Abstract: An object position region detection unit of an object detection device detects a position region of an object included in an input image on the basis of a first class definition in which a plurality of classes is defined in advance. A class identification unit identifies a class out of a plurality of classes to which the object belongs on the basis of the first class definition. An object detection result output unit outputs class information of the object as a detection result of the object on the basis of a second class definition in which a plurality of classes is defined in advance, the second class definition associated with the first class definition. The number of classes defined in the second class definition is smaller than the number of classes defined in the first class definition.

Abstract: The subject technology receives, at a client device, a selection of a selectable graphical item from a plurality of selectable graphical items, the selectable graphical item comprising an augmented reality content generator including a 3D effect. The subject technology applies, to image data and depth data, the 3D effect based at least in part on the augmented reality content generator, the applying the 3D effect. The subject technology generates a depth map using at least the depth data, generates a segmentation mask based at least on the image data, and performs background inpainting and blurring of the image data using at least the segmentation mask to generate background inpainted image data. The subject technology generates a 3D message based at least in part on the applied 3D effect.

Abstract: The input means 181 accepts inputs of test data, a hierarchical structure in which a node of bottom layer represents a target class, and a classification score of a seen class as the classification score indicating a probability that the test data is classified into each class. The unseen class score calculation means 182 calculates the classification score of an unseen class based on uniformity of the classification score of each seen class. The matching score calculation means 183 calculates a matching score indicating similarity between the test data and each class label. The final classification score calculation means 184 calculates a final classification score indicating a probability that the test data is classified into the class so that the larger the classification score of each class, and the matching score, the larger the final classification score.

Abstract: A constraint model that includes a representation of feasible viewing window placement within a source field of view of visual content may be generated by using a roll-pitch-yaw axes representation of viewing window placement and having a diagonal dimension of the viewing window that fit within vertical and horizonal dimensions of the source field of view. The constraint model may enable full horizon leveling of the visual content.

Abstract: A method of generating an aggregate saliency map using a convolutional neural network. Convolutional activation maps of the convolutional neural network model are received into a saliency map generator, the convolutional activation maps being generated by the neural network model while computing the one or more prediction scores based on unlabeled input data. Each convolutional activation map corresponds to one of the multiple encoding layers. The saliency map generator generates a layer-dependent saliency map for each encoding layer of the unlabeled input data, each layer-dependent saliency map being based on a summation of element-wise products of the convolutional activation maps and their corresponding gradients. The layer-dependent saliency maps are combined into the aggregate saliency map indicating the relative contributions of individual components of the unlabeled input data to the one or more prediction scores computed by the convolutional neural network model on the unlabeled input data.

Abstract: A method of performing object recognition is performed by an electronic device and includes obtaining a spatial map of a space, using a first recognition model, recognizing one or more objects in the space, to obtain first object information of the objects, and dividing the space into a plurality of subset spaces, based on the obtained spatial map and the obtained first object information. The method further includes determining at least one second recognition model to be allocated to each of the plurality of subset spaces into which the space is divided, based on characteristic information of each of the plurality of subset spaces, and using the determined at least one second recognition model allocated to each of the plurality of subset spaces, performing object recognition on each of the plurality of subset spaces, to obtain second object information.

Abstract: A system and method for automating and improving data extraction from a variety of document types, including both unstructured, structured, and nested content, is disclosed. The system and method incorporate an intelligent machine learning model that is designed to intelligently identify chunks of text, map the fields in the document, and extract multi-record values. The system is designed to operate with little to no human intervention, while offering significant gains in accuracy, data visualization, and efficiency. The architecture applies customized techniques including density-based adaptive text clustering, tabular data extraction based on hierarchical intelligent keyword searches, and natural language processing-based field value selection.

Abstract: A control method of an image signal processor for an artificial neural network may be configured to include a step of acquiring an image, a step of determining at least one image characteristic data corresponding to the image, and a step of determining an image correction parameter (SFR preset) for improving an inference accuracy of an artificial neural network model based on the at least one of image characteristic data and an inference accuracy profile of an artificial neural network model.

