Abstract: The present invention provides a motion capture with a high accuracy which can replace an optical motion capture technology, without attaching optical markers and sensors to a subject. A subject with an articulated structure has a plurality of feature points in the body of the subject including a plurality of joints wherein a distance between adjacent feature points is obtained as a constant. A spatial distribution of a likelihood of a position of each feature point is obtained based on a single input image or a plurality of input images taken at the same time. One or a plurality of position candidates corresponding to each feature point are obtained based on the spatial distribution of the likelihood of the position of each feature point. Each join angle is obtained by performing an optimization calculation based on inverse kinematics using the candidates and the articulated structure.

Abstract: A device may receive an image and may process the image, with a first model or a second model, to convert the image into a binary image. The device may generate an identifier that identifies the first model, or identifies the second model and a color removed from the image, and may utilize clustering to cluster pixels of the binary image and to generate a segmented image with a quantity of segments. The device may generate a particular number of segments to select, and may select the particular number of segments, as selected segments, from the quantity of segments. The device may mask the selected segments to generate a protected image with masked segments, and may associate the protected image with the identifier and with original pixel data of the masked segments. The device may store the protected image, the identifier, and the original pixel data in a data structure.

Abstract: Provided is a robot, including: a plurality of sensors; a processor; a tangible, non-transitory, machine readable medium storing instructions that when executed by the processor effectuates operations including: capturing, with an image sensor, images of a workspace as the robot moves within the workspace; identifying, with the processor, at least one characteristic of at least one object captured in the images of the workspace; determining, with the processor, an object type of the at least one object based on characteristics of different types of objects stored in an object dictionary, wherein possible object types comprise a type of clothing, a cord, a type of pet bodily waste, and a shoe; and instructing, with the processor, the robot to execute at least one action based on the object type of the at least one object.

Abstract: An eye detecting device is configured to: acquire a color image including a face of a person taken by an image taking device; generate a grayscale image by multiplying each of a red component value, a green component value, and a blue component value of each pixel of the color image by a predetermined ratio according to characteristics of a lens of glasses that the person wears; detect an eye of the person from the grayscale image; and output eye information on the eye of the person.

Abstract: Image denoising includes obtaining a saliency map for an image. The saliency map includes respective saliency scores for pixels of the image. Respective noise levels are assigned to the pixels using the respective saliency scores to obtain a noise level map. The image is denoised using the noise level map to obtain a denoised image. The denoised image is output, such as to a display or a storage device.

Abstract: In one aspect a computer-implemented method of processing a content item to extract information from the content item comprises steps of: receiving the content item at a processing stage; and determining whether the content item satisfies a predetermined content condition, by i) providing content of the content item to a plurality of content analysers, each of which applies machine learning content analysis thereto, in order to make an independent determination of whether that predetermined content condition is satisfied, and provides a confidence score for that independent determination, and ii) making an overall determination of whether the content item satisfies that predetermined content condition based on the confidence scores provided by the content analysers for their respective independent determinations.

Abstract: An image processing apparatus determines, based on results of extracting a foreground area from a captured image by learning, a threshold value that is used for generation of a foreground image based on a difference between the captured image and a background image corresponding to the captured image. The image processing apparatus generates a foreground image based on a difference between the captured image and the background image by using the determined threshold value.

Abstract: A system and method for dermal spraying includes a portable, hand-held dermal application device with disposable formulation capsules that spray a formulation unto the skin and a data transmission unit operatively connecting the dermal spray device to a mobile device. The mobile device communicates with a remote server and transmits anonymized data about the user's skin conditions and treatment history. The anonymized data may be labelled and classified by a dermatologist, and stored on a secure cloud server. The anonymized data is used to train models for serum and treatment plan recommenders.

Abstract: Disclosed are a model parameter training method and a terminal based on federation learning, and a medium. The method includes: determining a feature intersection of a first sample of the first terminal and a second sample of a second terminal, training the first sample based on the feature intersection to obtain a first mapping model, sending the first mapping model to the second terminal; receiving a second encryption mapping model sent by the second terminal, predicting a missing feature of the first sample of the first terminal according to the second encryption mapping model to obtain a first encryption supplementary sample; receiving a first encryption federation learning model parameter sent by a third terminal, training a federation learning model to be trained according to the first encryption federation learning model parameter, and calculating a first encryption loss value; and sending the first encryption loss value to the third terminal.

