Login| Sign Up| Help| Contact|

Patent Searching and Data


Title:
METHOD AND APPARATUS FOR ENCODING/DECODING DEEP LEARNING NETWORK
Document Type and Number:
WIPO Patent Application WO/2021/118140
Kind Code:
A1
Abstract:
The present invention relates to a method and apparatus for encoding/decoding a deep learning network. According to an embodiment of the present invention, the method for decoding a deep learning network, comprises the steps of: decoding network header information about the deep learning network; decoding layer header information about a plurality of layers included in the deep learning network; decoding layer data information about detailed information of the plurality of layers; and acquiring the deep learning network and the plurality of layers included in the deep learning network, wherein the layer header information includes layer classification information about classifying the plurality of layers.

Inventors:
CHOI BYEONG HO (KR)
LEE SANG SEOL (KR)
JANG SUNG JOON (KR)
KIM SUNG JEI (KR)
Application Number:
PCT/KR2020/017270
Publication Date:
June 17, 2021
Filing Date:
November 30, 2020
Export Citation:
Click for automatic bibliography generation   Help
Assignee:
KOREA ELECTRONICS TECHNOLOGY (KR)
International Classes:
H04N19/30; G06N3/04; G06N3/08; H04N19/433; H04N19/70
Foreign References:
US20190197420A12019-06-27
KR20180052069A2018-05-17
KR20180013674A2018-02-07
KR20180096779A2018-08-29
Other References:
PAL CHANDRAJIT; PANKAJ SUNIL; AKRAM WASIM; ACHARYYA AMIT; BISWAS DWAIPAYAN: "Modified Huffman based compression methodology for Deep Neural Network Implementation on Resource Constrained Mobile Platforms", 2018 IEEE INTERNATIONAL SYMPOSIUM ON CIRCUITS AND SYSTEMS (ISCAS), IEEE, 27 May 2018 (2018-05-27), pages 1 - 5, XP033434681, DOI: 10.1109/ISCAS.2018.8351234
Attorney, Agent or Firm:
SUNG, Byung Kee (KR)
Download PDF: