channel flat tube based on the instruction informati

Abstract: A machine-vision system for monitoring a quality metric for a product. The system includes a controller configured to receive a digital image from an image acquisition device. The controller is also configured to analyze the digital image using a first machine-vision algorithm to compute a measurement of the product. The system also includes a vision server connected to the controller, and configured to compute a quality metric and store the digital image and the measurement in a database storage. The system also includes a remote terminal connected to the vision server, and configured to display the digital image and the quality metric on the remote terminal.

Abstract: A defect inspection apparatus (100) is configured to approximate a difference value or an absolute value (I?) of the difference value between a pixel value in at least three captured images (A) captured by an imager in at least three different phases of an elastic wave and a pixel value in a reference image (Aave) separate from the captured images (A) so as to acquire an approximate value for defect inspection corresponding to an amount of change in the pixel value in the captured images (A).

Abstract: An inspection device according to one aspect of the present disclosure includes an image capturing unit configured to capture an image of an EUV mask provided with a pattern, a storage unit configured to store a database intermediate file including a gray image obtained by pixelating a binarized image rasterized from design data of the pattern, and a processing unit configured to inspect the EUV mask on the basis of a captured image obtained by the image capturing unit capturing an image of the EUV mask. The processing unit includes a conversion model generated by a learning machine configured to perform learning by deep learning, a reference image generation unit configured to generate a reference image from the gray image by using the conversion model, and a comparison unit configured to compare the reference image with the captured image.

Abstract: A method for monitoring welding of an electric resistance welded steel pipe of the present invention includes: an image acquisition step in which, when an electric resistance welded steel pipe is welded, a V-convergence portion including an edge detection area and a V-convergence point, and a welding portion including a position where a molten steel starts to be discharged from an inside of a wall thickness and a bead, are imaged to obtain a captured image; a determination information acquisition step in which the captured image is image-processed, and a molten length from the V-convergence point to the position where the molten steel starts to be discharged or a molten width of the welding portion or a combination thereof is acquired as a determination information; and a determining step in which it is determined whether a cold welding has occurred by comparing the determination information with a determination threshold value set in advance for each wall thickness of the electric resistance welded steel pip

Abstract: A teacher data generating method includes: causing a generation model to perform learning by using a learning image in which boundary information indicating in a chromatic color a range of a defect is superimposed on a defective product image in a gray scale so as to generate a generated defect image including a new image of a defect in the gray scale and an image of the boundary information in the chromatic color; generating the generated defect image by using the generation model; extracting a pixel having a pixel value corresponding to the chromatic color from the generated defect image, extracting the boundary information corresponding to the generated defect image, and acquiring a gray scale defect image without including an image of the boundary information; calculating a coordinate of the boundary information; and associating the gray scale defect image with the coordinate to obtain defective product teacher data.

Abstract: A method for use with a subterranean well can include positioning an x-ray unit so that x-rays emitted by the x-ray unit scan a tubular string, displacing the tubular string relative to the x-ray unit, and identifying a threaded connection in the tubular string. A system can include a torque application device configured to apply torque to a threaded connection in a tubular string, and an x-ray unit configured to project x-rays toward the tubular string.

Abstract: A computer-implemented method of detecting distracted driving comprises: determining, by one or more processors, a primary preview region (PPR) in a representation of an environment; determining, by the one or more processors, a gaze point for a driver based on a sequence of images of the driver; determining, by the one or more processors, that the gaze point is outside of the PPR; based on the determined gaze point being outside of the PPR, decreasing, by the one or more processors, an attention level for the PPR; based on the attention level for the PPR, generating, by the one or more processors, an alert.

Abstract: Provided is an image and an inspection system that make it possible to effectively utilize a result of past inspection. An audit position indicating a position where an audit is planned or a position where an audit is completed and a past audit position indicating a position where an audit was performed in the past within a predetermined range including the audit position are displayed over a map on a user terminal, the map being a drawing representing a configuration of a manufacturing site. The inspector references the map on the user terminal and effectively utilizes a past audit result to perform an audit.

Abstract: A computer system performs corrective operations during a three-dimensional (3D) printing job. The 3D printing job is initiated to produce a physical object. Printing of the physical object is monitored to identify a defect of the physical object during the printing. The defect is analyzed to determine one or more corrective operations to apply to the physical object. The one or more corrective operations are performed during the printing. Embodiments of the present invention further include a method and program product for performing corrective operations during a 3D printing job model in substantially the same manner described above.