Abstract: A method and apparatus that detects whether biometric information is spoofed is provided. The method receives, from a sensor, first feature information including a static feature associated with biometric information of a user, and a dynamic feature obtained based on images related to the biometric information, detects whether the biometric information is spoofed based on a first score calculated based on the first feature information, fuses the first score with a second score calculated based on second feature information extracted from the images, based on a result of the detecting that the biometric information is spoofed based on the first score, and detects that the biometric information is spoofed based on a fused score.

Abstract: The present disclosure relates to an object selection system that automatically detects and selects objects in a digital image based on natural language-based inputs. For instance, the object selection system can utilize natural language processing tools to detect objects and their corresponding relationships within natural language object selection queries. For example, the object selection system can determine alternative object terms for unrecognized objects in a natural language object selection query. As another example, the object selection system can determine multiple types of relationships between objects in a natural language object selection query and utilize different object relationship models to select the requested query object.

Abstract: A data labeling method, apparatus and system are provided. The method includes: sampling a data source according to an evaluation task for the data source to obtain sampled data; generating a labeling task from the sampled data; sending the labeling task to a labeling device; and receiving a labeled result of the labeling task from the labeling device. As such, an automatic evaluation of data can be implemented by using the evaluation task, and evaluation efficiency is improved.

Abstract: Techniques for user customized private label prediction are described. According to some embodiments, customers can train a classifier to detect new objects in image data. These new objects may not be included in a base model provided by a service provider system. The base model can be utilized to perform object detection and feature extraction from training images that are annotated by the customer to identify the new objects. Once trained, the new custom model can be used to identify the new objects in input images and label the images accordingly.

Abstract: Methods, systems, and media for analyzing spherical video content are provided. More particularly, methods, systems, and media for detecting two-dimensional videos placed on a sphere in abusive spherical video content are provided.

Abstract: Discloses are weakly supervised semantic segmentation device and method based on pseudo-masks. The device includes a localization map generator configured to generate a plurality of first localization maps by providing an image to a first classifier; a saliency map processor configured to calculate a saliency loss through a saliency map used to identify a boundary line and a co-occurring pixel based on the plurality of first localization maps; a multi-label processor configured to predict a multi-label based on the plurality of first localization maps and calculate a classification loss; and a pseudo-masks generator configured to generate a second classifier obtained by updating the first classifier based on the saliency loss and the classification loss, and generate the pseudo-masks based on a plurality of second localization maps by the second classifier.

Abstract: Dictionary learning method and means for zero-shot recognition can establish the alignment between visual space and semantic space at category layer and image level, so as to realize high-precision zero-shot image recognition. The dictionary learning method includes the following steps: (1) training a cross domain dictionary of a category layer based on a cross domain dictionary learning method; (2) generating semantic attributes of an image based on the cross domain dictionary of the category layer learned in step (1); (3) training a cross domain dictionary of the image layer based on the image semantic attributes generated in step (2); (4) completing a recognition task of invisible category images based on the cross domain dictionary of the image layer learned in step (3).

Abstract: An out-of-stock detection system notifies store management that a product is out of stock. The system captures images of a shelf and determines the position product labels thereon. For each product label, a bounding box is generated based on the position of each product label on the shelf. The system then identifies a product for each product label based on information within each product label and, for each product label, stores a product identified for each bounding box. Accordingly, the system performs an out-of-stock detection process that includes capturing additional image data of the shelf periodically that includes each bounding box, providing a portion of the additional image data for each bounding box to a model trained to determine whether the bounding box contains products, sending a notification for a product determined to be out of stock to a store client device based on output from the model.

Abstract: An image matching apparatus according to the present invention includes: a common region specification unit configured to specify a common region between a first image and a second image; a date replacement unit configured to generate a first replaced image in which a brightness value of the common region of the first image is replaced based on a pixel in the first image, and a second replaced image in which a brightness value of the common region of the second image is replaced based on a pixel in the second image; and a matching unit configured to perform matching between the first image and the second image based on frequency characteristics of the first replaced image and the second replaced image.

