公开(公告)号：US11948224B2

公开(公告)日：2024-04-02

申请号：US17978573

申请日：2022-11-01

申请人： Intel Corporation

发明人： Elmoustapha Ould-Ahmed-Vall ,  Sara S. Baghsorkhi ,  Anbang Yao ,  Kevin Nealis ,  Xiaoming Chen ,  Altug Koker ,  Abhishek R. Appu ,  John C. Weast ,  Mike B. Macpherson ,  Dukhwan Kim ,  Linda L. Hurd ,  Ben J. Ashbaugh ,  Barath Lakshmanan ,  Liwei Ma ,  Joydeep Ray ,  Ping T. Tang ,  Michael S. Strickland

IPC分类号： G06T1/20 ,  G06F3/14 ,  G06F7/483 ,  G06F9/30 ,  G06F9/38 ,  G06F9/50 ,  G06N3/044 ,  G06N3/045 ,  G06N3/063 ,  G06N3/08 ,  G06N3/084 ,  G06N20/00 ,  G06T1/60 ,  G06T15/00

CPC分类号： G06T1/20 ,  G06F7/483 ,  G06F9/30014 ,  G06F9/30185 ,  G06F9/3863 ,  G06F9/5044 ,  G06N3/044 ,  G06N3/045 ,  G06N3/063 ,  G06N3/084 ,  G06N20/00 ,  G06F3/14 ,  G06T1/60 ,  G06T15/005

摘要： One embodiment provides an apparatus comprising a memory stack including multiple memory dies and a parallel processor including a plurality of multiprocessors. Each multiprocessor has a single instruction, multiple thread (SIMT) architecture, the parallel processor coupled to the memory stack via one or more memory interfaces. At least one multiprocessor comprises a multiply-accumulate circuit to perform multiply-accumulate operations on matrix data in a stage of a neural network implementation to produce a result matrix comprising a plurality of matrix data elements at a first precision, precision tracking logic to evaluate metrics associated with the matrix data elements and indicate if an optimization is to be performed for representing data at a second stage of the neural network implementation, and a numerical transform unit to dynamically perform a numerical transform operation on the matrix data elements based on the indication to produce transformed matrix data elements at a second precision.

公开(公告)号：US20240104833A1

公开(公告)日：2024-03-28

申请号：US18525756

申请日：2023-11-30

申请人： Imagination Technologies Limited

发明人： Xile Yang ,  Robert Brigg ,  John W. Howson

IPC分类号： G06T15/40 ,  G06T1/60 ,  G06T15/00

CPC分类号： G06T15/40 ,  G06T1/60 ,  G06T15/005

摘要： Methods and primitive block generators for generating primitive blocks in a graphics processing system. The methods include receiving transformed position data for a current primitive, the transformed position data indicating a position of the current primitive in rendering space; determining a distance between the position of the current primitive and a position of a current primitive block based on the transformed position data for the current primitive; determining whether to add the current primitive to the current primitive block based on the distance and a fullness of the current primitive block; in response to determining that the current primitive is to be added to the current primitive block, adding the current primitive to the current primitive block; and in response to determining that the current primitive is not to be added to the current primitive block, flushing the current primitive block and adding the current primitive to a new current primitive block.

公开(公告)号：US11941743B2

公开(公告)日：2024-03-26

申请号：US17813709

申请日：2022-07-20

申请人： NVIDIA Corporation

发明人： Matthew Milton Pharr

IPC分类号： G06T15/00 ,  G06F9/38 ,  G06T11/40 ,  G06T15/06 ,  G06T15/08

CPC分类号： G06T15/005 ,  G06F9/3877 ,  G06T11/40 ,  G06T15/06 ,  G06T15/08

摘要： A system and method for generating a set of samples stratified across two-dimensional elementary intervals of a two-dimensional space is disclosed within the application. A computer-implemented technique for generating the set of samples includes selecting an elementary interval associated with a stratification of the two-dimensional space, initializing at least one data structure that indicates valid regions within the elementary interface based on other samples previously placed within the two-dimensional space, and generating a sample in a valid region of the elementary interval utilizing the at least one data structure to identify the valid region prior to generating the sample. In some embodiments, the data structures comprise a pair of binary trees. The process can be repeated for each elementary interval of a selected stratification to generate the set of stratified two-dimensional samples.