Abstract: A method and system for tracking attached objects for controlling a vehicle. The method includes receiving, by a processor, sensor data of an environment associated with the vehicle that includes an object. The method includes determining, based on the sensor data, whether the object corresponds to an attached object type, and determining whether the sensor data includes a second vehicle. The method includes determining an attached object region associated with the second vehicle based on a distance between the second vehicle and the vehicle, and determining whether a position of the object is within the attached object region. The method includes determining whether a difference between a velocity of the object and a vehicle velocity of the vehicle is within a velocity threshold based on the distance. The method includes determining that the object is an attached object and controlling the vehicle based on the determining.

Abstract: Provided is a method for monitoring additive manufacturing. The method comprises acquiring an image of a three-dimensional object that is being constructed using an object model. The method further includes isolating a layer of the three-dimensional object from the image. The method further includes generating a three-dimensional model from the layer. The method further includes comparing the three-dimensional model with the object model. The method further includes identifying a discrepancy between the three-dimensional model and the object model. The method further includes providing a notification of the identified discrepancy.

Abstract: Method and device for detecting and classifying a not-known object by a device having a two-dimensional field of view. The method includes a learning step, a detecting step and an inspection step. The learning step includes positioning a first known object in a field of view of the device and then positioning a second known object in the field of view of the device, to obtain images of the first and second known objects. Distinctive characteristics and common characteristics of the objects are identified and used to obtain an equality descriptor relative to common elements of the first and second known objects and a difference descriptor allowing association of the first and second known objects to two distinct classes. The device is then used to detect a third, not-known object. The equality descriptor and the difference descriptor are then used in an inspection step to classify the third not-known object.

Abstract: Provided is a dual-neighborhood wear angle automatic detection method, including constructing a proximal neighborhood and establishing grayscale differences between a pixel in a center area and pixels having first directions in the proximal neighborhood; calculating, based on the grayscale differences, deviation degrees of all pixels in the center area in the first directions; calculating a similar direction value based on the deviation degrees; constructing a distal neighborhood and focus pixels having second directions in the distal neighborhood; calculating grayscale differences between the pixel and the focus pixels; calculating deviation degrees of all pixels in the center area, in the second directions based on the grayscale differences between the pixel and the focus pixels; calculating a direction value of a wear angle based on the deviation degrees in the second directions; and calculating the wear angle based on the direction value of the wear angle.

Abstract: An apparatus includes one or more light source groups, a camera; one or more memories configured to store one or more first images acquired in advance, and one or more processors communicatively connected to the one or more light source groups, the camera, and the one or more memories, wherein the one or more processors are configured to acquire a second image by controlling a first light source group among the one or more light source groups and the camera and irradiating light beams to a substrate on which an object is coated, and determine a coating state of the object based on the one or more first images and the second image.

Abstract: When a machine learning network trained with both a non-defective product image and a defective product image is used, it is possible to stably exhibit high detection capability for the defective product image having an unknown defect while shortening a takt time during the operation time. A processor executes a first learning process of causing a machine learning network to learn a non-defective product image added with a noise, and a second learning process of causing the machine learning network to learn a defective product image, and detects both an unknown defect having a characteristic different from a characteristic of the non-defective product image and a known defect having a characteristic designated as a defective site, by inputting a workpiece image to the machine learning network whose parameter has been adjusted by the first learning process and the second learning process.

Abstract: An adaptable inspection and sorting unit includes an attachment mechanism, an inspection sensor device that is capable of inspecting a sample, a sorting device that is capable of deflecting the sample, a data port that is capable of receiving information, and a power port that is connectable to a power source. The attachment mechanism, the inspection device, the sorting device, the data port, and the power port are physically connected together. A memory circuit and a processor circuit read information received via the data port, control the inspection sensor device, and control the sorting device. The adaptable inspection and sorting unit is capable of inspecting and deflecting a sample traveling along the processing line. The speed of the sample is measured utilized to determine a delay between inspecting the sample and sorting the sample.

Abstract: An apparatus for processing a consolidated stack of fiber reinforced plies, comprising an inspection device, in particular an optical inspection device, arranged to inspect a consolidation state of the stack of fiber reinforced plies, and a handling device arranged to make at least an outer surface of the stack of fiber reinforced plies available for inspection by the inspection device.