Abstract: A mechanism to provide visual feedback regarding computing system command gestures. An embodiment of an apparatus includes a sensing element to sense a presence or movement of a user of the apparatus, a processor, wherein operation of the processor includes interpretation of command gestures of a user to provide input to the apparatus; and a display screen, the apparatus to display one or more icons on the display screen, the one or more icons being related to the operation of the apparatus. The apparatus is to display visual feedback for a user of the apparatus, visual feedback including a representation of one or both hands of the user while the one or both hands are within a sensing area for the sensing element.

Abstract: Example systems and methods may selection of video frames using a machine learning (ML) predictor program are disclosed. The ML predictor program may generate predicted cropping boundaries for any given input image. Training raw images associated with respective sets of training master images indicative of cropping characteristics for the training raw image may be input to the ML predictor, and the ML predictor program trained to predict cropping boundaries for raw image based on expected cropping boundaries associated training master images. At runtime, the trained ML predictor program may be applied to runtime raw images in order to generate respective sets of runtime cropping boundaries corresponding to different cropped versions of the runtime raw image. The runtime raw images may be stored with information indicative of the respective sets of runtime boundaries.

Abstract: A computer system, product, and method are provided. The computer system includes an artificial intelligence (AI) platform operatively coupled to a processor. The AI platform includes tools in the form of a machine learning model (MLM) manager, a metric manager, and a training manager. The MLM manager accesses a plurality of pre-trained source MLMs, and inputs a plurality of data objects of a test dataset into each of the source MLMs. The test dataset includes the plurality of data objects associated with respective labels. For each source MLM, associated labels are generated from the inputted data objects and a similarity metric is calculated. The MLM manager selects a base MLM to be used for transfer learning from the plurality of source MLMs based upon the calculated similarity metric. The training manager trains the selected base MLM with a target dataset for the target domain.

Abstract: The technology disclosed enables a user to optimize a sampling logic to increase the future sampling likelihood of those instances that are similar to the instances that the user believes are informative, and decrease the future sampling likelihood of those instances that are similar to the instances that the user believes are non-informative.

Abstract: The present invention is directed to making it possible to reduce a possibility of image stitching failures. An information processing apparatus includes a feature extraction unit configured to extract feature points from each of a plurality of images of an object to be inspected captured from a plurality of viewpoints, an image quality evaluation unit configured to, for each of the plurality of images, evaluate whether an image quality based on a predetermined index satisfies allowable conditions for inspection works of the object, and an image stitching unit configured to stitch at least a part of the images, among the plurality of images, having the image quality that satisfies the allowable conditions according to a positional relation based on the feature points.

Abstract: Techniques for generating text content arranged in a consistent read order from a source document including text corresponding to different read orders are disclosed. A system parses a binary file representing an electronic document to identify characters and metadata associated with the characters. The system pre-sorts a character order of characters in each line of the electronic document to generate an ordered list of characters arranged according to the right-to-left reading order. The system performs a layout-mirroring operation to change a position of characters within the modified document relative to a right edge of the document and a left edge of the document. Subsequent to performing layout-mirroring, the system identifies native left-to-right reading-order text in-line with the native right-to-left reading-order text.

Abstract: In one aspect, an example method for generating a candidate image for use as backdrop imagery for a graphical user interface is disclosed. The method includes receiving a raw image and determining an edge image from the raw image using edge detection. The method also includes identifying a candidate region of interest (ROI) in the raw image based on the candidate ROI enclosing a portion of the edge image having edge densities exceeding a threshold edge density. The method also includes manipulating the raw image relative to a backdrop imagery canvas for a graphical user interface based on a location of the candidate ROI within the raw image. The method also includes generating, based on the manipulating, a set of candidate backdrop images in which at least a portion of the candidate ROI occupies a preselected area of the backdrop imagery canvas, and storing the set of candidate backdrop images.

