Abstract: A method of living body detection includes generating a detection interface in response to a receipt of a living body detection request for a verification of a detection object. The detection interface includes a first region with a viewing target for the detection object to track, a position of the first region in the detection interface changes during a detection time according to a first sequence of a position change of the first region. The method also includes receiving a video stream that is captured during the detection time, determining a second sequence of a sight line change of the detection object based on the video stream, and determining that the detection object is a living body at least partially based on the second sequence of the sight line change of the detection object matching the first sequence of the position change of the first region.

Abstract: The present specification describes a computer-implemented method. According to the method, user-specific tags are generated for a virtual reality (VR) object displayed within a VR environment. The user-specific tags are generated based on an interaction of a first user with the VR object. Role-based access rights are assigned to the user-specific tags. A role of a second user accessing the VR environment is determined and the user-specific tags are presented to the second user, alongside the VR object, based on a comparison of the role of the second user and the role-based access rights.

Abstract: A target object detection method and apparatus are provided. The target object detection method and apparatus are applicable to fields such as artificial intelligence, object tracking, object detection, and image processing. An object is detected from a frame image of a video including a plurality of frame images based on a target template set including one or more target templates.

Abstract: A system to generate an image classifier and test it nearly instantaneously is described herein. Image embeddings generated by an image fingerprinting model are indexed and an associated approximate nearest neighbors (ANN) model is generated. The embeddings in the index are clustered and the clusters are labeled. Users can provide just a few images to add to the index as a labeled cluster. The ANN model is trained to receive an image embedding as input and return a score and label of the most similar identified embedding. The label may be applied if the score exceeds a threshold value. The image classifier can be tested efficiently using Leave One Out Cross Validation (“LOOCV”) to provide near-instantaneous quality indications of the image classifier to the user. Near-instantaneous indications of outliers in the provided images can also be provided to the user using a distance to the centroid calculation.

Abstract: Apparatus, systems and methods for the adaptive passage of a culture of cells and apparatus and methods for dissociating cell colonies are described. The systems may include an imaging module, a pipette module, a handling module, and/or a stage module. Coordinated operation of the modules, optionally in an automated manner, is effected by at least one processor based on one or more characteristics of the culture of cells calculated from one or more images captured at more than one time point. A first apparatus for adaptive passage of a culture cells includes an imaging module and at least one processor, which apparatus may be included in the systems or used in the methods. A second apparatus for dissociating cell colonies, may also be included in the systems or used in the methods, includes impact bumper(s) collidable with impact bracket(s) to transmit a dissociative force to a culture of cells.

Abstract: This application provides a digital person training method and system, and a digital person driving system. According to the method, human-body pose estimation data in training data is extracted, and the human-body pose estimation data is input into an optimized pose estimation network to obtain human-body pose optimization data. Generation losses of position optimization data and acceleration optimization data in the human-body pose optimization data are calculated based on a loss function of the optimized pose estimation network, so as to minimize errors between position estimation data and acceleration estimation data and a real value. In this way, the optimized pose estimation network is driven to update a network parameter to obtain an optimal driving model that is based on the optimized pose estimation network. The errors between the position estimation data and the acceleration estimation data and the real value are minimized.

Abstract: An object detection method includes: obtaining a video to be detected; preprocessing the video to be detected to obtain an image to be detected; inputting the image to be detected into an object detection network; extracting, by the object detection network, a feature map of the image to be detected; performing, by the object detection network, an object prediction on the extracted feature map to obtain a position of an object in the image to be detected and a confidence degree corresponding to the position; and generating a marked object video according to the position of the object in the image to be detected, the confidence degree corresponding to the position, and the video to be detected.

Abstract: A handheld imaging device has a data receiver that is configured to receive reference encoded image data. The data includes reference code values, which are encoded by an external coding system. The reference code values represent reference gray levels, which are being selected using a reference grayscale display function that is based on perceptual non-linearity of human vision adapted at different light levels to spatial frequencies. The imaging device also has a data converter that is configured to access a code mapping between the reference code values and device-specific code values of the imaging device. The device-specific code values are configured to produce gray levels that are specific to the imaging device. Based on the code mapping, the data converter is configured to transcode the reference encoded image data into device-specific image data, which is encoded with the device-specific code values.

Abstract: The provided is a tidal flat extraction method based on hyperspectral data. The method includes: acquiring and preprocessing hyperspectral data; calculating a constraint condition; calculating, according to the constraint condition, a normalized difference tidal flat index (NDTFI), generating a sample index statistical box chart, and determining a tidal flat extraction threshold according to the sample index statistical box chart; and removing a misclassified pixel based on a coastal buffer zone and a high spatial resolution image, thereby achieving final extraction of a tidal flat. The method has the following beneficial effects. The method is an effective supplement to the existing tidal flat extraction methods. The method adopts an easily realized process, improves the accuracy of tidal flat extraction, reflects the real spatial distribution of the tidal flat, and provides a scientific basis for tidal flat management and protection.

