Title:
CONVOLUTIONAL NEURAL NETWORK FOR IN-LOOP FILTER OF VIDEO ENCODER BASED ON DEPTH-WISE SEPARABLE CONVOLUTION
Document Type and Number:
WIPO Patent Application WO/2024/077740
Kind Code:
A1
Abstract:
A method of generating an enhanced frame at a video encoder is provided. The method may include receiving, by a weakly connected dense attention block (WCDAB) backbone (403) of a convolutional neural network (CNN) of an in-loop filter (400), a first set of feature extractions as an input. The first set of feature extractions may be associated with a reconstructed frame (402a). The method may include applying, by the WCDAB backbone (403) of the CNN of the in-loop filter (400), a plurality of depth-wise separable convolutions to the first set of feature extractions to generate a set of global features. The method also provides an efficient multi-stage training strategy based on progressive learning. The multi-stage training strategy maximizes the performance of the proposed network.
More Like This:
| WO/2018/171265 | IMAGE FILTERING METHOD AND APPARATUS |
| JP2021034848 | IMAGE DECODING DEVICE |
| JP2011049740 | IMAGE PROCESSING APPARATUS AND METHOD |
Inventors:
JUNG CHEOLKON (CN)
ZHANG HAO (CN)
ZHANG HAO (CN)
Application Number:
PCT/CN2022/136567
Publication Date:
April 18, 2024
Filing Date:
December 05, 2022
Export Citation:
Assignee:
GUANGDONG OPPO MOBILE TELECOMMUNICATIONS CORP LTD (CN)
International Classes:
H04N19/82
Domestic Patent References:
| WO2022072659A1 | 2022-04-07 |
Foreign References:
| US20210012537A1 | 2021-01-14 | |||
| CN114208203A | 2022-03-18 | |||
| CN113068031A | 2021-07-02 | |||
| CN113132729A | 2021-07-16 |
Other References:
DING DANDAN; KONG LINGYI; CHEN GUANGYAO; LIU ZOE; FANG YONG: "A Switchable Deep Learning Approach for In-Loop Filtering in Video Coding", IEEE TRANSACTIONS ON CIRCUITS AND SYSTEMS FOR VIDEO TECHNOLOGY, IEEE, USA, vol. 30, no. 7, 15 August 2019 (2019-08-15), USA, pages 1871 - 1887, XP011796773, ISSN: 1051-8215, DOI: 10.1109/TCSVT.2019.2935508
Attorney, Agent or Firm:
SCIHEAD IP LAW FIRM (CN)
Download PDF: