专利检索 ap:("SEOUL NATIONAL UNIVERSITY R&DB FOUNDATION") AND inv:"Ji Yea CHON" 第 1 页

1.

发明公开
ALGORITHM AND METHOD FOR DYNAMICALLY VARYING QUANTIZATION PRECISION OF DEEP LEARNING NETWORK 审中-公开

公开(公告)号：US20240062537A1

公开(公告)日：2024-02-22

申请号：US18270638

申请日：2020-12-31

申请人： SEOUL NATIONAL UNIVERSITY R&DB FOUNDATION

发明人： Ook Sang YOO , Hyuk Jae LEE , Soo Jung RYU , Ji Yea CHON , Kyeong Jong LIM

IPC分类号： G06V10/70 , G06V10/82 , G06V10/764

CPC分类号： G06V10/87 , G06V10/82 , G06V10/764

摘要： An image recognition method includes the steps of: for a deep learning network that carries out object recognition on a random image, carrying out quantization corresponding to the number of a plurality of different bits to generate a plurality of quantization models respectively corresponding to the number of bits; receiving image data as an input for the deep learning network; determining the uncertainty of the input image data; selecting any one of the plurality of quantization models on the basis of the determined uncertainty; and recognizing an object from the image data by using the selected quantization model, and outputting, as the result of the object recognition, a label corresponding to the image data.

2.

发明公开
ALGORITHM AND METHOD FOR DYNAMICALLY CHANGING QUANTIZATION PRECISION OF DEEP-LEARNING NETWORK 审中-公开

公开(公告)号：US20240071070A1

公开(公告)日：2024-02-29

申请号：US18270649

申请日：2021-12-30

申请人： SEOUL NATIONAL UNIVERSITY R&DB FOUNDATION

发明人： Ook Sang YOO , Ji Yea CHON , Hyuk Jae LEE , Kyeong Jong LIM

IPC分类号： G06V10/70 , G06V10/28 , G06V10/764 , G06V10/82 , G06V20/70

CPC分类号： G06V10/87 , G06V10/28 , G06V10/764 , G06V10/82 , G06V20/70

摘要： Provided is an image recognition method including the steps of: for a deep learning network that carries out object recognition on a random image, carrying out quantization corresponding to the number of a plurality of different bits to generate a plurality of quantization models respectively corresponding to the number of bits; receiving image data as an input for the deep learning network; determining the uncertainty of the input image data; selecting any one of the plurality of quantization models on the basis of the determined uncertainty; and recognizing an object from the image data by using the selected quantization model, and outputting, as the result of the object recognition, a label corresponding to the image data.