公开(公告)号：US20230007282A1

公开(公告)日：2023-01-05

申请号：US17897808

申请日：2022-08-29

Applicant: HUAWEI TECHNOLOGIES CO., LTD.

Inventor： Cui HU ,  Jingwei CHEN ,  Liang CHEN ,  Juanpeng HE ,  Shuo LIAN ,  Shuangquan WANG

IPC: H04N19/30 ,  H04N19/436 ,  H04N19/172 ,  H04N19/136

Abstract: Embodiments of this application provide an image transmission method and apparatus. The method includes: converting a first high-resolution image into a first low-resolution image, where first resolution of the first high-resolution image is higher than second resolution of the first low-resolution image; encoding the first low-resolution image to obtain a first bitstream; obtaining a second high-resolution image, where third resolution of the second high-resolution image is higher than the second resolution, and the second high-resolution image includes high-frequency information of the first high-resolution image and excludes low-frequency information of the first high-resolution image; obtaining an image residual between the first high-resolution image and the second high-resolution image, where the image residual is used to reflect the low-frequency information of the first high-resolution image; encoding the image residual to obtain a second bitstream; and sending the first bitstream and the second bitstream.

公开(公告)号：US20210192349A1

公开(公告)日：2021-06-24

申请号：US17197677

申请日：2021-03-10

Applicant: HUAWEI TECHNOLOGIES CO., LTD.

Inventor： Shuo LIAN ,  Jingwei CHEN

IPC: G06N3/08 ,  G06K9/62

Abstract: Example methods and apparatus for quantizing a neural network model in a device are described. In one example method, user calibration data is obtained and input into the neural network model to calculate a quantization parameter of each of a plurality of layers of the neural network model. To-be-quantized data is input into the neural network model to quantize input data of each of the plurality of layers by using the quantization parameter of each of the plurality of layers. Because the user calibration data is generated based on data generated by the device used by a user, the quantization parameter can be obtained online, and the quantization parameter matches user data in the device.