Abstract: Device, systems, and methods are provided that can be used to monitor, examine, and/or analyze hair conditions and make suggestions to improve the hair conditions.

Abstract: Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for segmenting a medical image. In one aspect, a method comprises: receiving a medical image that is captured using a medical imaging modality and that depicts a region of tissue in a body; and processing the medical image using a segmentation neural network to generate a segmentation output. The segmentation neural network can include a sequence of multiple encoder blocks and a decoder subnetwork. Training the segmentation neural network can include determining a set of error values for a segmentation channel; identifying the highest error values from the set of error values for the segmentation channel; and determining a segmentation loss based on the highest error values identified for the segmentation channel.

Abstract: A method and system is provided for detection of carbonate core features from core images. An input carbonate core image is separated into a plurality of first blocks, each of the plurality of first blocks having a first block size. An image of each of the separated plurality of first blocks is input into an artificial intelligence (AI) model. The AI model being trained to predict for each first block, one of a plurality of carbonate core features and a corresponding confidence value indicating a confidence of the predicted carbonate core feature being imaged in the first block. Any bounding boxes of a first set of bounding boxes are detected in the input core image based on the predicted one of the plurality of carbonate core features and the corresponding confidence values for each first block.

Abstract: An image capture device may capture visual content during a capture duration. The context of capture of the visual content by the image capture device may be assessed and used to determine values of stabilization parameters for the visual content.

Abstract: An inpainting method includes obtaining image information at an electronic device, where the image information identifies an area corresponding to a removed object within an image. The method also includes reconstructing the area corresponding to the removed object by (i) applying a semantic mask and a surface normal map to identify and rank neighboring contexts of the area and (ii) sampling, using an attention mechanism, the ranked contexts to generate pixel information for the area. The method further includes rendering the image with the reconstructed area.

Abstract: The present disclosure provides systems and methods for building an inventory database with automatic labeling. A system can maintain a hierarchical concept tree including labels. Each of the labels is associated with a set of attributes and a respective embedding. The system can receive, from a provider device, a request to generate labels for an item of media content. The request can include a request attribute. The system can generate, using a gated categorical model, document embeddings for the item of media content. The system can select a subset of the labels based on the request attribute. The system can determine a respective label score for each label of the subset of the labels based on the document embeddings and the respective embedding of the label. The system can provide a selected label of the subset of the labels based on the respective label score of the selected label.

Abstract: A conferencing endpoint selects a background for a conferencing system. The conferencing endpoint captures an initial series of images of a foreground object in front of a background image, and segments at least one frame of the initial series of images into the foreground object and the background image according to a first segmentation technique. The conferencing endpoint generates one or more test backgrounds and evaluates the test backgrounds according to a second segmentation technique. The conferencing endpoint selects a final background from the test backgrounds for segmenting a subsequent series of images according to the second segmentation technique.

Abstract: A commodity detection terminal is disclosed, the commodity detection terminal includes: an image segmentation unit configured to obtain position information of each grid of a goods shelf in an image of the goods shelf based on a first image, which is an image of the goods shelf not placed with commodities; a detector unit configured to obtain a current grid where each commodity is located and a current quantity of each commodity based on a second image, which is a current image of the goods shelf placed with commodities; and a determination unit configured to compare the current grid and the current quantity with a preset grid and a preset quantity of the each commodity, so as to determine whether a status of each commodity satisfies a preset condition. A commodity detection method, an intelligent goods shelf system, a computer device, and a readable medium are also disclosed.

Abstract: An image processing apparatus includes a superpixel extractor configured to extract a plurality of superpixels from an input original image, a backbone network including N feature extracting layers (here, N is a natural number of two or more) which divide the input original image into grids including a plurality of regions and generate an output value including a feature value for each of the divided regions, and a superpixel pooling layer configured to generate a superpixel feature value corresponding to each of the plurality of superpixels using a first output value to an Nth output value output from each of the N feature extracting layers.

