Hardware environment-based data quantization method and apparatus, and readable storage medium

    公开(公告)号:US11748970B2

    公开(公告)日:2023-09-05

    申请号:US17794110

    申请日:2020-11-16

    IPC分类号: G06V10/00 G06V10/28

    CPC分类号: G06V10/28

    摘要: A hardware environment-based data quantization method includes: parsing a model file under a current deep learning framework to obtain intermediate computational graph data and weight data that are independent of a hardware environment; performing calculation on image data in an input data set through a process indicated by an intermediate computational graph to obtain feature map data; separately performing uniform quantization on the weight data and the feature map data of each layer according to a preset linear quantization method, and calculating a weight quantization factor and a feature map quantization factor (S103); combining the weight quantization factor and the feature map quantization factor to obtain a quantization parameter that makes hardware use shift instead of division; and finally, writing the quantization parameter and the quantized weight data to a bin file according to a hardware requirement so as to generate quantized file data (S105).

    ENCODING SYSTEM, ENCODING METHOD, AND COMPUTER-READABLE RECORDING MEDIUM STORING ENCODING PROGRAM

    公开(公告)号:US20230260162A1

    公开(公告)日:2023-08-17

    申请号:US18302834

    申请日:2023-04-19

    申请人: FUJITSU LIMITED

    IPC分类号: G06T9/00 G06V10/28

    CPC分类号: G06T9/00 G06V10/28

    摘要: An encoding system includes: a memory; and a processor coupled to the memory and configured to: calculate, for each area, for a first image, a quantization value that has a compression ratio according to a degree of influence on recognition accuracy during recognition processing; set, when setting the quantization value calculated for each area, for each area of a second image that is acquired after the first image, a quantization value that has a compression ratio lower than the compression ratio, for a specific area other than an area that corresponds to an area of an object to be recognized included in the first image; and encode the second image, using the quantization value.

    IMAGE FRAME SUPER-RESOLUTION IMPLEMENTATION METHOD AND APPARATUS

    公开(公告)号:US20230162323A1

    公开(公告)日:2023-05-25

    申请号:US18151588

    申请日:2023-01-09

    发明人: Linge LI Hao WANG

    IPC分类号: G06T3/40 G06T7/00 G06V10/28

    摘要: This application provides an image frame super-resolution implementation method and apparatus, the method includes: obtaining super-resolution reference information including a quantization parameter and an image quality score set, and the image quality score set includes image quality scores of a plurality of image frames; selecting M image frames from the plurality of image frames based on the image quality score set, where M is greater than or equal to 1; obtaining a video super-resolution network corresponding to the quantization parameter, where the video super-resolution network has a super-resolution function; and inputting the M image frames and a first image frame into the video super-resolution network, where the video super-resolution network is configured to perform super-resolution processing on the first image frame based on the M image frames to obtain a second image frame, and resolution of the second image frame is higher than resolution of the first image frame.