Abstract: An apparatus is provided to estimate two-dimensional poses. The apparatus includes a communications interface to receive raw data. The raw data includes a representation of first and second objects. In addition, the apparatus includes a memory storage unit to store the raw data. Furthermore, the apparatus includes a neural network engine to apply a first convolution to the raw data to extract first features from a first output, to downsample the first output to extract a first set of subfeatures from a first suboutput, to apply a second convolution to the first output to extract a second set of features from a second output, and to apply the second convolution the first suboutput. The second output and the second suboutput are to be merged to generate joint heatmaps of the first object and the second object, and bone heatmaps of the first object and the second object.

Abstract: A method, system, and article is directed to automatic content-dependent image processing algorithm selection.

Abstract: A method and apparatus for processing or compressing image data are disclosed. A method of processing or compressing image data, performed by an apparatus, according to an embodiment of the present disclosure may include obtaining a plurality of low-resolution images of a specific object through a microscope; and processing or compressing the plurality of low-resolution images, based on at least one of i) a first correlation between adjacent low-resolution images among the plurality of low-resolution images, or ii) a second correlation between a virtual low-resolution image corresponding to a location of a specific light emitting diode (LED) obtained through the plurality of low-resolution images and an actual low-resolution image corresponding to a location of the specific LED among the plurality of low-resolution images.

Abstract: A processor-implemented method includes generating an adjusted reference patch by adjusting a position of a reference patch in a reference image based on a pixel value of a ground truth (GT) patch of a GT image and a pixel value of the reference patch, wherein the GT patch corresponds to a specific region of an input image; generating a super-resolution (SR) image of the input image using a SR model provided an input that is based on the generated adjusted reference patch; and training the SR model based on the SR image and the GT image.

Abstract: The present disclosure relates to a medical image processing system that is capable of implementing low-latency image processing, a medical image processing method, and a program. An image processing unit executes, at each predetermined processing timing, image processing on each of a plurality of medical images input asynchronously in units of strip images obtained by dividing each of the medical images into a plurality of pieces. The present disclosure can be applied to the medical image processing system.

Abstract: A method and system for correcting a difference in contrast agent density in a sequence of contrast-enhanced image frames. A reference image frame is defined in the sequence of contrast-enhanced image frames, and segmentation is performed on the reference image frame to determine a location of a region of interest within the reference image frame. The region of interest is a region of the reference image frame that contains contrast agent. Other image frames in the sequence of contrast-enhanced images are corrected based on a difference in contrast agent density/image intensity in the region of interest relative to the region of interest in the reference image frame.

Abstract: An image processing method includes: acquiring a first image containing a target object; inputting the first image into an image processing model to obtain a second image, the second image being a mask image of the target object in the first image, a-a value for each pixel in the second image being in a range of 0 to 1, inclusive ([0, 1]), and the range of 0 to 1, inclusive ([0, 1]) indicating a degree of relation between each pixel in the second image and a pixel in the target object; fusing the first image and a background image according to the second image to obtain a fused image; and providing a first interface and displaying the fused image on the first interface.

Abstract: Systems, methods and non-transitory computer readable media for attributing generated visual content to training examples are provided. A first visual content generated using a generative model may be received. The generative model may be associated with a plurality of training examples. Each training example may be associated with a visual content. Properties of the first visual content may be determined. Each visual content associated with a training example may be analyzed to determine properties of the visual content. The properties of the first visual content and the properties of the visual contents associated with the plurality of training examples may be used to attribute the first visual content to a subgroup of the plurality of training examples. The visual contents associated with the training examples of the subgroup may be associated with a source. A data-record associated with the source may be updated based on the attribution.

Abstract: It is provided a method, configured to modify medical image data acquired by different vendor-specific or type-specific medical imaging devices of a same medical imaging technology to look more like a common reference.

