Abstract: Disclosed is an image processing apparatus that includes one or more processors; and a memory storing instructions, which when executed by the one or more processors, cause the one or more processors to: generate distribution data indicating a frequency distribution of horizontal distance values of a range image based on the range image having pixel values according to distance of an object in a plurality of captured images; detect an object based on the distribution data with respect to a range image; predict a predicted position of the object in a current frame based on the distribution data with respect to range images of a plurality of previous frames; and integrate a plurality of objects detected by the detector based on the predicted position to track an integrated object in subsequent frames.

Abstract: A style transfer neural network may be used to generate stylized synthetic images, where real images provide the style (e.g., seasons, weather, lighting) for transfer to synthetic images. The stylized synthetic images may then be used to train a recognition neural network. In turn, the trained neural network may be used to predict semantic labels for the real images, providing recognition data for the real images. Finally, the real training dataset (real images and predicted recognition data) and the synthetic training dataset are used by the style transfer neural network to generate stylized synthetic images. The training of the neural network, prediction of recognition data for the real images, and stylizing of the synthetic images may be repeated for a number of iterations. The stylization operation more closely aligns a covariate of the synthetic images to the covariate of the real images, improving accuracy of the recognition neural network.

Abstract: There is provided an information processing apparatus, an information processing method, and a program capable of further reducing the unpleasant feeling caused by the disturbance of the display of the virtual object, the information processing apparatus including: a prediction accuracy estimation unit that estimates a degree of prediction accuracy related to prediction of a position or a posture of a real object; and an output control unit that outputs information for causing a display unit to display a virtual object corresponding to the real object on a basis of an estimation result of the prediction accuracy estimation unit.

Abstract: A device may receive images of an object and information identifying the object, process the images using an artificial intelligence technique to identify parts of the object that are depicted in the images, and receive information identifying a location of damage on the object and information regarding the damage on the object. The device may process the information identifying the location of damage to identify a damaged part of the object, identify images depicting the damaged part, and identify, in the images, a location of the damaged part. The device may generate a first content item for display at the location of the damaged part in the images and generate a second content item for display with the images based on user interaction with the first content item, where the second content item includes information based on the information regarding the damage on the object.

Abstract: Low spatial frequencies of an original image and an upscaled filtered image are analyzed. Differences will be observed in the low frequency components of the two images in the general case since the pixel art upscaler filter as a side effect introduces low frequency changes. A modification to images produced by the PAU is applied to attempt to match the brightness of the original images in the low frequency spectrum. From a viewer perspective (e.g., based on typical blurring visual effects), the original image and the modified filtered image will look the same—demonstrating that there is no low frequency brightness creep or resulting increased photosensitivity concerns.

Abstract: One embodiment can provide a system for detecting outlines of objects in images. During operation, the system receives an image that includes at least one object, generates a random noise signal, and provides the received image and the random noise signal to a shape-regressor module, which applies a shape-regression model to predict a shape outline of an object within the received image.

Abstract: Dental images processed with artificial intelligence for e-commerce is described. In an example scenario, a processing device receives a dental image. The dental image is processed with a deep neural network and matched to anatomy and pathology datasets in real time. The matched and identified dental image is matched to a dental treatment recommendation dataset to further produce a real time dental product recommendation for a patient. One or more artificial intelligence entities may provide in real time dental treatment recommendations and/or real time dental product recommendations to a dental professional and/or an individual. The process may also produce a diagnostic treatment aid and/or a product recommendation such as an orthodontic aligner.

Abstract: Systems and methods for automatically excluding artifacts from an analysis of a biological specimen image are disclosed. An exemplary method includes obtaining an immunohistochemistry (IHC) image and a control image, determining whether the control image includes one or more artifacts, upon a determination that the control image includes one or more artifacts, identifying one or more artifact regions within the IHC image by mapping the one or more artifacts from the control image to the IHC image, and performing image analysis of the IHC image where any identified artifact regions are excluded from the image analysis.

Abstract: Apparatus and methods are described including receiving, via a computer processor, at least one image of a portion of a subject's body. One or more features that are present within the image of the portion of the subject's body, and that were artificially added to the image subsequent to acquisition of the image, are identified. In response thereto, an output is generated on an output device.

Abstract: Method for processing of a grey scale image, in particular a dim grey scale image, comprising the following steps: a) receiving an initial grey scale image, said initial grey scale image having a plurality of pixels at an initial resolution, b) calculating parameters characterizing the luminance (gain, median_grey, var_grey) and the noise level (X, noise_estimate, radius_spatial_summation, grid_size, threshold_var) of the initial grey scale image of step a), c) creating a basic intermediate image, d) creating an averaged intermediate image, and e) creating an enhanced grey scale image by interpolation of pixels based on the averaged receptors (greyAvg) of the averaged intermediate image of step d).