公开(公告)号：US11941741B2

公开(公告)日：2024-03-26

申请号：US17708965

申请日：2022-03-30

申请人： MediaTek Inc.

发明人： Chien-Chih Wang ,  Ying-Chieh Chen

IPC分类号： G06T15/00 ,  G06T1/20 ,  G06T1/60 ,  G06T15/06

CPC分类号： G06T15/005 ,  G06T1/20 ,  G06T15/06 ,  G06T1/60

摘要： A graphics system includes an effect engine and a graphics pipeline. The graphics pipeline performs pipeline operations on graphical objects in a frame. The graphics pipeline includes at least a fragment shader stage. An application programming interface (API) provides an instruction that specifies a subset of the graphical objects in the frame for the effect engine to execute. When detecting the instruction, the graphics pipeline invokes the effect engine to perform a predefined set of graphics operations on the subset of the graphical objects in the frame. The predefined set of graphics operations has a higher computational complexity than the pipeline operations.

公开(公告)号：US20240096026A1

公开(公告)日：2024-03-21

申请号：US17844541

申请日：2022-06-20

申请人： Jan Bajana ,  Bernhard Jung ,  Daniel Wagner

发明人： Jan Bajana ,  Bernhard Jung ,  Daniel Wagner

IPC分类号： G06T19/00 ,  G06T7/20 ,  G06T15/00

CPC分类号： G06T19/006 ,  G06T7/20 ,  G06T15/005 ,  G06T2200/28 ,  G06T2207/30196

摘要： A method for reducing motion-to-photon latency for hand tracking is described. In one aspect, a method includes accessing a first frame from a camera of an Augmented Reality (AR) device, tracking a first image of a hand in the first frame, rendering virtual content based on the tracking of the first image of the hand in the first frame, accessing a second frame from the camera before the rendering of the virtual content is completed, the second frame immediately following the first frame, tracking, using the computer vision engine of the AR device, a second image of the hand in the second frame, generating an annotation based on tracking the second image of the hand in the second frame, forming an annotated virtual content based on the annotation and the virtual content, and displaying the annotated virtual content in a display of the AR device.

公开(公告)号：US20240095994A1

公开(公告)日：2024-03-21

申请号：US17946193

申请日：2022-09-16

申请人： NVIDIA Corporation

发明人： Gregory MUTHLER ,  John BURGESS ,  Magnus ANDERSSON ,  Ian KWONG ,  Edward BIDDULPH

IPC分类号： G06T15/06 ,  G06T15/00 ,  G06T15/40

CPC分类号： G06T15/06 ,  G06T15/005 ,  G06T15/40 ,  G06T2210/12 ,  G06T2210/21

摘要： Techniques applicable to a ray tracing hardware accelerator for traversing a hierarchical acceleration structure with reduced false positive ray intersections are disclosed. The reduction of false positives may be based upon one or more of selectively performing a secondary higher precision intersection test for a bounding volume, identifying and culling bounding volumes that degenerate to a point, and parametrically clipping rays that exceed certain configured distance thresholds.

公开(公告)号：US11936883B2

公开(公告)日：2024-03-19

申请号：US17557812

申请日：2021-12-21

申请人： SAMSUNG ELECTRONICS CO., LTD.