Abstract: This application discloses a method for generating an image processing model performed by a computer device. The method includes: performing training by using a first source image sample, a first template image sample, and a first standard synthesized image, to obtain a first parameter adjustment model, and combining the first parameter adjustment model and a first resolution update layer into a first update model; adjusting the first update model into a second parameter adjustment model using a second source image sample and a second template image sample and a second standard synthesized image; combining the second parameter adjustment model and a second resolution update layer into a second update model; and adjusting the second update model into a target image fusion model using a third source image sample, a third template image sample and a third standard synthesized image.

Abstract: An image processing method includes: performing a first division on an acquired preprocessed image to obtain a plurality of first sub-image regions; when it is determined that image parameters of the plurality of first sub-image regions do not meet a first preset condition, performing a second division on the plurality of first sub-image regions, to obtain a plurality of second sub-image regions; performing attribute identification on each second sub-image region of the plurality of second sub-image regions to obtain an identification attribute of each second sub-image region; according to an image processing method corresponding to the identification attribute of each second sub-image region, performing image processing on the second sub-image region to obtain a third sub-image region; and merging a plurality of third sub-image regions to obtain a target image.

Abstract: There is provided an image generation system and method. The method comprises obtaining a runtime image of a semiconductor specimen with a low Signal-to-noise ratio (SNR), and processing the runtime image using a machine learning (ML) model to obtain an output image with a high SNR. The ML model is previously trained using a training set comprising a plurality of low SNR images associated with a high SNR image. The plurality of low SNR images correspond to a plurality of sequences of frames acquired in a plurality of runs of scanning a first site of the specimen. The high SNR image is generated based on the plurality of low SNR images. The training comprises, for each low SNR image: processing the low SNR image by the ML model to obtain predicted image data, and optimizing the ML model based on the predicted image data and the high SNR image.

Abstract: Some embodiments provide a neural network inference circuit for executing a neural network that includes multiple nodes that use state data from previous executions of the neural network. The neural network inference circuit includes (i) a set of computation circuits configured to execute the nodes of the neural network and (ii) a set of memories configured to implement a set of one or more registers to store, while executing the neural network for a particular input, state data generated during at least two executions of the network for previous inputs. The state data is for use by the set of computation circuits when executing a set of the nodes of the neural network for the particular input.

Abstract: An image processing method includes computing a blur kernel of an input image using a kernel estimation model, performing kernel-based deblurring on the input image using the blur kernel to obtain a deconvolved image, and generating an output image by performing kernel-free deblurring based on the deconvolved image.

Abstract: A computer-implemented method of configuring an electronic device for inpainting source three-dimensional (3D) scenes, includes: receiving the source 3D scenes and a user's input about a first object of the source 3D scenes; generating accurate object masks about the first object of the source 3D scenes; and generating inpainted 3D scenes of the source 3D scenes by using an inpainting neural radiance field (NeRF) based on the accurate object masks.

Abstract: The present disclosure relates to systems, methods, and non-transitory computer-readable media that modify digital images using an intelligent user interface tool that determines the intent of a user interaction. For instance, in some embodiments, the disclosed systems receive, via a graphical user interface of a client device, a user interaction with a set of pixels within a digital image. The disclosed systems determine, based on the user interaction, a user intent for targeting one or more portions of the digital image for deletion, the one or more portions including an additional set of pixels that differs from the set of pixels.

Abstract: Methods and systems are provided for removing visual artifacts from a medical image acquired during a scan of an object, such as a patient. In one example, a method for an image processing system comprises receiving a medical image; performing a wavelet decomposition on image data of the medical image; performing one or more 2-D Fourier transforms on wavelet coefficients resulting from the wavelet decomposition; removing image artifacts from the Fourier coefficients determined from the 2-D Fourier transforms using a filter; reconstructing the medical image using the filtered Fourier coefficients; and displaying the reconstructed medical image on a display device of the image processing system.