Abstract: A system and method of combining different sensor data for higher reliable diagnosis information on a portable mobile device includes using a camera for imaging a region of interest of a subject to obtain an image signal, a microphone for capturing acoustic information from the subject, and one or more processors. The one or more processors can be configured to spectrally analyze the image signal, estimate a first vital-sign of the subject corresponding to a diagnosis using the image signal, analyze the acoustic information, estimate a second vital-sign of the subject corresponding to the diagnosis using the acoustic information, and combine the first vital sign with the second vital-sign to provide a higher confidence level diagnostic of the diagnosis.

Abstract: An illustrative point cloud compression system accesses an input point cloud dataset representative of a point cloud comprising a plurality of points. The point cloud compression system identifies a first attribute dataset and a second attribute dataset within the input point cloud dataset. Based on an application of a transform algorithm to the first and second attribute datasets, respectively, the point cloud compression system generates 1) a first low-frequency component and a first high-frequency component of the first attribute dataset, and 2) a second low-frequency component and a second high-frequency component of the second attribute dataset. The point cloud compression system then generates an output point cloud dataset that prioritizes both the first and second low-frequency components above both the first and second high-frequency components. Corresponding methods and systems are also disclosed.

Abstract: The present invention relates to a method of extracting a roof edge image for solar panel installation by using machine learning, the method comprising: a training step for passing original rooftop image data through a second generation unit of an image extraction system to output an image similar to a target image, and passing image data, from which a rooftop edge has been detected, through a first generation unit of the system to identify the image data from an original image; a step for segmenting an obstruction hiding a roof edge, and receiving, by a second discriminator unit, an image in which the roof edge has been detected; a step for optimizing the weight of a parameter, and training the second generation unit and the second discriminator unit again; and a step for automatically connecting edge portions after extracting edges, and generating a complete roof edge image.

Abstract: A wafer inspection method includes acquiring an inspection image from an edge region of a wafer, generating a color profile of the inspection image in a radial direction of the wafer, detecting a side surface of a layer formed on the wafer based on a change in the color profile in a first direction from a side surface of the wafer toward a center of the wafer, and calculating a distance between the side surface of the layer and the side surface of the wafer.

Abstract: The invention relates to an apparatus for generating or processing an image signal. A first image property pixel structure is a two-dimensional non-rectangular pixel structure representing a surface of a view sphere for the viewpoint. A second image property pixel structure is a two-dimensional rectangular pixel structure and is generated by a processor (305) to have a central region derived from a central region of the first image property pixel structure and at least a first corner region derived from a first border region of the first image property pixel structure. The first border region is a region proximal to one of an upper border and a lower border of the first image property pixel structure. The image signal is generated to include the second image property pixel structure and the image signal may be processed by a receiver to recover the first image property pixel structure.

Abstract: An object detection apparatus comprises circuitry configured to obtain one or more feature maps of a measurement space generated from sensor data of a scene, the sensor data including position information and a feature map representing feature values of a feature at a plurality of positions in the measurement space, transform the one or more feature maps into a template coordinate system, compute a likelihood that the one or more feature maps correspond to an object template and a given candidate object configuration, and iteratively compute likelihoods for different candidate object configurations and determine the object configuration with the highest likelihood.

Abstract: In implementations for free form radius editing, a computing device implements a radius editing system, such as may be integrated with an image editing application. The radius editing system can determine the edge segments for outlines of image objects depicted in a digital image, where the edge segments include corner segments of the image objects. The radius editing system can also determine the radius values of the corner segments of the image objects, and the radius values of the corner segments are maintained in a cache as part of object data corresponding to the image objects depicted in the digital image. The radius editing system can also identify one or more similar corner segments of the image objects that have an equivalent radius value as a selected corner segment responsive to an editing input of a radius of the selected corner segment of an image object.

Abstract: The present disclosure describes techniques for facial expression recognition. A first loss function may be determined based on a first set of feature vectors associated with a first set of images depicting facial expressions and a first set of labels indicative of the facial expressions. A second loss function may be determined based on a second set of feature vectors associated with a second set of images depicting asymmetric facial expressions and a second set of labels indicative of the asymmetric facial expressions. The first loss function and the second loss function may be used to determine a maximum loss function. The maximum loss function may be applied during training of a model. The trained model may be configured to predict at least one asymmetric facial expression in a subsequently received image.

Abstract: Provided is a Face-aware Offset Calculation (FOC) module and method for facial frame interpolation and enhancement and a face video deblurring system and method using the same. The system comprises: a facial frame enhancement device, including a FOC module, for enhancing a target frame; a facial frame interpolation device, including the FOC module, for interpolating the target frame; and a combination device for combining the enhanced target frame with the interpolated target frame.

Abstract: An apparatus for detecting a dimension error obtains an image of a target object, estimates dimensional data for a region of interest (ROI) for which dimensions are to be measured from the image of the target object using a learned dimensional measurement model, and determines whether there is a dimension error in the ROI from the estimated dimension data using a learned dimension error determination model.

Abstract: A method for detecting a traffic sign using a traffic sign detection device is provided. The method includes acquiring point data from a lidar sensor detecting a vicinity of the lidar sensor; clustering a plurality of points based on positions of the plurality of points included in the point data; determining one or more plane object clusters, each of which corresponds to a plane object, among one or more clusters, each of which is formed by clustering the plurality of points; and detecting one or more traffic sign object clusters, each of which corresponds to the traffic sign, among the one or more plane object clusters based on a predetermined traffic sign detection factor.