Intel Corporation

Beijing Xiaomi Mobile Software Co., Ltd.

Lenovo (Beijing) Limited

Ming Lu

Anbang Yao

Yurong Chen

Shandong Wang

Wenlong Li

Xiaofeng Tong

Yiwen Guo

Yikai Wang

Qiang Li

Hao Zhao

Yikai Fang

Chen Ling

Yumeng Wang

Yangyuxuan Kang

Haihua Lin

Feng Chen

Sungye Kim

JianSheng Chen

Liwei Liao

Bin Zhang

Scott R Janus

Xiaoming Chen

Yang Liu

Chao Yu

Runyu Shi

Tao Kong

Attila Tamas Afra

Junjie Huang

Hongzhi Tao

Wenjian Shao

Zhiming Wang

Zhiqiao Zheng

Haojun Xu

Yiwei He

Chong Wong

Yuanke Luo

Tao Lv

Techniques related to 3D pose estimation from a 2D input image are discussed. Such techniques include incrementally adjusting an initial 3D pose generated by applying a lifting network to a detected 2D pose in the 2D input image by projecting each current 3D pose estimate to a 2D pose projection, applying a residual regressor to features based on the 2D pose projection and the detected 2D pose, and combining a 3D pose increment from the residual regressor to the current 3D pose estimate.

Incremental 2D-to-3D pose lifting for fast and accurate human pose estimation

A data processing method includes, based on a transfer control parameter corresponding to a storage location transfer event of distributed business data in a distributed storage system within a first time, predicting a first transfer control parameter usable by the distributed storage system within a second time for the transfer event, obtaining a second transfer control parameter that the distributed storage system is capable of providing within the second time in a case that the storage location transfer event does not occur, and controlling a transfer of the distributed business data within the second time based on the first transfer control parameter and the second transfer control parameter.

Data processing method and apparatus for data transfer in distributed storage system, device, and storage medium

Techniques related to implementing and training image classification networks are discussed. Such techniques include applying shared convolutional layers to input images regardless of resolution and applying normalization selectively based on the input image resolution. Such techniques further include training using mixed image size parallel training and mixed image size ensemble distillation.

Input image size switchable network for adaptive runtime efficient image classification

Apparatus and methods for three-dimensional pose estimation are disclosed herein. An example apparatus includes an image synchronizer to synchronize a first image generated by a first image capture device and a second image generated by a second image capture device, the first image and the second image including a subject; a two-dimensional pose detector to predict first positions of keypoints of the subject based on the first image and by executing a first neural network model to generate first two-dimensional data and predict second positions of the keypoints based on the second image and by executing the first neural network model to generate second two-dimensional data; and a three-dimensional pose calculator to generate a three-dimensional graphical model representing a pose of the subject in the first image and the second image based on the first two-dimensional data, the second two-dimensional data, and by executing a second neural network model.

Apparatus and methods for three-dimensional pose estimation

Embodiments are generally directed to methods and apparatuses of spatially sparse convolution module for visual rendering and synthesis. An embodiment of a method for image processing, comprising: receiving an input image by a convolution layer of a neural network to generate a plurality of feature maps; performing spatially sparse convolution on the plurality of feature maps to generate spatially sparse feature maps; and upsampling the spatially sparse feature maps to generate an output image.

Method and apparatus of spatially sparse convolution module for visual rendering and synthesis

Techniques are disclosed for providing improved pose tracking of a subject using a 2D camera and generating a 3D image that recreates the pose of the subject. A 3D skeleton map is estimated from a 2D skeleton map of the subject using, for example, a neural network. A template 3D skeleton map is accessed or generated having bone segments that have lengths set using, for instance, anthropometry statistics based on a given height of the template 3D skeleton map. An improved 3D skeleton map is then produced by at least retargeting one or more of the plurality of bone segments of the estimated 3D skeleton map to more closely match the corresponding template bone segments of the template 3D skeleton map. The improved 3D skeleton map can then be animated in various ways (e.g., using various skins or graphics) to track corresponding movements of the subject.

Kinematic interaction system with improved pose tracking

Methods and apparatus for discrimitive semantic transfer and physics-inspired optimization in deep learning are disclosed. A computation training method for a convolutional neural network (CNN) includes receiving a sequence of training images in the CNN of a first stage to describe objects of a cluttered scene as a semantic segmentation mask. The semantic segmentation mask is received in a semantic segmentation network of a second stage to produce semantic features. Using weights from the first stage as feature extractors and weights from the second stage as classifiers, edges of the cluttered scene are identified using the semantic features.

Methods and apparatus for discriminative semantic transfer and physics-inspired optimization of features in deep learning

The present disclosure provides an apparatus and method of guided neural network model for image processing. An apparatus may comprise a guidance map generator, a synthesis network and an accelerator. The guidance map generator may receive a first image as a content image and a second image as a style image, and generate a first plurality of guidance maps and a second plurality of guidance maps, respectively from the first image and the second image. The synthesis network may synthesize the first plurality of guidance maps and the second plurality of guidance maps to determine guidance information. The accelerator may generate an output image by applying the style of the second image to the first image based on the guidance information.