Abstract: An Augmented-Reality which performs operations that include: accessing a data object that comprises image data, location data, and orientation data; applying a transformation to the data object to produce a rectified data object; generating a point cloud based on the rectified data object; assigning the point cloud to a location based on at least the location data of the data object; detecting a client device at the location; and loading the point cloud to the client device in response to the detecting the client device at the location.

Abstract: A method for reducing a dynamic range of an image, whereby first and second images have pixel values of a first bit depth. One or more peak regions and a representative pixel value for each peak region is identified in a histogram of the pixel values in the first image. For pixel values in the first image that are within a peak region, a pixel value difference is calculated as a difference between the pixel value and the representative pixel value of the peak region. For pixel values in the first image that are outside of the peak regions, the pixel value difference is zero. The pixel values of the second image are transformed to reduce the bit depth to a second bit depth. The pixel value differences calculated from the first image are then added to the pixel values of the transformed second image.

Abstract: This application provides a tone mapping method and apparatus for a panoramic image. The tone mapping method includes: determining one or more target metadata information units of a first pixel from a plurality of metadata information units, where the plurality of metadata information units are obtained by parsing a bitstream, the first pixel is any pixel in a to-be-processed panoramic video two-dimensional planar projection, the plurality of metadata information units correspond to a plurality of segmented regions included in a panoramic video three-dimensional spherical representation panoramic image, and there is a mapping relationship between the panoramic video two-dimensional planar projection and the panoramic video three-dimensional spherical representation panoramic image; and performing tone mapping on a pixel value of the first pixel based on the one or more target metadata information units, to obtain a target tone mapping value of the first pixel.

Abstract: Disclosed herein are system, apparatus, article of manufacture, method and/or computer program product embodiments, and/or combinations and sub-combinations thereof, for dynamic tone mapping of video content. An example embodiment operates by identifying, by a dynamic tone mapping system executing on a media device, characteristics of a first video signal having a first dynamic range based on a frame-by-frame analysis of the first video signal. The example embodiment further operates by modifying, by the dynamic tone mapping system, a tone mapping curve based on the characteristics of the first video signal to generate a modified tone mapping curve. Subsequently, the example embodiment operates by converting, by the dynamic tone mapping system, the first video signal based on the modified tone mapping curve to generate a second video signal having a second dynamic range that is less than the first dynamic range.

Abstract: Disclosed are various techniques for generating gain maps. According to some embodiments, one technique for generating a gain map includes the steps of (1) accessing a high dynamic range (HDR) image, (2) accessing a standard dynamic range (SDR) image, (3) for each pixel shared between the HDR and SDR images: plotting the pixel onto a graph, wherein an HDR value of the pixel is plotted on a Y-axis of the graph and an SDR value of the pixel is plotted on an X-axis of the graph, (4) establishing a first curve that represents an approximation of the plotted pixels, (5) inverting the first curve to establish a second curve, (6) applying the second curve against the plotted pixels to establish replotted pixels, (7) generating the gain map based on the replotted pixels, and (8) embedding the gain map into the HDR image.

Abstract: Disclosed are various techniques for utilizing gain maps. According to some embodiments, one technique for utilizing a gain map comprises (1) accessing an enhanced image that includes a high dynamic range (HDR) image and the gain map, (2) extracting the HDR image and the gain map from the enhanced image, (3) generating a standard dynamic range (SDR) image using the HDR image and the gain map, (4) receiving and applying first modification instructions against the HDR image, (5) generating second modification instructions based on at least the first modification instructions, (6) applying the second modification instructions to the SDR image, (7) generating a second gain map by comparing the HDR image against the SDR image or vice-versa, and (8) embedding the second gain map into the HDR image or the SDR image.

Abstract: This application describes techniques for preprocessing images to improve gain map compression outcomes. A gain map can be generated by comparing a first image to a second image and subsequently compressed to form a compressed gain map. The compressed gain map can combined with a compressed version of the first image to form a compressed enhanced image. The compressed enhanced image can be later uncompressed to generate an uncompressed version of the first image and the gain map applied to the uncompressed version of the first image to generate a version of the second image to provide for reproduction on a target display. The compressed version of the first image and the compressed gain map can be generated separately by an image compression module and a gain map compression module or jointly by a combined gain map generation and compression module.

