IP Library › Granted Patent US 12,190,548
Granted Patent B2
US 12,190,548 · App. 17/482,067 · Granted Jan 7, 2025

Method, apparatus, system and computer-readable recording medium for feature information

Inventors: Ji-Hoon Do (Daejeon, KR); Hyoung-Jin Kwon (Daejeon, KR); Dong-Hyun Kim (Daejeon, KR); Youn-Hee Kim (Daejeon, KR); Joo-Young Lee (Daejeon, KR); Se-Yoon Jeong (Daejeon, KR); Jin-Soo Choi (Daejeon, KR); Tae-Jin Lee (Daejeon, KR); Jee-Hoon Kim (Seoul, KR); Dong-Gyu Sim (Seoul, KR); Seoung-Jun Oh (Seongnam-si Gyeonggi-do, KR); Min-Hun Lee (Uijeongbu-si Gyeonggi-do, KR); Yun-Gu Lee (Seongnam-si Gyeonggi-do, KR); Han-Sol Choi (Dongducheon-si Gyeonggi-do, KR); Kwang-Hwan Kim (Gwangju, KR)
Assignee: ELECTRONICS AND TELECOMMUNICATIONS RESEARCH INSTITUTE
G06T9/002G06N3/02G06T3/4046G06V10/40
View Patent ↗
Loading inventors, assignments & file history…
Monitor This Case
Get email alerts when status or documents change.
Order Certified Copies
Most orders are placed with the USPTO same day — all within 24 business hours.
Order via The Patent Place →
Pre-filled with this patent's details
Quick Facts
Patent No.
US 12,190,548
App. No.
17/482,067
Filed
Sep 22, 2021
Granted
Jan 7, 2025
Kind
B2
Art Unit
2668
USPC
382/232
Abstract

There are provided an apparatus, method, system, and recording medium for performing selective encoding/decoding on feature information. An encoding apparatus generates residual feature information. The encoding apparatus transmits the residual feature information to a decoding apparatus through a residual feature map bitstream. The residual feature information is the difference between feature information extracted from an original image and feature information extracted from a reconstructed image. Feature information of the reconstructed image is generated using the reconstructed image. Reconstructed feature information is generated using the feature information of the reconstructed image and reconstructed residual feature information.

Claims (35)

1. An encoding method comprising:

extracting first feature information from a reconstructed image;

extracting second feature information from an original image; and

generating residual feature information using the first feature information and the second feature information, wherein

the first feature information and the second feature information are extracted using feedback information from a decoding apparatus, and

the feedback information is used to determine a specific part of the original image to be encoded.

2. An encoding method comprising:

extracting first feature information from a reconstructed image;

extracting second feature information from an original image; and

generating residual feature information using the first feature information and the second feature information, wherein

a preprocessed image is generated by performing first preprocessing on the original image,

the first feature information is extracted from the preprocessed image,

the first preprocessing comprises at least one of a resolution subsampling, a temporal subsampling and a spatial subsampling.

3. The encoding method of claim 2 , further comprising:

generating preprocessed residual feature information by performing second preprocessing on the residual feature information, wherein

the second preprocessing includes one or more of sub-sampling, domain transformation, and domain rearrangement.

4. A non-transitory computer-readable recording medium in which a program for performing the encoding method of claim 2 is recorded.

5. A decoding method, comprising:

generating reconstructed residual feature information by performing decoding on information of a residual feature information bitstream; and

generating reconstructed feature information using feature information of a reconstructed image and the reconstructed residual feature information, wherein

postprocessed reconstructed residual feature information is generated by performing postprocessing on the reconstructed residual feature information, and

the reconstructed feature information is generated using the feature information of the reconstructed image and the postprocessed reconstructed residual feature information;

the postprocessing includes one or more of inverse rearrangement, and inverse domain transformation.

6. The decoding method of claim 5 , further comprising:

generating the reconstructed image using an image bitstream; and

generating the feature information of the reconstructed image from the reconstructed image.

7. The decoding method of claim 6 , further comprising:

generating a postprocessed reconstructed image by performing postprocessing on the reconstructed image.

8. The decoding method of claim 7 , wherein the postprocessing includes one or more of inverse transformation of a color format and up-sampling.

9. The decoding method of claim 5 , wherein the feature information of the reconstructed image is a result from an intermediate layer of a neural network, among multiple layers of the neural network.

10. The decoding method of claim 5 , wherein the postprocessing includes up-sampling, and

the up-sampling is performed using a sampling phase and a sampling rate.

11. The decoding method of claim 5 , wherein the postprocessing includes up-sampling, and

the up-sampling is fixed sampling or non-fixed sampling.

12. A non-transitory computer-readable recording medium in which a program for performing the decoding method of claim 5 is recorded.

Assignments (1)
ASSIGNMENT OF ASSIGNOR'S INTEREST Recorded Sep 22, 2021
From: DO, JI-HOON; KWON, HYOUNG-JIN; KIM, DONG-HYUN; KIM, YOUN-HEE; LEE, JOO-YOUNG; JEONG, SE-YOON; CHOI, JIN-SOO; LEE, TAE-JIN; KIM, JEE-HOON; SIM, DONG-GYU; OH, SEOUNG-JUN; LEE, MIN-HUN; LEE, YUN-GU; CHOI, HAN-SOL; KIM, KWANG-HWAN
To: ELECTRONICS AND TELECOMMUNICATIONS RESEARCH INSTITUTE
Reel/Frame 057565/0965 →
Priority Claims (2)
KR 10-2020-0123346 · Sep 23, 2020 · national
KR 10-2021-0124646 · Sep 17, 2021 · national
Continuity (1)
Related Publication 20220092827A1 · Mar 24, 2022
References Cited (25)
US 8126282B2 · Jung et al. · 2012 [cited by applicant]
US 9961344B2 · Im et al. · 2018 [cited by applicant]
US 10666944B2 · Navarrete Michelini et al. · 2020 [cited by applicant]
US 11190784B2 · Kim et al. · 2021 [cited by applicant]
US 11683490B2 · Zhao · 2023 [cited by examiner]
US 11706426B2 · Choi · 2023 [cited by examiner]
US 11783508B2 · Mammou · 2023 [cited by examiner]
US 11792438B2 · Li · 2023 [cited by examiner]
US 11798196B2 · Kim · 2023 [cited by examiner]
US 11805244B2 · Choi · 2023 [cited by examiner]
US 20190246102A1 · Cho et al. · 2019 [cited by applicant]
US 20210203997A1 · Veselov · 2021 [cited by examiner]
US 20230119758A1 · Li · 2023 [cited by examiner]
US 20230308665A1 · Choi · 2023 [cited by examiner]
US 20230353790A1 · Wang · 2023 [cited by examiner]
US 20230412787A1 · Kim · 2023 [cited by examiner]
US 20240007602A1 · Kim · 2024 [cited by examiner]
US 20240007653A1 · Kim · 2024 [cited by examiner]
US 20240007671A1 · Kim · 2024 [cited by examiner]
US 20240015282A1 · Choi · 2024 [cited by examiner]
KR 1020090083716A · 2009 [cited by applicant]
KR 1020160096460A · 2016 [cited by applicant]
KR 1020200009118A · 2020 [cited by applicant]
WO WO2020055279A1 · 2020 [cited by examiner]
Ling-Yu Duan et al., Video Coding for Machines: A Paradigm of Collaborative Compression and Intelligent Analytics, Jan. 13, 2020, pp. 1-15, Computer Science IEEE Transactions on Image Processing. [cited by applicant]