Abstract: This application provides a method for training a pose recognition model performed at a computer device. The method includes: inputting a sample image labeled with human body key points into a feature map model included in a pose recognition model, to output a feature map of the sample image; inputting the feature map into a two-dimensional (2D) model included in the pose recognition model, to output 2D key point parameters used for representing a 2D human body pose; input a target human body feature map cropped from the feature map and the 2D key point parameter into a three-dimensional (3D) model included in the pose recognition model, to output 3D pose parameters used for representing a 3D human body pose; constructing a target loss function based on the 2D key point parameters and the 3D pose parameters; and updating the pose recognition model based on the target loss function.

Abstract: In a method of obtaining a reference image for optical object recognition, first light sources among a plurality of light sources included in a display panel are driven where the first light sources are disposed to correspond to an object recognition window which is a partial region of the display panel, a plurality of images are contemporaneously obtained based on reflected light received by an object recognition sensor through the object recognition window while driving the first light sources, at least one of the plurality of images is obtained as a first reference image for optical object recognition associated with a first object, and the first reference image is stored.

Abstract: Wireless communication devices, systems, and methods relate to selecting transmission parameters that optimize a dimension (e.g., time, frequency, and/or spatial) under a set of constraints. A UE may receive a resource allocation mode indication and a traffic parameter indication from a BS. The resource allocation mode indication informs the UE to optimize along a time dimension, a frequency dimension, and/or a spatial domain. The traffic parameter indication informs the UE what constraints to consider for the transmission parameters. The UE may use this information upon receipt of a reference signal in selecting transmission parameters based on the received reference signal, resource allocation mode indication, and traffic parameter indication. The resource allocation mode may correspond to optimization within a single slot or across multiple slots. Once selected, the UE may transmit the transmission parameters to the BS for use in sending the message to the UE.

Abstract: The present invention refers to automatics and computing technology, namely to the field of processing images and video data, namely to correction the eyes image of interlocutors in course of video chats, video conferences with the purpose of gaze redirection. A method of correction of the image of eyes wherein the method obtains, at least, one frame with a face of a person, whereupon determines positions of eyes of the person in the image and forms two rectangular areas closely circumscribing the eyes, and finally replaces color components of each pixel in the eye areas for color components of a pixel shifted according to prediction of the predictor of machine learning. Technical effect of the present invention is rising of correction accuracy of the image of eyes with the purpose of gaze redirection, with decrease of resources required for the process of handling a video image.

Abstract: An image classification method is provided to an electronic device. The method includes: receiving a target image and a reference image about the target image, the target image being a medical image; determining a first image feature of the target image and a second image feature of the reference image; fusing the first image feature and the second image feature to determine a to-be-classified image feature; and determining, by using the to-be-classified image feature, a probability that the target image belongs to a preset category.

Abstract: An image processing apparatus includes an acquisition unit that acquires N-valued image data corresponding to an image including at least one object, and edge data indicating an edge area of the object, an image processing unit that generates M-valued (N>M?2) quantized data where each of a plurality of pixels is associated with one of a value indicating that no dot is to be recorded and a value indicating that a dot is to be recorded, based on the acquired image data, using an image processing parameter, and a changing unit that changes at least one pixel among pixels that indicate the edge area of the object in the edge data and have the value indicating that no dot is to be recorded in the quantized data so that the at least one pixel has the value indicating that the dot is to be recorded.

Abstract: Sound signals that are not audible are generated by one or more speakers disposed in the vehicle. Reflected sound signals from a driver or a passenger of the vehicle that are not audible are received by one or more microphones disposed in the vehicle. A behavior-induced acoustic pattern is detected based on the reflected ultrasound signals. The behavior-induced acoustic pattern is analyzed to recognize a behavior of the driver or the passenger of the vehicle. A response or an alert is generated according to the recognized behavior of the driver or the passenger in the vehicle.

Abstract: A fusion method and a fusion system for multiple depth information are provided. A first depth information is obtained by a first image capturing module. A second depth information is obtained by a second image capturing module. Through a non-linear transformation, the first depth information is transformed into a first transformation information, and the second depth information is processed according to the first transformation information to generate a transformed information; or the first depth information and the second depth information are merged to be a fusion information, which is then transformed into the transformed information through the nonlinear transformation.