Abstract: A system for extracting and transplanting live video avatar images including a depth sensor for creating a depth map based first live video avatar of a user or object disposed in a heterogeneous first environment with an arbitrary background; a processor coupled to the depth sensor; code fixed in a tangible medium for execution by the processor for extracting the depth map from the first environment to provide an extracted depth map based live video avatar; and a display system coupled to the processor for showing the extracted depth map based live video avatar in a second environment diverse from the first environment. In a second embodiment, the system includes a camera coupled to the processor to provide live video images of the user in the first environment and code for spatially filtering the images to provide a spatially filtered extracted second live video avatar.

Abstract: The present subject matter is related in general to the field of image processing, disclosing method and system for evaluating an image quality for Optical Character Recognition (OCR) Image evaluation system receives image comprising optical character data. The image evaluation system determines image parameter value for each of one or more image parameters of the image. The image parameter value for each of the one or more image parameters is determined for plurality of binary image segments identified in the image. The image evaluation system determines suitability value and impact value of the image, based on the image parameter value for each of the image parameters determined for the image. The image evaluation system determines quality score for the image, based on the suitability value and the impact value. The image is transmitted for processing before the OCR, upon determining the quality score to be above overall pre-defined threshold value.

Abstract: A system for generating a set of digital image features, comprising at least one hardware processor adapted for: producing a plurality of input groups of features, each produced by extracting a plurality of features from one of a plurality of digital images; computing an output group of features by inputting the plurality of input groups of features into at least one prediction model trained to produce a model group of features in response to at least two groups of features, such that a model set of labels indicative of the model group of features is similar, according to at least one similarity test, to a target set of labels computed by applying at least one set operator to a plurality of input sets of labels each indicative of one of the at least two groups of features; and providing the output group of features to at least one other processor.

Abstract: An image or a video may include a spherical capture of a scene. A punchout of the image or the video may provide a panoramic view of the scene.

Abstract: A label modification unit may receive a label modification input associated with an image. The label modification unit may obtain a bitmap associated with a modification to the rewriteable label. The label modification unit may determine, based on the bitmap and using a heat dissipation model, temperature profiles for a plurality of raster print paths of the laser. The label modification unit may select, based on the temperature profiles and from the plurality of raster print paths, a raster print path for the modification. The label modification unit may control at least one of the laser, the optic, and the reflector system to: cause the light beam to follow the raster print path, and emit the light beam according to an array of power factors that are associated with the raster print path.

Abstract: Enhanced methods and systems for the semantic segmentation of images are described. A refined segmentation mask for a specified object visually depicted in a source image is generated based on a coarse and/or raw segmentation mask. The refined segmentation mask is generated via a refinement process applied to the coarse segmentation mask. The refinement process correct at least a portion of both type I and type II errors, as well as refine boundaries of the specified object, associated with the coarse segmentation mask. Thus, the refined segmentation mask provides a more accurate segmentation of the object than the coarse segmentation mask. A segmentation refinement model is employed to generate the refined segmentation mask based on the coarse segmentation mask. That is, the segmentation model is employed to refine the coarse segmentation mask to generate more accurate segmentations of the object. The refinement process is an iterative refinement process carried out via a trained neural network.

Abstract: According to an embodiment, a computing device includes a processing circuitry. The processing circuitry receives an input of tensor data. The processing circuitry sets a window in the tensor data. The processing circuitry compares, for each pair of coordinates in the tensor data within the window, a pixel value at the pair of coordinates with one or more thresholds, and selects a weight value corresponding to a comparison result. The processing circuitry adds the weight values selected for the respective pairs of coordinates to obtain a cumulative value. The processing circuitry derives a value based at least in part on the cumulative value.

Abstract: An approach is provided for feature point detection and representation. The approach, for example, involves processing (e.g., using a neural network or equivalent) image data associated with a grid cell of an image to determine a feature point corresponding to a position of a feature detected in the image data. The approach also involves encoding the position of the feature with respect to a coordinate system referenced to the grid cell. The output comprises one or more parameters indicating the encoded position, one or more attributes of the feature, or a combination thereof.