Abstract: In an example, an apparatus comprises logic, at least partially including hardware logic, to save one or more outputs of a deep learning neural network in a storage system of an autonomous vehicle and upload the one or more outputs to a remote server. Other embodiments are also disclosed and claimed.

Abstract: Disclosed herein are system, method, and computer program product embodiments for providing object detection and filtering operations. An embodiment operates by receiving an image comprising a plurality of pixels and pixel information for each pixel. The pixel information indicates a bounding box corresponding to an object within the image associated with a respective pixel and a confidence score associated with the bounding box for the respective pixel. Pixels that do not correspond to a center of at least one of the bounding boxes are iteratively removed from the plurality of pixels until a subset of pixels each of which correspond to a center of at least one of the bounding boxes remains. Based on the subset, a final bounding box associated with each object of the image is determined based on an overlapping of the bounding boxes of the subset of pixels and the corresponding confidence scores.

Abstract: A facial tracking method can include receiving a first vector of a first frame, and second vectors of second frames that are prior to the first frame in a video. The first vector is formed by coordinates of first facial feature points in the first frame and determined based on a facial registration method. Each second vector is formed by coordinates of second facial feature points in the respective second frame and previously determined based on the facial tracking method. A second vector of the first frame is determined according to a fitting function based on the second vectors of the first set of second frames. The fitting function has a set of coefficients that are determined by solving a problem of minimizing a function formulated based on a difference between the second vector and the first vector of the current frame, and a square sum of the coefficients.

Abstract: Described is a system for visual activity recognition. In operation, the system detects a set of objects of interest (OI) in video data and determines an object classification for each object in the set of OI, the set including at least one OI. A corresponding activity track is formed for each object in the set of OI by tracking each object across frames. Using a feature extractor, the system determines a corresponding feature in the video data for each OI, which is then used to determine a corresponding initial activity classification for each OI. One or more OI are then detected in each activity track via foveation, with the initial object detection and foveated object detection thereafter being appended into a new detected-objects list. Finally, a final classification is provided for each activity track using the new detected-objects list and filtering the initial activity classification results using contextual logic.

Abstract: A guidance processing apparatus (100) includes an information acquisition unit (101) that acquires a plurality of different pieces of guidance information on the basis of states of a plurality of people within one or more images, and a control unit (102) that performs control of a plurality of target devices present in different spaces or time division control of a target device so as to set a plurality of different states corresponding to the plurality of pieces of guidance information.

Abstract: Visual data is used to locate a device in the real world. The location and orientation of the device is detected by matching the camera view of the device to a computer-generated view of an area surrounding the location. The computer-generated view of the area is constructed by combining various remotely sensed geospatial data including satellite imagery, aerial LIDAR, map data, GIS inventory data, and/or the like. In various embodiments, the disclosed system can not only locate the device but can also provide measurements regarding where the user is looking and 3D measurement of the scene in the device camera view. In various embodiments, the presented invention enables devices to have a system that provides spatial intelligence to any camera.

Abstract: Photorealistic image stylization concerns transferring style of a reference photo to a content photo with the constraint that the stylized photo should remain photorealistic. Examples of styles include seasons (summer, winter, etc.), weather (sunny, rainy, foggy, etc.), lighting (daytime, nighttime, etc.). A photorealistic image stylization process includes a stylization step and a smoothing step. The stylization step transfers the style of the reference photo to the content photo. A photo style transfer neural network model receives a photorealistic content image and a photorealistic style image and generates an intermediate stylized photorealistic image that includes the content of the content image modified according to the style image. A smoothing function receives the intermediate stylized photorealistic image and pixel similarity data and generates the stylized photorealistic image, ensuring spatially consistent stylizations.

Abstract: A method for image reconstruction may include: obtaining a plurality of sets of scan data captured by a magnetic resonance imaging (MRI) device, each set of scan data corresponding to a same scanning area of an object and corresponding to a plurality of scanning characteristics; generating one or more shareable data sets based on the plurality of sets of scan data; generating, based on the one or more shareable data sets, at least one optimized data set for each of the plurality of scanning characteristics; and reconstructing, based on at least one optimized data set for at least one of the plurality of scanning characteristics, the plurality of sets of scan data to obtain a reconstructed image for the at least one scanning characteristic.

Abstract: An image processing apparatus includes: a calculation unit configured to calculate deformation information of an object deformed from a first deformation state to a second deformation state; a degree-of-deviation calculation unit configured to calculate a degree of deviation of the deformation information with respect to a deformation model representing a deformation state of the object; and a display control unit configured to display the calculated degree of deviation.

Abstract: The present application relates to a method for acquiring maximum principal strain or a maximum principal stress status of a vessel wall.