Abstract: A method for enhancing movement or autonomy, including movement or flight of vehicles or flight vehicles in communication and or GPS degraded and or denied and or non-optimal environments is presented, where 3D area maps and or area maps may be employed, and 3D area maps and or area maps employed may include embedded position, location or GPS date, where the location or GPS data may be used in the present invention to determine vehicle location through image recognition, and using the GPS coordinates or location from the 3D area map of the image recognized, or measuring distance to other know aspects of a 3D area map in the real world, and context or system level information may also be used to estimate obstacle shapes and optimal routes in a potential route of travel or for understanding an area.

Abstract: A method of generating an image recognition model for recognising an input image and a system thereof are provided. The method includes appending at least one feature extraction layer to the image recognition model, extracting a plurality of feature vectors from a set of predetermined images, grouping the plurality of feature vectors into a plurality of categories, clustering the plurality of feature vectors of each of the plurality of categories into at least one cluster, determining at least one centroid for each of the at least one cluster, such that each of the at least one cluster comprises at least one centroid, such that each of the at least one centroid is represented by a feature vector, generating a classification layer based on the feature vector of the at least one centroid of the plurality of categories, and appending the classification layer to the image recognition model. In addition, a method of classifying an input image and a system thereof are provided.

Abstract: The electronic device includes a memory storing one or more instructions, and a processor configured to execute the one or more instructions to: receive a selection of a location of a synthesis region in an input image, obtain a target object image to be located in the synthesis region of the input image, by using at least one object detected in the input image, generate a composite image by combining the input image with the target object image, and control a display of the electronic device to display the composite image.

Abstract: A system for detects and manages recurring transactions incurred by members of a client organization. The system includes a computing server that is configured to classify named entities into a category of transacting named entities. For example, the computing server classifies a list of named entities that offer subscription services and hence, create recurring transactions with the client organization. The computing server receives transaction data and determines that a target transacting named entity extracted from the received data belongs to the category (e.g., there is a named entity in the received transaction data that has sold a service to the client organization more than once). The computing server may then determine the presence and frequency of a subscription transaction series in the received transaction data, and cause a predicted a timing of an upcoming transaction in the subscription transaction series to be displayed in a graphical user interface (GUI).

Abstract: The technology relates to selecting and displaying images captured at different points in time. As an example, a user of a computing device may view a first street level image as viewed from a particular location and oriented in a particular direction. The user may select other time periods for which similar images are available. Upon selecting a particular time period, a second street level image may be displayed concurrently with the first street level image, wherein the second street level image was captured on or around the selected time period. If the user changes the perspective of the first image an automatic change in perspective of the second image may occur.

Abstract: Provided is a meat inspection apparatus including a light source configured to emit a plurality of inspection lights to a plurality of regions of meat, respectively, a light detector configured to generate a plurality of emission spectrum signals based on measuring a plurality of fluorescences emitted from the plurality of regions, and a processor configured to receive the plurality of emission spectrum signals from the light detector, generate hyperspectral images of the meat based on the plurality of emission spectrum signals, and obtain a state of the meat based on the hyperspectral images, wherein the processor is further configured to obtain the state of the meat for a plurality of sub-regions in each of the plurality of regions.

Abstract: A modulated segmentation system can use a modulator network to emphasize spatial prior data of an object to track the object across multiple images. The modulated segmentation system can use a segmentation network that receives spatial prior data as intermediate data that improves segmentation accuracy. The segmentation network can further receive visual guide information from a visual guide network to increase tracking accuracy via segmentation.

Abstract: A system includes an imaging device configured to capture image data of a monitored area and a computing device having at least one processor. The computing device processor is configured to receive first image data from the imaging device and second image data from a satellite imagery system, where the second image data relates to the monitored area. The processor is also configured to determine a first set of points of interest in the first image data and a second set of points of interest in the second image data, wherein members of each set of points of interest are represented by a respective vector. The processor is further configured to generate a homography matrix based on the two vectors and determine, using the homography matrix, latitude and longitude coordinates of at least one object represented in the first image data but not the second image data.

Abstract: The present disclosure provides a human-computer interaction method and system. The method includes following operations: by a motion sensor, sensing a hand movement to generate a movement signal; by a camera module, capturing a plurality of images of the hand movement; by at least one processor, generating a plurality of feature values according to the movement signal adjusted based on hand movement information, wherein the hand movement information is generated by analyzing the plurality of images of the hand movement; and by the at least one processor, controlling a virtual object by a first hand gesture determined by a hand gesture determination model according to the plurality of feature values.