Abstract: Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for segmenting a medical image. In one aspect, a method comprises: receiving a medical image that is captured using a medical imaging modality and that depicts a region of tissue in a body; and processing the medical image using a segmentation neural network to generate a segmentation output, wherein the segmentation neural network comprises a sequence of multiple encoder blocks, wherein: each encoder block is a residual neural network block comprising one or more two-dimensional convolutional neural network layers, one or more three-dimensional convolutional neural network layers, or both, and each encoder block is configured to process a respective encoder block input to generate a respective encoder block output wherein a spatial resolution of the encoder block output is lower than a spatial resolution of the encoder block input.

Abstract: A method for extracting of data contained in a fixed format electronic document is disclosed. The method is particularly applicable to extracting data from tables in electronic documents and includes reading, by a computer system, the electronic document as a computer image file; segmenting, by the computer system, the computer image file into document sections representative of distinct portions of data; applying a label to each distinct document section; and executing, by the computer system, an optical character recognition algorithm to convert the image file into computer-readable text, wherein segments of the converted text is associated with a respective label indicative of each distinct document section.

Abstract: An electronic apparatus is provided. The electronic apparatus includes: a storage configured to store a plurality of filters each corresponding to a plurality of image patterns; and a processor configured to classify an image block including a target pixel and a plurality of surrounding pixels into one of the plurality of image patterns based on a relationship between pixels within the image block and to obtain a final image block in which the target pixel is image-processed by applying at least one filter corresponding to the classified image pattern from among the plurality of filters to the image block, wherein the plurality of filters are obtained by learning, through an artificial intelligence algorithm, a relationship between a plurality of first sample image blocks and a plurality of second sample image blocks corresponding to the plurality of first sample image blocks based on each of the plurality of image patterns.

Abstract: A method and a system are described for counting objects in an image. The method includes receiving at least one image comprising one or more objects. The method includes determining contours of the one or more objects using one or more morphological operations on the at least one image. The method further includes identifying shapes of the one or more objects based on counting a number of contours associated with each of the one or more objects. The method includes comparing the shapes of the one or more objects with one or more predefined training images to identify one or more objects of interest. The method includes counting the one or more objects of interest based on the shapes of the one or more objects of interest.

Abstract: Values of pixels in an image are mapped to a binary space using a first function that preserves characteristics of values of the pixels. Labels are iteratively assigned to the pixels in the image in parallel based on a second function. The label assigned to each pixel is determined based on values of a set of nearest-neighbor pixels. The first function is trained to map values of pixels in a set of training images to the binary space and the second function is trained to assign labels to the pixels in the set of training images. Considering only the nearest neighbors in the inference scheme results in a computational complexity that is independent of the size of the solution space and produces sufficient approximations of the true distribution when the solution for each pixel is most likely found in a small subset of the set of potential solutions.

Abstract: Methods and systems may include logic to identify a plurality of blocks in image data having one or more top-left dependent pixels, and select the plurality of blocks in a wavefront order for processing. In addition, the logic may process a plurality of pixels in each block in the wavefront order. The system may also include a display device to output a result associated with processing the plurality of pixels.

Abstract: A method, computer program product, and computer system for analyzing an image to detect a plurality of geometric shapes in the image. The method may also include building a graph data structure resembling the image based upon, at least in part, analyzing the image. In some embodiments, building the graph data structure may include traversing the image to generate one or more graph data structure clauses.

Abstract: Implementations relate to generating and displaying blur in images. In some implementations, a method includes generating a plurality of mipmap images based on an input image, including applying a blur to a respective plurality of pixels derived from the input image for each mipmap image. In some examples, the blur is at least partially based on depth data for the image. Parameter data is obtained that indicates an output focal plane depth for an output focal plane of an output image and an output focal range in front of the output focal plane. Output pixel values of the output image are generated, including determining blurred pixel values based on one or more of the mipmap images selected based on the output focal plane depth and the output focal range. The blurred pixel values are based on particular pixels associated with a depth outside the output focal range.