Abstract: According to some embodiments, a system includes: a memory, an encoder; a decoder, wherein the system is operable to: receive, at the encoder, an input video; divide, by the encoder, the input video into a plurality of video patches; select, by the encoder, codes corresponding to the plurality of video patches of the input video, from a codebook comprising the codes; determine, by the encoder, an assigned code matrix comprising the codes corresponding to the plurality of video patches of the input video; receive, by the decoder, the assigned code matrix from the encoder; and generate, by the decoder, a reconstructed video based on the assigned code matrix.

Abstract: An image processing apparatus is configured to determine presence or absence of print abnormality in a target image by comparing a reference image or reference images, each of which is a print result to be a reference, and the target image, which is a print result to be an inspection target. The image processing apparatus includes a setting unit configured to set an inspection condition for a printed product, and a comparison unit configured to compare the reference image or reference images and the target image using a number of the reference images set based on the inspection condition.

Abstract: A method for detecting defects in products from images thereof and an electronic device applying the method inputs a defect image repair data set into an autoencoder to train the autoencoder, and generates a reconstructed image, calculates a reference error value between the sample image and the reconstructed image by a preset error function, and set a threshold value based on the reference error value. The electronic device inputs an image possibly revealing a defect into the autoencoder and generates the reconstructed image corresponding to the image to be detected, and uses the preset error function to calculate the reconstruction error between the image and the reconstructed image, thereby determining whether the image being analyzed does reveal defects. When the reconstruction error is greater than the threshold value, a determination is made that a defect is revealed.

Abstract: A method of classification of a film non-uniformity on a substrate includes obtaining a color image of a substrate with the color image comprising a plurality of color channels, obtaining a standard color for the color image of the substrate, for each respective pixel along a path in the color image determining a difference vector between the a color of the respective pixel and the standard color to generate a sequence of difference vectors, sorting the pixels along the path into a plurality of regions including at least one normal region and at least one abnormal region based on the sequence of difference vectors, and classifying the at least one abnormal region as overpolished or underpolished based on at least one difference vector of a pixel at a boundary between the abnormal region and an adjacent normal region.

Abstract: In a case where a three-dimensional model of a construction stored in a storage is to be displayed on a display, an information display apparatus 10 accepts a view operation including an operation for enlarging the three-dimensional model from an operation interface and at least enlarges and displays the three-dimensional model. After a desired position on the three-dimensional model enlarged and displayed on the display is specified in accordance with a user operation, the information display apparatus identifies a three-dimensional position on the three-dimensional model corresponding to the position and identifies a management segment that is an inspection unit of the construction on the basis of the identified three-dimensional position. After the management segment is thus identified, the information display apparatus reads management information corresponding to the identified management segment from the storage and displays the management information on the display.

Abstract: A computer implemented method is disclosed herein for monitoring and determining a quality level of incoming raw material from one or more sources. The method includes (1) receiving visual data associated with the incoming raw material; (2) determining an indication of quality level associated with the incoming raw material; and (3) transmitting, to at least one of a graphical user interface (GUI) and a computer log, the indication of quality level and at least one timestamp associated with the visual data. The visual data may include a plurality of images received from one or more cameras configured for monitoring the incoming raw material. A related system is also disclosed herein.

Abstract: In an approach to non-destructive inspection of IC devices, one or more images of a Device Under Test (DUT) are received from one or more imaging devices. Observed features are detected in the one or more images and producing a first synthetic representation of a part design of the DUT that includes the observed features. The presence of one or more first unobserved features are inferred, where the one or more first unobserved features are inferred using a mapping and inference model (MIM). The one or more first unobserved features are added to the first synthetic representation of the part design of the DUT.

Abstract: A vehicle comprising a track system can be monitored to obtain information regarding the vehicle, including information regarding the track system, such as an indication of a physical state of a track and/or other component of the track system based on at least on 3D recognition and/or 2D recognition of image data of a track system component.

Abstract: A collision performance evaluation test with few variations in test results with high accuracy in which a complicated phenomenon that actually occurs can be reproduced in a simple manner by considering the history of deformation in both the press forming and a collision. A collision performance evaluation test method for a metal sheet material for an automobile body is characterized in that a press working apparatus first forms a flat test specimen made of a metal sheet material to be tested into a V shape by primary bending, a bending test apparatus then deforms, by secondary bending, the test specimen formed by the primary bending in a direction intersecting with the primary bending direction, and a bending load and a bending stroke for the test specimen during the secondary bending deformation are recorded and evaluated.

Abstract: An information processing device according to the present disclosure including: a mesh division unit (102) that divides at least part of first and second images captured at different times of a target area for detecting chips generated from a workpiece, into a plurality of mesh regions, the images each being an inside image of a machine tool; and an information processing unit (103) that performs processing to associate (a) information on a first chip corresponding to a specific mesh region among the plurality of the mesh regions corresponding to the first image, (b) information on a second chip corresponding to the specific mesh region among the plurality of the mesh regions corresponding to the second image, (c) a first time related to the first image, and (d) a second time related to the second image.

Abstract: A wafer defect detection apparatus and a method of fabricating an IC using the same. Images of a plurality of semiconductor wafers forming a wafer lot are captured at a targeted process step of a fabrication flow and preprocessed, wherein a medoid image is identified as a reference wafer image. In one arrangement, preprocessed wafer images of a semiconductor wafer lot may be analyzed for defects based on an ensemble of image analysis techniques using at least one of the reference wafer image from the wafer lot and a template patch to enhance the predictive power of defect detection.

Abstract: A system includes a robotic arm of a robotic device, an imaging device coupled to the robotic arm, and a processor configured to control movement of the robotic arm to move the imaging device along a preprogrammed measurement path of a transparency while prompting the imaging device to record images of localized portions of the transparency. The processor is configured to determine one or more localized transparency characteristics based on an analysis of the images of the localized portions.

Abstract: The present invention proposes a fabric defect detection method based on multi-modal deep learning. First, a tactile sensor is placed onto the fabric surface with different defects to collect the fabric texture images, a camera is used to collect the corresponding fabric external images, and a fabric external image and a fabric texture image constitute a set of fabric detection data; then, a feature extraction network and a multi-modal fusion network are connected to establish a classification model based on multi-modal deep learning, which uses the fabric texture image and fabric external image in each set of collected fabric detection data as input, and the fabric defect as output; said classification model is trained using the collected fabric detection data; finally, the trained classification model is used to detect the fabric defect. The present invention employs vision-touch complementary information, which can greatly improve the accuracy and robustness of detection.

Abstract: A display defect detection system and a detection method thereof are disclosed. A display defect detection system includes a preprocessing circuit receiving a capture image of a test pattern, displayed by a display panel, as a panel image including diagonal Mura and preprocessing the panel image to output a preprocessing image and a Mura detection circuit calculating a final feature value of the diagonal Mura by multiplying a first feature value, in which a luminance difference of the diagonal Mura is reflected, by a second feature value in which a shape ratio of the diagonal Mura is reflected, on the basis of an edge map of the preprocessing image, and detecting a display position of the diagonal Mura on the basis of the final feature value of the diagonal Mura.

Abstract: An image capturing unit captures an image of a surface of a member treated by a surface treatment unit. A comparison unit obtains a correlation coefficient between a reference image stored in a storage unit and the image captured by the image capturing unit. A cross-correlation coefficient between power spectra of the reference image and the image captured by the image capturing unit is obtained. A control unit stops a surface treatment performed by the surface treatment unit when the correlation coefficient obtained by the comparison unit reaches a set value.

Abstract: An example embodiment of the present invention provides a method of assessing the condition of a pavement site, comprising: (a) acquiring aerial images of the site from above, for example by an unmanned aerial system (UAS); (b) using photogrammetry tools to generate an orthomosaic that represents the airport pavement surface; (c) using image analysis tools and machine learning methods to determine the location and extent of defects in the pavement; (c) producing an image representation of the site and the defects, where the location and extent of defects are discernible from the image; (d) using software application techniques to store and present defect data and other related information for client-side user access.

Abstract: A method includes the steps collecting measurement data of a sample utilizing an adaptable inspection unit or while the sample is in-flight, determining a volume or area of the sample based at least in part on the measurement data, and calculating a weight of the sample based at least in part on the volume or area of the sample. The measurement data includes a captured image that includes a plurality of pixels. The determining of the volume of the sample includes determining the number of pixels in the captured image that display a portion of the sample, or determining the maximum number of consecutive pixels that display a portion of the sample in two or three dimensions.