Abstract: An output device including: a memory; and a processor coupled to the memory, the processor configured to: acquire a satellite image in which a target region is photographed and event information relating to events that have occurred in the target region; and output, from among roads in the target region, a road for which a safety degree calculated based on the satellite image and the event information is less than a threshold value.

Abstract: Methods and system of characterizing perforations in a tubular. An imaging tool having an acoustic imaging probe transmits waves and receives acoustic reflections from the tubular. Image segments are processed from the acoustic data. A first trained neural network operates on the image segments to identify processed image segments containing a perforation. A second trained neural network operates on processed image segments containing a perforation to output the location of perforations along the tubular. Another trained neural network can calculate a geometric size of the perforations, such as their size or points along their contour.

Abstract: An image processing device receives a caption for an image before, during, or after capture of the image by an image capture device. The image processing device generates image processing settings based on the caption, for instance based on a mood indicated in the caption or an object identified in the caption. If the caption is received before image capture, the image processing settings may include image capture settings that the image capture device may use to alter exposure or focus during image capture. Once the image is captured, the image processing device may process the image based on the image processing settings, for instance by applying filters or adjusting gain, brightness, contrast, saturation, or colors. For instance, brightness and saturation may be altered if the caption indicates a happy or sad mood, and focus may be altered to focus on an object identified in a caption.

Abstract: A method and system for detecting anomalies associated with a plurality of data objects of a technical installation is provided. The method includes receiving a request to detect anomaly associated with the plurality of data objects in an image data. The image data can be a static image data or a dynamic image data. The method includes classifying the plurality of data objects using a multimodal knowledge graph. The multimodal knowledge graph includes a plurality of pre-classified data objects, each of pre-classified data object is connected using one or more paths. Further, the method includes detecting whether anomaly exists in the plurality of classified data objects. The anomaly is detected by evaluating the plurality of classified data objects with reference data corresponding to the plurality of classified data objects obtained from the multimodal knowledge graph. Furthermore, the method includes generating a notification indicating the detected anomaly based on the evaluation.

Abstract: The present disclosure is directed to devices and techniques for auto-qualifying images on new hardware or infrastructure configurations. The systems and methods include initiating an auto-qualification process for pre-testing one or more images registered within a bare metal system having one or more new infrastructure configurations, discovering all of the one or more images registered for use within the bare metal system, booting each of the one or more registered images into an isolated infrastructure having the one or more new infrastructure configurations, probing instances of each of the one or more registered images booted on the isolated infrastructure to determine stability of each of the one or more registered images on the one or more new infrastructure configurations, and marking each of the one or more registered images as stable or unstable.

Abstract: In an inspection device, a classification means classifies temporal captured images which capture a target object, into a plurality of groups. A recognition means recognizes the captured images belonging to each of the groups, and outputs a determination result for each of the groups. An integration means integrates respective determination results of the groups, and outputs a final determination result.

Abstract: A method of predicting virtual metrology data for a wafer lot that includes receiving first image data from an imager system, the first image data relating to at least one first wafer lot, receiving measured metrology data from metrology equipment relating to the at least one first wafer lot, applying one or more machine learning techniques to the first image data and the measured metrology data to generate at least one predictive model for predicting at least one of virtual metrology data or virtual cell metrics data of wafer lots, and utilizing the at least one generated predictive model to generate at least one of first virtual metrology data or first virtual cell metrics data for the first wafer lot.

Abstract: A method, apparatus, system, and computer program product for detecting inconsistencies in an unconsolidated composite material. Fiber inconsistencies are detected in sections of the fibers input into a composite material manufacturing system in real time during operation of the composite material manufacturing system. Material inconsistencies are detected in the unconsolidated composite material in real time as the unconsolidated composite material is being manufactured by the composite material manufacturing system using the fibers. A determination is made as to whether the material inconsistencies are present in a number of the sections in which the fiber inconsistencies are detected in real time as the unconsolidated composite material is being manufactured by the composite material manufacturing system.

