Abstract: Provided is a method for classifying data in an electronic apparatus, including obtaining target data, obtaining first classification information by using a classifier set including a plurality of classifiers based on the target data, obtaining second classification information by using a neural network model based on the target data, comparing the first classification information and the second classification information, and verifying the classifier set based on a result of comparing the first classification information and the second classification information.

Abstract: Disclosed is a method of processing information in an electronic apparatus, the method including acquiring a neural network model, determining a reference format for conversion of the neural network model, and converting the neural network model to a model of the reference format, wherein the model converted into the reference format is executed in a neural processing unit (NPU).

Abstract: A method, apparatus and computer program product are provided to generate a panoramic view derived from multiple cameras and automatically place a seam in that panoramic view in a computationally efficient manner. In regards to a method, images captured by at least two cameras are received. Each camera has a different, but partially overlapping field of view. The method determines a seam location and scale factor to be used when combining the images together to minimize errors at the seam between the two images. In some example implementations, the seam location and scale factor may be recalculated in response to a manual or automatic trigger. In some additional example implementations, motion associated with an image element near a seam location is detected, and the seam location is moved in a direction opposite that of the direction of motion.

Abstract: A method, apparatus and computer program product are provided to define the location of a center point associated with each frame within a stream of image data. Metadata associated with the orientation, such as the pitch and yaw, of a camera is synchronized on a frame-by-frame basis with a related stream of image data. In connection with receiving the image data and the orientation data, the method defines a center point associated with a video frame images and transmits a control signal causing the reorientation of at least a subset of video image frames. In some example implementations, arising in the context of 360° video streams, the rotation of the head of a viewer of virtual reality content or other 360° video streams may be taken into account when defining the location of a center point.

Abstract: A method including determining an aggregate cost for pixels in a frame under construction; applying a disparity computation for the pixels in the frame under construction; and storing the frame in a memory. Determining the aggregate cost for the pixels in a frame under construction includes use of at least one previous aggregate cost for the pixels in a previous frame, where the aggregate cost is determined with use of edge-aware filtering and a minimum spanning tree (MST). The disparity computation includes use of a confidence based temporal filtering of disparities at least partially using a disparity computation for the pixels in the previous frame.

Abstract: Various methods are provided for facilitating live virtual reality streaming, and more specifically, for facilitating dynamic metadata transmission, stream tiling, and attention based active view processing, encoding, and rendering. One example method may receiving an indication of a position of a display unit, determining, based on the indication of the position of the display unit, at least one active view associated with the position of the display, the at least one active view being a first view of a plurality of views, and causing transmission of first video content corresponding to the at least one active view, the first video content configured for display on the display unit.

Abstract: A method including determining an aggregate cost for pixels in a frame under construction; applying a disparity computation for the pixels in the frame under construction; and storing the frame in a memory. Determining the aggregate cost for the pixels in a frame under construction includes use of at least one previous aggregate cost for the pixels in a previous frame, where the aggregate cost is determined with use of edge-aware filtering and a minimum spanning tree (MST). The disparity computation includes use of a confidence based temporal filtering of disparities at least partially using a disparity computation for the pixels in the previous frame.

Abstract: Cluster centers of an image are initialized. The image includes a plurality of pixels. Pixels of the plurality of pixels in a pattern are labeled. The cluster centers are recomputed based on the labeling of the plurality of pixels.

Abstract: In an example embodiment, a method, apparatus and computer program product are provided. The method includes facilitating receipt of an image, and partitioning the image into a plurality of sub-images. The method further includes super-pixels in the plurality of sub-images, wherein determining the super-pixels includes determining a plurality of super-pixels in each sub-image of the plurality of sub-images. An image graph including a plurality of connections between the super-pixels is determined, wherein a connection of the plurality of connections is determined between a first super-pixel and a second super-pixel. The first super-pixel belongs to a sub-image and the second super-pixel belongs to the sub-image or a neighboring sub-image of the sub-image located in an immediate vicinity of the sub-image in the image.

Abstract: Disclosed herein is a method. Cluster centers of an image are initialized. The image includes a plurality of pixels. Pixels of the plurality of pixels in a pattern are labeled. The cluster centers are recomputed based on the labeling of the plurality of pixels.