Abstract: An inspection device is such that the inspection device executes a mounting inspection of a component using an inspection image of a board on which the component is mounted and includes an imaging device and an image processing device. The imaging device obtains three monochromatic images of the board in R, G, B. The image processing device sets a board color using brightness values of the three monochromatic images obtained by the imaging device, calculates complementary colors of the board color so set, determines on a color component having a largest brightness value in individual color components of the complementary colors so calculated, and sets the monochromatic image having the color component so determined in the obtained three monochromatic images as the inspection image.

Abstract: The present disclosure includes edge defect detection via image analytics. A method includes identifying an image of an edge of a susceptor pocket formed by a susceptor of a substrate processing system. The method further includes predicting, based on the image, whether property values of the edge of the susceptor meet threshold values. The method further includes, responsive to the property values of the edge meeting threshold values, causing performance of a corrective action associated with the susceptor.

Abstract: A method for specifying a position through touch input in a screen displayed on a touch panel display. The position is specified by acquiring an initial contact position with the touch panel display, displaying a position displaying cursor in a position according to the initial contact position, displaying the position displaying cursor in the position according to the initial contact position during the period for which the contact position sensing continues until the distance from the initial contact position to the contact position reaches a predetermined distance and terminating the display of the position displaying cursor where the contact position sensing is terminated before the distance reaches the predetermined distance, and displaying the position displaying cursor in such a way that the position displaying cursor follows movement of the contact position after the distance reaches the predetermined distance.

Abstract: One variation of a method for predicting manufacturing defects includes: accessing a first set of inspection images of a first set of assembly units recorded by an optical inspection station over a first period of time; generating a first set of vectors representing features extracted from the first set of inspection images; grouping neighboring vectors in a multi-dimensional feature space into a set of vector groups; accessing a second inspection image of a second assembly recorded by the optical inspection station at a second time succeeding the first period of time; detecting a second set of features in the second inspection image; generating a second vector representing the second set of features in the multi-dimensional feature space; and, in response to the second vector deviating from the set of vector groups by more than a threshold difference, flagging the second assembly unit.

Abstract: A method for detecting a display panel defect, including: collecting a panel image of a to-be-detected display panel, a plurality of first pixels of the display panel corresponding to a plurality of second pixels in the panel image; converting the panel image into a binary image; dilating each bright spot region in the binary image such that adjacent bright spot regions communicate with each other to form at least one closed communication region in the binary image; determining a region of interest mask image in the binary image in accordance with the at least one closed communication region; determining a region of interest in accordance with the region of interest mask image and the panel image; and performing feature identification on the region of interest to determine a defect of the display panel.

Abstract: An apparatus for forming and applying artificial fingernails is disclosed herein. In one or more embodiments, the apparatus includes a housing; a rotatable carousel disposed at least partially within the housing, the rotatable carousel including a plurality of fingernail molds that are configured to be rotated with the carousel, the plurality of fingernail molds configured to form a plurality of different fingernail shapes using an artificial fingernail molding material; a dispensing device disposed at least partially within the housing, the dispensing device configured to dispense the artificial fingernail molding material onto a fingertip portion of a user within a selected one of the plurality of fingernail molds; and a curing device, the curing device configured to cure the artificial fingernail molding material on the fingertip portion of the user so as to form an artificial fingernail having a particular shape.

Abstract: An imaging system is for recording images of a raceway of a bearing ring, the ring having two opposing axial ends, a central axis extending between the axial ends, and inner and outer circumferential surfaces. The raceway is formed on the inner circumferential surface or the outer circumferential surface of the ring and has first and second ends spaced along the central axis. The imaging system includes a frame movably coupleable with the bearing ring and having a centerline. A drive is mounted to the frame and is configured to angularly displace the frame about the bearing central axis. An optical imager has a lens and an adjustable positioner is mounted on the frame and is configured to position the optical imager such that the lens is disposed axially between the raceway first and second ends and faces generally perpendicular to the bearing raceway.

Abstract: A system and method are provided for super-resolution imaging beyond the Rayleigh spatial resolution limit of an optical system. In the context of a system, first and second pinhole assemblies are configured to be controllably positioned. The first and second pinhole assemblies define respective pinholes and being configured to be backlit. The system also includes a collimating lens configured to collimate at least a portion of the signals passing through the respective pinholes of the first and second pinhole assemblies. The system further includes an amplitude/phase mask configured to provide amplitude and phase modulation to signals received from the collimating lens and an imaging lens configured to focus the signals received from the amplitude/phase mask upon an image plane to permit objects to be separately identified.