Abstract: A computer-implemented method for processing images may include obtaining at least one image for analyzing; inputting the at least one image to at least one of a plurality of image plugins; analyzing the at least one image via the at least one of the plurality of image plugins; determining metadata related to the at least one image based on the at least one of the plurality of image plugins; filtering the at least one image based on one or more rule sets to generate at least one filtered image; sorting the at least one filtered image to generate at least one sorted image; displaying the at least one sorted image based on an organizational sequence of a webpage; displaying navigation controls via the webpage; and displaying the at least one sorted image according to a user interaction with the navigation controls on the webpage.

Abstract: In a learning device, an acquisition means acquires captured images in a time series which capture a target object. Next, a learning means simultaneously trains a group discrimination model for discriminating a plurality of groups from the captured images based on features in each image and a plurality of recognition models each for recognizing captured images belonging to a corresponding group.

Abstract: A visual inspection apparatus inspects a surface of a piston based on a captured image acquired by a camera and a learning result acquired by conducting machine learning using a plurality of defective product sample images, which is each generated by combining a two-dimensional image of a defect image that is generated based on a three-dimensionally pre-generated defect model with an image of the surface of the piston.

Abstract: A method, an apparatus and a device for testing an adhesive tape pasted onto a surface of an electrode plate, and a machine for applying an adhesive tape to an electrode plate are described. In the method, a first surface image and a second surface image are acquired, where a color distance between a background color away from a tab side in the first surface image and a color of a first adhesive tape is set to be greater than or equal to a first preset value and a color distance between a background color away from the tab side in the second surface image and a color of a second adhesive tape is set to be greater than or equal to a second preset value, such that the colors of the first adhesive tape and the second adhesive tape are significantly different from the corresponding background colors.

Abstract: A non-contact detection method for a nut is provided. The method includes the following steps. The nut is photographed to obtain a threaded hole image of the nut. A thread area comparison between the threaded hole image and a standard threaded hole image is performed. An area difference is obtained according to the result of the thread area comparison. Whether the nut is a good nut is determined according to the area difference.

Abstract: The invention relates to a method for testing quality of an object in a real environment using a camera, an optical display device, and a processing apparatus, the method including the following steps: defining a test geometry and a reference geometry in a computer-assisted data mode; defining a test pose, in which the camera should be placed by the user as target positioning for a quality test to be carried out of the object to be tested; and visualizing the test pose on the optical display device. In a second phase, at least one image of the real environment is captured by the camera, the pose of which camera is in a range that includes the test pose, and the test geometry and the reference geometry in the image are tracked. Furthermore, a pose of the tracked test geometry in relation to the reference geometry and at least one parameter are determined on the basis of how the pose of the tracked test geometry is in relation to a target pose of the test geometry defined within the data model.

Abstract: An apparatus including a processor configured to execute a plurality of instructions; and a memory storing the plurality of instructions, wherein execution of the plurality of instructions configures the processor to generate a defect prediction score of an input image through the use of a neural network provided reference image, the input image, and an enhanced image. The neural network may include an attention map modulator configured to adaptively adjust an intensity of an attention map generated during the use of the neural network.

Abstract: A texture pattern is reduced from an image easily. An image processing apparatus matches the spatial frequency component of a first texture pattern taken to be a reference with the spatial frequency component of a second texture pattern included in a target image and reduces the spatial frequency component of the second texture pattern included in the target image based on results of the matching.

Abstract: A medical information processing system comprises: a data store storing a medical ontology, knowledge graph or other knowledge base; and processing circuitry configured to: receive medical image data; receive an input regarding position on the medical image data; specify a concept in the medical ontology, knowledge graph or other knowledge base which includes or is related to the position; and specify text data including a term included in or related to the specified concept.

