公开(公告)号：US20240161488A1

公开(公告)日：2024-05-16

申请号：US18479611

申请日：2023-10-02

Applicant: HUAWEI TECHNOLOGIES CO., LTD.

Inventor： Timofey Mikhailovich Solovyev ,  Elena Alexandrovna Alshina ,  Biao Wang ,  Alexander Alexandrovich Karabutov ,  Mikhail Vyacheslavovich Sosulnikov ,  Georgy Petrovich Gaikov ,  Han Gao ,  Panqi Jia ,  Esin Koyuncu ,  Sergey Yurievich Ikonin ,  Semih Esenlik

IPC: G06V10/82 ,  G06V10/77 ,  G06V20/40 ,  H04N19/513 ,  H04N19/91

CPC classification number: G06V10/82 ,  G06V10/7715 ,  G06V20/46 ,  H04N19/521 ,  H04N19/91

Abstract: This application provides methods and apparatuses for processing of picture data or picture feature data using a neural network with two or more layers. The present disclosure may be applied in the field of artificial intelligence (AI)-based video or picture compression technologies, and in particular, to the field of neural network-based video compression technologies. According to some embodiments, two kinds of data are combined during the processing including processing by the neural network. The two kinds of data are obtained from different stages of processing by the network. Some of the advantages may include greater scalability and a more flexible design of the neural network architecture which may further lead to better encoding/decoding performance.

公开(公告)号：US20240037802A1

公开(公告)日：2024-02-01

申请号：US18479507

申请日：2023-10-02

Applicant: HUAWEI TECHNOLOGIES CO., LTD.

Inventor： Timofey Mikhailovich Solovyev ,  Biao Wang ,  Elena Alexandrovna Alshina ,  Han Gao ,  Panqi Jia ,  Esin Koyuncu ,  Alexander Alexandrovich Karabutov ,  Mikhail Vyacheslavovich Sosulnikov ,  Semih Esenlik ,  Sergey Yurievich Ikonin

IPC: G06T9/00 ,  G06T3/40

CPC classification number: G06T9/002 ,  G06T3/4046

Abstract: This application provides methods and apparatuses for processing of picture data or picture feature data using a neural network with two or more layers. The present disclosure may be applied in the field of artificial intelligence (AI)-based video or picture compression technologies, and in particular, to the field of neural network-based video compression technologies. According to some embodiments, position within the neural network, at which auxiliary information may be entered for processing is selectable based on a gathering condition. The gathering condition may assess whether some prerequisite is fulfilled. Some of the advantages may include better performance in terms of rate and/or disclosure due to the effect of increased flexibility in neural network configurability.

公开(公告)号：US20250008128A1

公开(公告)日：2025-01-02

申请号：US18884321

申请日：2024-09-13

Applicant: HUAWEI TECHNOLOGIES CO., LTD.

Inventor： Timofey Mikhailovich Solovyev ,  Esin Koyuncu ,  Alexander Alexandrovich Karabutov ,  Maxim Borisovitch Sychev ,  Mikhail Vyacheslavovich Sosulnikov ,  Sergey Yurievich Ikonin ,  Elena Alexandrovna Alshina

IPC: H04N19/189 ,  H04N19/13 ,  H04N19/42

Abstract: A neural network including at least one neural network layer and an activation function connected to an output of the at least one neural network layer. The activation function is implemented as an approximation function of a mathematically defined real valued non-linear activation function, wherein the approximation function allows for integer-only processing of fixed-point representations of input values of the approximation function.

公开(公告)号：US20250005331A1

公开(公告)日：2025-01-02

申请号：US18885411

申请日：2024-09-13

Applicant: HUAWEI TECHNOLOGIES CO., LTD.

Inventor： Timofey Mikhailovich Solovyev ,  Sergey Yurievich Ikonin ,  Elena Alexandrovna Alshina ,  Johannes Sauer ,  Esin Koyuncu ,  Maxim Borisovitch Sychev ,  Alexander Alexandrovich Karabutov ,  Mikhail Vyacheslavovich Sosulnikov ,  Kirill Igorevich SOLODSKIKH ,  Vladimir Mikhailovich Kryzhanovskiy ,  Alexander Nikolaevich Filippov

IPC: G06N3/0455 ,  G06T9/00

Abstract: The present disclosure relates to a method of operating a neural network with clipped input data. The method includes defining lower and upper threshold values for integer numbers in data entities of input data for at least one neural network layer. If a value of an integer number in a data entity of the input data is smaller than the defined lower threshold value, the method includes clipping the value of the integer number comprised in the data entity of the input data to the defined lower threshold value. If a value of an integer number in a data entity of the input data is larger than the defined upper threshold value, the method includes clipping the value of the integer number comprised in the data entity of the input data to the defined upper threshold value. Integer overflow of an accumulator register is thereby avoided.

公开(公告)号：US20250005330A1

公开(公告)日：2025-01-02

申请号：US18883907

申请日：2024-09-12

Applicant: HUAWEI TECHNOLOGIES CO., LTD.

Inventor： Timofey Mikhailovich Solovyev ,  Sergey Yurievich Ikonin ,  Elena Alexandrovna Alshina ,  Esin Koyuncu ,  Maxim Borisovitch Sychev ,  Alexander Alexandrovich Karabutov ,  Mikhail Vyacheslavovich Sosulnikov

IPC: G06N3/0455

Abstract: A method of operating a neural network based on conditioned weights includes defining integer lower and upper threshold values for values of integer numbers comprised in data entities of input data for the neural network layer. If a value of an integer numbers comprised in a data entity of the input data is smaller than the lower threshold value, the value of the integer number comprised in the data entity of the input data is clipped to the lower threshold value, or if a value of an integer number comprised in a data entity of the input data is larger than the upper threshold value, the value of the integer number comprised in the data entity of the input data is clipped to the upper threshold value. Integer valued weights are determined based on the lower threshold value, the upper threshold value, and a pre-defined accumulator register size, such that integer overflow of the accumulator register can be avoided.