Abstract: An individual identification apparatus includes a setting means for setting an imaging condition for imaging a matching area of an object by a camera, an acquisition means for acquiring a matching image that is an image obtained by imaging the matching area of the object by the camera, a matching means for performing matching between the matching image and a registration image registered in advance, and a control means. The control means repeats processing of acquiring the matching image by the acquisition means and matching between the matching image and the registration image by the matching means, while changing the imaging condition set by the setting means, and determines an imaging condition in which a matching degree between the matching image and the registration image becomes a preset threshold or larger. The matching means identifies an individual based on the matching image captured under the determined imaging condition.

Abstract: Systems and methods are provided for automating property assessment using probable roof loss confidence scores. More particularly, the systems and methods may generate a base-line probable roof loss confidence score based upon buildings, roof, weather, hail, and climate data. The systems and methods may predict a specific event and a set of characteristics of the specific event. The systems and methods may predict a level of roof damage and a cost of roof damage based upon the predicted level of roof damage. The systems and methods may adjust a risk-related factor, and may further adjust policy parameters based upon the risk-related factor. A probable roof loss confidence score may be generated, as well as property insurance claims data or property insurance loss mitigation data based upon probable roof loss confidence score data.

Abstract: A computer system is provided for converting images through use of a trained neural network. A source image is divided into blocks and context data is added to each pixel block. The context blocks are split into channels and each channel from the same context block is added to the same activation matrix. The action matrix is then executed against a trained neural network to produce a changed activation matrix. The changed activation matrix is then used to generate a converted image.

Abstract: The present application discloses a method and device for controlling misalignment of electrode plates, electrode plates, an electrochemical cell, and a battery. The method includes: acquiring cut hole position information of an anode electrode plate, and calculating a first width deviation value corresponding to the anode electrode plate according to the cut hole position information and a calibration width of the anode electrode plate; acquiring tab position information of a cathode electrode plate, and calculating a second width deviation value corresponding to the cathode electrode plate according to the tab position information and a calibration width of the cathode electrode plate; and when the anode electrode plate and the cathode electrode plate are stacked, adjusting a feed position of the cathode electrode plate according to the first width deviation value, and compensating for a cut width of the cathode electrode plate according to the second width deviation value.

Abstract: A system and method for image series transformation for optimal compressibility with neural upsampling. It incorporates a novel AI deblocking network composed of convolutional layers for feature extraction and a channel-wise transformer with attention to capture complex inter-channel dependencies. Additionally, the system includes an angle optimizer to further enhance the compressibility of an image. The convolutional layers extract multi-dimensional features from the image, while the channel-wise transformer learns global inter-channel relationships. This hybrid approach addresses both local and global features, mitigating compression artifacts and improving image quality. The model's outputs enable effective image reconstruction, achieving advanced compression while preserving crucial information for accurate analysis.

Abstract: An image defect detection method applied to an electronic device is provided. The method includes dividing a detecting image into a plurality of detecting areas, generating a detection accuracy value for each detecting area based on a defective image and a non-defective image, and obtaining a plurality of detection accuracy values. An autoencoder is selected for each detection accuracy value. A model corresponding to each detecting area is obtained by training the autoencoder based on the non-defective image. A plurality of reconstructed image blocks is obtained by inputting each of a plurality of detecting blocks into the corresponding model, and a reconstruction error value between each reconstructed image block and the corresponding detecting block is obtained. A detection result of a product contained in the image to be detected is obtained based on the reconstruction error value corresponding to each detecting block.

Abstract: An inspection system is provided and includes a camera and controller. The controller can include one or more processors in communication with the camera and receive a plurality of images of a target captured by the camera. The controller can also determine, using a first computer vision algorithm, a first prediction and corresponding confidence level for substantially all of the images. The controller can select a subset of the images having the first prediction confidence level greater than or equal to a first prediction threshold value. The controller can additionally determine, using a second computer vision algorithm, a second prediction and corresponding second prediction confidence level for each of the selected images. The at least one second prediction can require more time to determine than the at least one first prediction. The controller can output the second prediction and the second prediction confidence level for each of the selected images.