Abstract: A method for training an artificial neural network for detecting a prostate cancer from a TURP pathological image, includes: acquires a plurality of acquiring pathological images for primary training, each being a prostate needle biopsy pathological image or a radical prostatectomy pathological image; using the pathological images to primarily train an artificial neural network for determining prostate cancer; acquiring TURP pathological images; and using the TURP pathological images to secondarily train the primarily trained artificial neural network, wherein each TURP pathological image includes a non-prostate tissue region and/or a cauterized prostate tissue region, and does not include any prostate cancer lesion region.

Abstract: An imaging module to obtain an image of an entire bottle in one image frame. The imaging module places the bottle in an auxiliary mirror module (AMM). The module has a quasi-conical shaped mirror which offers a reflection of the bottle placed in the AMM. That reflection is the source of an image obtained by an imaging assembly. The imaging assembly can be a lens/camera assembly or a photo sensor that will detect photo fluorescence.

Abstract: The electronic device for recommending an AI-based melanoma biopsy site according to an exemplary embodiment of the present invention includes a processor which classifies a skin image that is input by using a classification model as melanoma or nevus, identifies melanoma features in the skin image by using a generation model when the skin image is classified as melanoma to generate an image from which the melanoma features are removed, and compares the skin image with the generated image to identify at least one candidate biopsy site.

Abstract: Devices, systems, kits, software and methods for using deep learning for enhanced identification of anatomical structures or tissue types during eye surgery.

Abstract: The image processing device 1X includes an acquisition means 31X, a selection means 32X, and a determination means 33X. The acquisition means 31X is configured to acquire data obtained by applying Fourier transform to an endoscopic image of an examination target photographed by a photographing unit provided in an endoscope. The selection means 32X is configured to select partial data that is a part of the data. The determination means 33X is configured to make a determination regarding an attention point to be noticed in the examination target based on the partial data.

Abstract: Image processing performed by a processor and including acquiring a first fundus image of an examined eye, acquiring a position for acquiring a tomographic image of the examined eye's fundus, acquiring a second fundus image of the examined eye, determining the acquired position is included in a specific range of the first fundus image, and computing a first movement amount of the examined eye in a case in which the acquired position is included in the specific range, using first registration processing to positionally align the first fundus image and the second fundus image, and computing a second movement amount of the examined eye in a case in which the acquired position falls outside the specific range, using second registration processing to positionally align the first fundus image and the second fundus image, the second registration processing being different from the first registration processing.

Abstract: Systems, methods, and storage media for automatically identifying medical conditions based on images and other data using machine learning methods and algorithms that encode expert knowledge of medical conditions. The systems, methods, and storage media may also leverage infrastructure that facilitates the collection, storage, labeling, dynamic organization, and expert review of existing and incoming data.

Abstract: A method and apparatus for identifying blood vessels in ultrasound images and displaying blood vessels in ultrasound images are described. In some embodiments, the method is implemented by a computing device and includes assigning, with a neural network implemented at least partially in hardware of the computing device, one of a vein classification and an artery classification to one or more blood vessels in ultrasound images. The method also includes determining a misclassification for one blood vessel that denotes the neural network assigning the one of the vein classification and the artery classification to the one blood vessel in one ultrasound image and the other of the vein classification and the artery classification to the one blood vessel in additional ultrasound images. The method includes displaying, in the one ultrasound image, an indication of the other of the vein classification and the artery classification for the one blood vessel.

Abstract: Methods, systems, and apparatus for an imaging-based MicroOrganoSphere drug assay. In one aspect, a method includes obtaining image data of a well plate comprising a plurality of MicroOrganoSpheres; in response to applying a machine learning model configured to identify instances of at least some of the plurality of MicroOrganoSpheres in the image data, obtaining (i) indications indicative of each instance of the MicroOrganoSpheres and (ii) attributes of each instance of the MicroOrganoSpheres; and normalizing, based on the indications and the attributes, a well-to-well variation in the well plate.