Title:
FEATURE MAP COMPRESSION METHOD AND APPARATUS
Document Type and Number:
WIPO Patent Application WO/2023/158267
Kind Code:
A1
Abstract:
A neural network-based image processing method and apparatus according to an embodiment of the present invention may: acquire a feature tensor from an input image by using a first neural network including a plurality of neural network layers; acquire a symbol tensor by performing quantization on the acquired feature tensor; and generate a bitstream by performing entropy encoding on the basis of the symbol tensor.
Inventors:
AHN YONGJO (KR)
LEE JONGSEOK (KR)
LEE JONGSEOK (KR)
Application Number:
PCT/KR2023/002344
Publication Date:
August 24, 2023
Filing Date:
February 17, 2023
Export Citation:
Assignee:
INTELLECTUAL DISCOVERY CO LTD (KR)
International Classes:
H04N19/119; G06N3/045; G06N3/08; G06N5/04; H04N19/124; H04N19/13; H04N19/176; H04N19/96
Domestic Patent References:
WO2021260269A1 | 2021-12-30 |
Foreign References:
KR20210136476A | 2021-11-17 | |||
KR20200134155A | 2020-12-01 | |||
US20210211733A1 | 2021-07-08 | |||
KR20190110068A | 2019-09-27 |
Attorney, Agent or Firm:
CHOE, Yun Seo (KR)
Download PDF: