Abstract: A first computing device acquires video data representing a user performing an activity. The first device uses a first pose extraction algorithm to determine a pose of the user within a frame of video data. If the pose is determined to be potentially inaccurate, the user is prompted for authorization to send the frame of video data to a second computing device. If authorization is granted, the second computing device may use a different algorithm to determine a pose of the user and send data indicative of this pose to the first computing device to enable the first computing device to update a score or other output. The second computing device may also use the frame of video data as training data to retrain or modify the first pose extraction algorithm, and may send the modified algorithm to the first computing device for future use.

Abstract: Systems and techniques for producing image-based radiology reports including contextual cropping of image data and radiologist supplied notes and annotations are provided herein. Computer vision and natural language processing algorithms may enable processing of image data and language inputs to identify objects associated with annotations, aid in cropping the image data according to the annotations and object identification and in producing a final text and image laden report.

Abstract: Video output is generated based on first video data that depicts the user performing an activity. Poses of the user during performance of the activity are compared with second video data that depicts an instructor performing the activity. Corresponding poses of the user's body and the instructor's body may be determined through comparison of the first and second video data. The video data is used to determine the rate of motion of the user and to generate video output in which a visual representation of the instructor moves at a rate similar to the that of the user. For example, video output generated based on an instructional fitness video may be synchronized so that movement of the presented instructor matches the rate of movement of the user performing an exercise, improving user comprehension and performance.

Abstract: First image data representing a first human wearing a first article of clothing may be received. The first image data, when rendered on a display, may include a first photometric artifact. A first generator network may be used to generate second image data from the first image data. The first photometric artifact may be removed from the second image data. A second generator network may be used to generate third image data from the second image data, the third image data representing the first human in a different pose relative to the first image data. Fourth image data representing the first article of clothing segmented from the first human may be generated and displayed on a display.

Abstract: Characteristics of a user's movement are evaluated based on performance of activities by a user within a field of view of a camera. Video data representing performance of a series of movements by the user is acquired by the camera. Pose data is determined based on the video data, the pose data representing positions of the user's body while performing the movements. The pose data is compared to a set of existing videos that correspond to known errors to identify errors performed by the user. The errors may be used to generate scores for various characteristics of the user's movement. Based on the errors, exercises or other activities to improve the movement of the user may be determined and included in an output presented to the user.

Abstract: Devices and techniques are generally described for catalog normalization and segmentation for fashion images. First image data representing a first human wearing a first article of clothing may be received. The first image data, when rendered on a display, may include a first photometric artifact. A first generator network may be used to generate second image data from the first image data. The first photometric artifact may be removed from the second image data. A second generator network may be used to generate third image data from the second image data, the third image data representing the first human in a different pose relative to the first image data. Fourth image data representing the first article of clothing segmented from the first human may be generated and displayed on a display.

Abstract: Asymmetries are detected in one or more images by partitioning each image to create a set of patches. Salient patches are identified, and an independent displacement for each patch is identified. The techniques used to identify the salient patches and the displacement for each patch are combined in a function to generate a score for each patch. The scores can be used to identify possible asymmetries.

Abstract: Asymmetries are detected in one or more images by partitioning each image to create a set of patches. Salient patches are identified, and an independent displacement for each patch is identified. The techniques used to identify the salient patches and the displacement for each patch are combined in a function to generate a score for each patch. The scores can be used to identify possible asymmetries.

Abstract: Asymmetries are detected in one or more images by partitioning each image to create a set of patches. Salient patches are identified, and an independent displacement for each patch is identified. The techniques used to identify the salient patches and the displacement for each patch are combined in a function to generate a score for each patch. The scores can be used to identify possible asymmetries.

Abstract: A method comprising using at least one hardware processor for computing a patch distinctiveness score for each of multiple patches of a medical image, computing a shape distinctiveness score for each of multiple regions of the medical image, and computing a saliency map of the medical image, by combining the patch distinctiveness score and the shape distinctiveness score.

Abstract: A method comprising using at least one hardware processor for computing a patch distinctiveness score for each of multiple patches of a medical image, computing a shape distinctiveness score for each of multiple regions of the medical image, and computing a saliency map of the medical image, by combining the patch distinctiveness score and the shape distinctiveness score.

Abstract: A method, executed by one or more processors, includes computing a saliency surface for an image, conducting a mean shift algorithm on the saliency surface to provide mean shift data for the saliency surface, producing a mode voting map for the saliency surface from the mean shift data, and classifying features in the image according to the mode voting map. The features may correspond to medical conditions. In some embodiments, computing the saliency surface comprises determining a distinctiveness score for each of a plurality of image patches. In some embodiments, producing the mode voting map comprises determining a plurality of modes and an area of influence for each mode of the plurality of modes where the area of influence corresponds to mean shift data that leads to a particular mode. A corresponding computer program product and computer system are also disclosed herein.

Abstract: A method for selecting a motion vector includes selecting a candidate block included in a second image that corresponds to a processing block of a first image, computing a first probability for the processing block and the candidate block, selecting a random block for the candidate block, computing a second probability for the processing block and the random block, saving a greater of the first probability and the second probability as a first comparison result, computing a third probability for the processing block, a first neighboring block of the processing block, a matching block in the second image that is matched to the first neighboring block, and a second neighboring block of the matching block, and saving a greater of the first comparison result and the third probability as a second comparison result.

Abstract: A method comprising using at least one hardware processor for computing a patch distinctiveness score for each of multiple patches of a medical image, computing a shape distinctiveness score for each of multiple regions of the medical image, and computing a saliency map of the medical image, by combining the patch distinctiveness score and the shape distinctiveness score.

Abstract: Asymmetries are detected in one or more images by partitioning each image to create a set of patches. Salient patches are identified, and an independent displacement for each patch is identified. The techniques used to identify the salient patches and the displacement for each patch are combined in a function to generate a score for each patch. The scores can be used to identify possible asymmetries.

Abstract: A method, executed by one or more processors, includes computing a saliency surface for an image, conducting a mean shift algorithm on the saliency surface to provide mean shift data for the saliency surface, producing a mode voting map for the saliency surface from the mean shift data, and classifying features in the image according to the mode voting map. The features may correspond to medical conditions. In some embodiments, computing the saliency surface comprises determining a distinctiveness score for each of a plurality of image patches. In some embodiments, producing the mode voting map comprises determining a plurality of modes and an area of influence for each mode of the plurality of modes where the area of influence corresponds to mean shift data that leads to a particular mode. A corresponding computer program product and computer system are also disclosed herein.

Abstract: A method, system and product for image classification. The method comprising obtaining a set of encoding functions and signature values which corresponds to a set of class images, wherein each pair of an encoding function and a signature value corresponds to a class image of the set of class images, wherein the signature value is a value produced by applying the encoding function on the class image; obtaining an image to be classified to a class associated with a class image of the set of class images; with respect to each class image of the set of class images: determining a transformation from the image to the class image; and applying the encoding function using the transformation on the image to produce a value; and automatically determining a class to which the image is classified based on the values and the signature values.

Abstract: A motion estimation system comprises: an initializing module generating a first initial motion field by allocating an initial motion vector to each block of a first motion field related to a first motion change of an image which accompanies a change from an (n?1)-th frame to an n-th frame and generating a second initial motion field allocating an initial motion vector to each block of a second motion field related to a second motion change of the image which accompanies a change from the n-th frame to the (n?1)-th frame; and a candidate test module generating first and second random motion fields based on a similarity function and each block of each of the first and second initial motion fields and random motion vectors, generating first and second spatial propagation motion fields based on the similarity function, and generating first and second optimum motion fields based on the similarity function.

Abstract: A method, system and product for image classification. The method comprising obtaining a set of encoding functions and signature values which corresponds to a set of class images, wherein each pair of an encoding function and a signature value corresponds to a class image of the set of class images, wherein the signature value is a value produced by applying the encoding function on the class image; obtaining an image to be classified to a class associated with a class image of the set of class images; with respect to each class image of the set of class images: determining a transformation from the image to the class image; and applying the encoding function using the transformation on the image to produce a value; and automatically determining a class to which the image is classified based on the values and the signature values.

Abstract: A method comprising using at least one hardware processor for computing a patch distinctiveness score for each of multiple patches of a medical image, computing a shape distinctiveness score for each of multiple regions of the medical image, and computing a saliency map of the medical image, by combining the patch distinctiveness score and the shape distinctiveness score.