Abstract: A system and a method for training and utilizing deep convolutional neural networks to perform diagnostic and image enhancement operations on images of dentition uses digital images and other auxiliary parameters as inputs for a convolutional neural network. The neural network can output a tooth segmentation map, tooth identifiers, a probability map indicating the presence of caries, cavities or other dental anomalies/conditions, and recommended parameters for use in image enhancement algorithms.

Abstract: A method of generating a three-dimensional topological surface representation of a tyre on a vehicle, the method comprising: using a tread depth measurement device to record tread depth data for a tyre surface moving relative to the tread depth measurement device; generating a movement profile of the tyre surface; and using the movement profile of the tyre surface to map the tread depth data onto a base tyre structure, thereby generating a three-dimensional topological surface representation of the tyre.

Abstract: A media summary is generated to include portions of media items. The portions of media items identified for inclusion in the media summary is determined based on the length of the media summary and classification of content depicted within the media items. Classification of content depicted within the media items includes number of smiles depicted within the media items.

Abstract: Unsupervised learning for video classification. One or more features from one or more video clips are extracted using a spatial-temporal encoder. The one or more extracted features are processed, using a video instance discrimination task, to generate a classification label, the classification label indicating whether two of the video clips are from a same video. The one or more extracted features are processed, using a pair-wise speed discrimination task, to generate a comparison label, the comparison label indicating a relative playback speed between two given video clips. A search is performed in a video database for a video that is similar to a given video based on the comparison label.

Abstract: A system for updating linear traffic feature data is provided. The system, for example, determines, from sensor data, the linear traffic feature data associated with a matched link. Further, the system, for example, generates heading data based on the linear traffic feature data and updates the linear traffic feature data based on the generated heading data. The heading data is generated by dropping a perpendicular to at least one of the matched link and a second link connected to the matched link.

Abstract: A method for retrieving footprint images is provided, comprising: pre-training models; cleaning footprint data and conducting expansion pre-processing by using the pre-trained models, dividing the footprint data into multiple data sets; adjusting full connection layers and classification layers of the models; training the models again by using the data sets through the parameters of the pre-trained models; saving the models trained twice, removing the classification layer, executing a feature extraction for images in an image library and a retrieval library to form a feature index library; connecting the features extracted by three models to form fused features, establishing a fused feature vector index library; extracting the features of the images in the image library to be retrieved in advance, and establishing a feature vector library; calculating distances in the retrieval library and the image library when a single footprint image is inputted, thereby outputting the image with the highest similarity.

Abstract: Methods, devices, and processor-readable media for adjusting the control-display gain of a gesture-controlled device are described. Adjusting the control-display gain may facilitate user interaction with content or UI elements rendered on a display screen of the gesture-controlled device. The control-display gain may be adjusted based on a property of how a mid-air dragging gesture is being performed by a user's hand. The property may be the location of the gesture, the orientation of the hand performing the gesture, or the velocity of the gesture. A hand that becomes stationary for a threshold time period while performing the dragging gesture may adjust the control-display gain to a different level. Control-display gain may be set to a different value based on the current velocity of the hand performing the gesture. The control-display gain levels may be selected from a continuous range of values or a set of discrete values. Devices for performing the methods are described.

Abstract: A method for establishing a complex motion controller includes following steps: obtaining a source controller and a destination controller, wherein the source controller is configured to generate a source motion, and the destination controller is configured to generate a destination motion; determining a transition tensor between the source controller and the destination controller, wherein the transition tensor has a plurality of indices, one of the plurality of indices corresponds to a plurality of phases of the source motion; calculating a plurality of transition outcomes of the transition tensor and recording the plurality of transition outcomes according to the plurality of indices; calculating a plurality of transition qualities according to the plurality of transition outcomes; and searching for an optimal transition quality from the plurality of transition qualities to establish a complex motion controller for generating a complex motion corresponding to one of the plurality of phases.

Abstract: Embodiments of the present disclosure provide a method and an apparatus for searching for an image and a related storage medium. The method includes obtaining reference keypoint data of a reference person in a reference image, and searching, based on the reference keypoint data, a set of candidate images for at least one target image containing at least one candidate person that has a pose similar to the reference person.