Abstract: An image or a video may include a spherical capture of a scene. A punchout of the image or the video may provide a panoramic view of the scene.

Abstract: A processor-implemented learning method for an image segmentation includes training first duplicate layers, as duplications of trained first layers of a pre-trained model, so that a second feature extracted from a target image by the trained first duplicate layers is matched to a first feature extracted from a training image by the trained first layers; regularizing the trained first duplicate layers so that a similarity between the first feature and a third feature extracted from the training image by the regularized first duplicate layers meets a threshold; and training second duplicate layers, as duplications of trained second layers of the pre-trained model, to be configured to segment the target image based on the regularized first duplicate layers, the trained second layer being configured to segment the training image.

Abstract: A method and apparatus for identifying an intersection in an electronic map, and a computer readable medium are provided. An embodiment of the method includes: acquiring boundary information related to road boundaries from an electronic map; determining a topological relationship between the road boundaries in an area having a predetermined size in the electronic map based on the boundary information; and determining a distribution of an intersection in the area based on the topological relationship. The apparatus corresponding to the method, the device implementing the method of the present disclosure, and the computer readable medium are also provided. Through the technical solutions, the intersection may be automatically identified by detecting the road boundaries, which improves the efficiency of producing a high-precision map, and has the advantage of high accurate recall rate, strong universality, or simple method.

Abstract: A method of object detection includes obtaining a set of images depicting overlapping regions of an area containing a plurality of objects. Each image includes input object indicators defined by input bounding boxes, input confidence level values, and object identifiers. The method includes identifying candidate subsets of input object indicators in adjacent images. Each candidate subset has input overlapping bounding boxes in a common frame of reference, and a common object identifier. The method includes adjusting the input confidence levels for each input object indicator in the candidate subsets; selecting clusters of the input object indicators satisfying a minimum input confidence threshold, having a common object identifier, and having a degree of overlap satisfying a predefined threshold; and detecting an object by generating a single output object indicator for each cluster, the output object indicator having an output bounding box, an output confidence level value, and the common object identifier.

Abstract: A depth sensing system receives an image pair showing objects in a scene. The system generates binary hashes for pixels in the image pair by performing a random walk. The system matches pixels in the first image to pixels in the second image that depict the same point in the scene by generating cost values representing differences between the binary hashes for pairs of pixels in the images. The system generates a disparity map containing disparity vectors representing coordinate differences between matched pixels in the first and second images. The system generates and outputs a depth map based on the disparity map. The depth map represents the distances between an image acquisition system that acquired the image pair and the objects in the scene.

Abstract: An active learning system classifies multiple objects in an input image from the set of images with a classification metric indicative of uncertainty of each of the classified object to belong to one or different classes and determines a diversity metric of the input image indicative of diversity of the objects classified in the input image. The active learning system evaluates the diversity metric of the input image and to cause rendering of the input image on a display device based on a result of the evaluation and trains the classifier using the labelled objects of the input image.

Abstract: A computer software that provides the user with the means to import an image of a paper financial document for data extraction. The extracted data automatically populates a financial datasheet and can be synchronized with a company financial record being kept on an external accounting software. The present invention provides the user with the convenience of automatic data input and eliminates the traditional method of individually inputting financial transactions into the accounting software.

Abstract: Systems and methods are provided for implementing a parallel Expectation Minimization algorithm for generalized latent variable models. Item response data that is based on responses to items from multiple respondents is accessed. The item response data includes data for multiple response variables. The item response data is analyzed using a generalized latent variable model, and the analysis includes an application of a Parallel-E Parallel-M (PEPM) algorithm. In a parallel Expectation step of the PEPM algorithm, the respondents are subdivided into N groups of respondents, and computations for the N groups are performed in parallel using the N processor cores. In a parallel Maximization step of the PEPM algorithm, the response variables are subdivided into N groups of response variables, and computations for the N groups of response variables are performed in parallel using the N processor cores.