Qualcomm Incorporated

Yunke Pan

Shuxue Quan

Ying Hai Chen

Ning Bi

Lei Xu

Dashan Gao

Yingen Xiong

Xin Zhong

Stephen Michael Verrall

Jian Wei

Bolan Jiang

Xin Li

Provided are systems, methods, and computer-readable medium for operating a neural network. In various implementations, the neural network can receive an input image that includes an object to be identified. The neural network can generate a plurality of initial feature maps using a convolution layers, wherein a first initial feature maps is generated using the input image. The neural network can generate an up-sampled feature map using a de-convolution layer that takes an initial feature map as input, where the up-sampled feature map has a same resolution as the previous initial feature map. The neural network can combine the up-sampled feature map and the previous initial feature map, and use the combined feature map to more accurate identify the object.

Combining convolution and deconvolution for object detection

Methods and an apparatus for background subtraction are described. According to some implementations, the temporal history of a pixel location can be used to classify whether a current input pixel of a current input video frame is a background pixel or a foreground pixel. The classification can be made based on the number of matching pixels in corresponding pixel locations in background history planes. One of the background history planes can be updated according to the classified background pixels and persistent classified foreground pixels.

Methods and apparatus for video background subtraction

A structured light active sensing systems may be configured to transmit and received codewords to generate a depth map by analyzing disparities between the locations of the transmitted and received codewords. To determine the locations of received codewords, an image of the projected codewords is identified, from which one or more codeword boundaries are detected. The codeword boundaries may be detected based upon a particular codeword bit of each codeword. Each detected codeword boundary may be constrained from overlapping with other detected codeword boundaries, such that no pixel of the received image is associated with more than one codeword boundary.

Methods and apparatus for codeword boundary detection for generating depth maps

A method performed by an electronic device is described. The method includes interleaving multiple input image channels to produce an interleaved multi-channel input. The method also includes loading the interleaved multi-channel input to a single-instruction multiple data (SIMD) processor. The method further includes convolving the interleaved multi-channel input with a multi-channel filter.

Systems and methods for image processing in a deep convolution network

A method performed by an electronic device is described. The method includes determining overlapping areas from neighboring images. The method also includes determining a difference measure between the overlapping areas. The method further includes determining a constraint measure corresponding to at least one of the overlapping areas. The method additionally includes determining a seam based on a combination of the difference measure and the constraint measure.

Growing community of inventors

San Diego, CA, United States of America

Yunke Pan