Abstract: In various examples, the present disclosure describes methods and systems for generating hand state predictions. A hand state prediction system includes a machine learning-based model, such as a neural network model, that is trained to convert inertial motion measurements and surface contact data into predictions of a corresponding hand gesture or gripping posture. For each window of data sampled, motion data and contact data are obtained, processed and fused to generate a fused prediction. The hand state prediction system can operate in continuous mode to automatically detect a start and an end of a hand action, or a user can designate a start and an end of a hand action. A hand state prediction is generated by a multimodal classifier by processing the fused prediction. Instructions represented by a hand action can be acted upon through a command action performed by a computing device or a computer application.

Abstract: Devices and methods for processing detected signals at a detector using a processor are provided. The system involves (i) a data compressor that implements an algorithm for converting a set of data into a compressed set of data, (ii) a machine learning (ML) module coupled to the data compressor, the ML module transforming the compressed set of data into a vector and filtering the vector, (iii) a data encryptor coupled to the ML module that encrypts the filtered vector, and (iv) an integrity protection module coupled to the ML module, wherein the integrity protection module protects the integrity of the filtered vector.

Abstract: A method and device for determining whether an object includes a defect are provided. The method includes the following steps. A tested image of a tested object is obtained. Selected good product sample data corresponding to the tested object is obtained. A dissimilarity value between the tested image and the selected good product sample data is calculated by using a dissimilarity model, and whether the tested object is a good product or a defective product is determined according to the dissimilarity value.

Abstract: Methods, systems, and apparatus for ground plane filtering of video events are disclosed. A method includes obtaining a first set of images of a scene from a camera; determining a ground plane from the first set of images of the scene; obtaining a second set of images of the scene after the first set of images of the scene is obtained; determining that movement shown by a group of pixels in the second set of images of the scene satisfies motion criteria; determining that the ground plane corresponds with at least a portion of the group of pixels; and in response to determining that movement shown by the group of pixels in the second set of images of the scene satisfies motion criteria, and that the ground plane corresponds with at least a portion of the group of pixels, classifying the group of pixels as showing ground plane based motion.

Abstract: There is provided an optical navigation device including an image sensor and a processing unit. The image sensor outputs successive image frames. The processing unit calculates a contamination level and a motion signal based on filtered image frames, and determines whether to update a fixed pattern noise (FPN) stored in a frame buffer according to a level of FPN subtraction, the calculated contamination level and the calculated motion signal to optimize the update of the fixed pattern noise.

Abstract: Implementations described herein relate to methods, devices, and computer-readable media to generate and provide image-based creations. A computer-implemented method includes obtaining a plurality of episodes, each episode associated with a corresponding time period and including a respective set of images and person identifiers for each image. The method further includes forming a respective cluster for each episode that includes at least two person identifiers. The method further includes determining whether one or more person identifiers are included in less than a threshold number of clusters, and in response, removing the one or more person identifiers from the clusters that the one or more person identifiers that are included in. The method further includes merging identical clusters to obtain a plurality of people groups that each include two or more person identifiers and providing a user interface that includes an image-based creation based on a particular people group.

Abstract: Various embodiments described herein support or provide for annotation of a media asset, such as an audio asset or a video asset, based on one or more events identified within content of the media asset. In particular, some embodiments can determine one or more of the following details with respect to content of a given media asset, which can represent annotations that enable determination of contextual information for the given media asset: events; event classification labels for events; subclassifications labels for events; scenes comprising events; attributes of scenes; themes presented by the content; and title-level attributes of the given media asset.

Abstract: An eye center localization method includes performing an image sketching step, a frontal face generating step, an eye center marking step and a geometric transforming step. The image sketching step is performed to drive a processing unit to sketch a face image from the image. The frontal face generating step is performed to drive the processing unit to transform the face image into a frontal face image according to a frontal face generating model. The eye center marking step is performed to drive the processing unit to mark a frontal eye center position information on the frontal face image. The geometric transforming step is performed to drive the processing unit to calculate two rotating variables between the face image and the frontal face image, and calculate the eye center position information according to the two rotating variables and the frontal eye center position information.