Apparatus and method of guided neural network model for image processing

Embodiments are generally directed to methods and apparatuses for determining a frontal body orientation. An embodiment of a method for determining a three-dimensional (3D) orientation of frontal body of a player comprises: detecting each of a plurality of players in each of a plurality of frames captured by a plurality of cameras; for each of the plurality of cameras, tracking each of the plurality of players between continuous frames captured by the camera; and associating the plurality of frames captured by the plurality of cameras to generate the 3D orientation of each of the plurality of players.

Method and apparatus for determining a frontal body orientation

Embodiments are generally directed to an adaptive deformable kernel prediction network for image de-noising. An embodiment of a method for de-noising an image by a convolutional neural network implemented on a compute engine, the image including a plurality of pixels, the method comprising: for each of the plurality of pixels of the image, generating a convolutional kernel having a plurality of kernel values for the pixel; generating a plurality of offsets for the pixel respectively corresponding to the plurality of kernel values, each of the plurality of offsets to indicate a deviation from a pixel position of the pixel; determining a plurality of deviated pixel positions based on the pixel position of the pixel and the plurality of offsets; and filtering the pixel with the convolutional kernel and pixel values of the plurality of deviated pixel positions to obtain a de-noised pixel.

Adaptive deformable kernel prediction network for image de-noising

Methods, systems and apparatuses may provide for technology that selects a player from a plurality of players based on an automated analysis of two-dimensional (2D) video data associated with a plurality of cameras, wherein the selected player is nearest to a projectile depicted in the 2D video data. The technology may also track a location of the selected player over a subsequent plurality of frames in the 2D video data and estimate a location of the projectile based on the location of the selected player over the subsequent plurality of frames.

Technology to automatically locate ball-holding players in multi-camera video feeds

An information transmission method and an electronic device are provided. The information transmission method is applied to an ultrasonic wave transmitting device, and includes: transmitting a Bluetooth wake-up signal; obtaining an ultrasonic transmission signal based on information to be transmitted; and transmitting the ultrasonic transmission signal to an ultrasonic wave receiving device, wherein the Bluetooth wake-up signal is configured to wake up the ultrasonic wave receiving device to receive the ultrasonic transmission signal. In the ultrasonic wave receiving device, such elements as a decoder and an acoustic wave receiver may be in a normally-off state; and when it is needed to transmit information through an ultrasonic wave, the ultrasonic wave transmitting device first transmits the Bluetooth wake-up signal; and upon the reception of the Bluetooth wake-up signal, the ultrasonic wave receiving device wakes up related elements to receive the ultrasonic signal.

Information transmission method and electronic device

Embodiments are generally directed to methods and apparatuses for determining a game status. An embodiment of a method for determining a game status comprises: detecting players and a ball in a plurality of continuous frames captured by a plurality of cameras; obtaining tracking data of the players and the ball of the plurality of continuous frames; and determining a game status for each of the plurality of continuous frames based on the tracking data of the players and the ball of the plurality of continuous frames.

Method and apparatus for determining a game status

A multi-camera architecture for detecting and tracking a ball in real-time. The multi-camera architecture includes network interface circuitry to receive a plurality of real-time videos taken from a plurality of high-resolution cameras. Each of the high-resolution cameras simultaneously captures a sports event, wherein each of the plurality of high-resolution cameras includes a viewpoint that covers an entire playing field where the sports event is played. The multi-camera architecture further includes one or more processors coupled to the network interface circuitry and one or more memory devices coupled to the one or more processors. The one or more memory devices includes instructions to determine the location of the ball for each frame of the plurality of real-time videos, which when executed by the one or more processors, cause the multi-camera architecture to simultaneously perform one of a detection scheme or a tracking scheme on a frame from each of the plurality of real-time videos to detect the ball used in the sports event and perform a multi-camera build to determine a location of the ball in 3D for the frame from each of the plurality of real-time videos using one of detection or tracking results for each of the cameras.

Multi-cam ball location method and apparatus

Methods and apparatus are disclosed for enhancing a binary weight neural network using a dependency tree. A method of enhancing a convolutional neural network (CNN) having binary weights includes constructing a tree for obtained binary tensors, the tree having a plurality of nodes beginning with a root node in each layer of the CNN. A convolution is calculated of an input feature map with an input binary tensor at the root node of the tree. A next node is searched from the root node of the tree and a convolution is calculated at the next node using a previous convolution result calculated at the root node of the tree. The searching of a next node from root node is repeated for all nodes from the root node of the tree, and a convolution is calculated at each next node using a previous convolution result.

Growing community of inventors

Beijing, China

Ming Lu