发明人： Sumit Panwar ,  Ashish Kumar ,  Daljeet Kaur ,  Harsh Aggarwal

IPC分类号： H04N19/176 ,  G06T1/20 ,  G06T15/00 ,  H04N19/127 ,  H04N19/174

CPC分类号： H04N19/176 ,  G06T1/20 ,  G06T15/005 ,  H04N19/127 ,  H04N19/174

摘要： A differential video rendering system, including a graphics processing unit (GPU); a graphical display coupled to the GPU; a video decoder configured to decode a bitstream of encoded data into a plurality of sets of decoded blocks; at least one processor configured to: generate, based on a first set of the plurality of sets of decoded blocks, a first differential video frame comprising a plurality of sets of differential regions, normalize each set of the plurality of sets of differential regions to a fixed size block to provide a normalized plurality of sets of differential regions, map a respective set of the normalized plurality of sets of differential regions to align with a respective tile size region of a plurality of tile size regions conforming with the GPU, generate a hierarchal region tree based on the normalized plurality of sets of differential regions mapped to the plurality of tile size regions, and generate a plurality of optimal regions based on the hierarchal region tree satisfying a predefined criteria corresponding to a pre-defined optimal number of regions and a predefined efficiency parameter; and a graphics rendering engine configured to render the first differential video frame on the graphical display based on the plurality of optimal regions and a group of differential regions.

公开(公告)号：US11935218B2

公开(公告)日：2024-03-19

申请号：US17646884

申请日：2022-01-04

申请人： City University of Hong Kong

发明人： Tak Wu Sam Kwong ,  Jing Liao ,  Jialu Huang

IPC分类号： G06T15/00 ,  G06N3/045 ,  G06T5/50 ,  G06T7/33

CPC分类号： G06T5/50 ,  G06N3/045 ,  G06T7/33 ,  G06T2207/20081 ,  G06T2207/20084 ,  G06T2207/30201

摘要： A method of providing an image to image translation using a style based generator architecture for generative adversarial networks. The method includes the step of applying a feature alignment loss function defined by:

LossFA=Σi=mnwi·L1(ϕ(GSM)i,ϕ(GTM)i)

wherein GSM and GTM are respective generators of a source model and a target model, ϕ(·)i is the extraction of the ith layer of features and wi represents the weight of L1 loss at the ith layer, the L1 loss weight wi being reduced as layers decrease such that L1 loss at higher layers has a higher weight than the L1 loss at lower layers.

公开(公告)号：US11935176B2

公开(公告)日：2024-03-19

申请号：US18060128

申请日：2022-11-30

申请人： Beijing Bytedance Network Technology Co., Ltd.

发明人： Jiacheng Liu

IPC分类号： G06T15/00

CPC分类号： G06T15/005

摘要： The present disclosure provides a face image displaying method and apparatus, an electronic device and a storage medium. According to the face image displaying method provided in the present application, the interactivity during a face image displaying process is enhanced by means of displaying dynamically a face mask sequence in accordance with a preset motion mode at a preset relative position of a face of an object, and fusing, after a target face mask of the face mask sequence is triggered by a user, the target face mask to the face of the object for display; and the effect of displaying a specific face mask on a face of an object can be achieved by means of fusing, after a target face mask is triggered by a user, the target face mask to the face of the object for display.

公开(公告)号：US11935174B2

公开(公告)日：2024-03-19

申请号：US17809177

申请日：2022-06-27

申请人： DISH Network L.L.C.

发明人： Eric Pleiman ,  Jesus Flores Guerra

IPC分类号： G06T15/00 ,  G06K7/14 ,  G06Q10/1093

CPC分类号： G06T15/00 ,  G06K7/1417 ,  G06Q10/1095

摘要： Processes, systems, and devices generate a training set comprising a first presentation having a first visual aid and a first audio description. The first visual aid and the first audio description are based on initial data retrieved from a first data source using a first indexing technique. The machine-learning system is trained using the first presentation and the initial data retrieved from the first data source using the first indexing technique. The machine-learning system generates a second presentation having a second visual aid and a second audio description. The second visual aid and the second audio description are based on refreshed data retrieved from the first data source using the first indexing technique. The machine-learning system presents the second presentation via an avatar in a virtual meeting room. The avatar is generated by the machine-learning system to present the second visual aid and the second audio description.