IP Library Granted Patent US 12,659,486
Granted Patent B2
US 12,659,486 · App. 18/844,628 · Granted Jun 16, 2026

Feature encoding/decoding method, device, recording medium storing bitstream, and method for transmitting bitstream

Inventors: Chulkeun Kim (Seoul, KR); Jaehyun Lim (Seoul, KR); Donggyu Gwak (Seoul, KR)
Assignee: LG Electronics Inc.
H04N19/136G06V10/7715H04N19/169H04N19/70
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 12,659,486
App. No.
18/844,628
Granted
Jun 16, 2026
Kind
B2
Abstract

Provided are a feature encoding/decoding method and device, a recording medium on which a bitstream generated by the feature encoding method is stored, and a method for transmitting the bitstream. The feature decoding method according to the present disclosure comprises the steps of: determining whether or not an input source is feature data; determining a feature map type of the feature data based on the input source being the feature data; obtaining encoding information of the feature data from a bitstream; and decoding the feature data based on the feature map type and the encoding information, and whether or not the input source is the feature data may be determined based on input format identification information that is obtained from a sequence parameter set.

Claims (21)

1 . A feature decoding method performed by a feature decoding apparatus, the feature decoding method comprising:

determining, based on input format identification information signaled from a bitstream, at least one of whether or not an input source is feature data or a feature map type of the feature data,

wherein the input format identification information equal to a first value indicates that the input source is image data, the input format identification information equal to a second value indicates that the input source is the feature data and the feature map type of the feature data is a first feature map type, and the input format identification information equal to a third value indicates that the input source is the feature data and the feature map type of the feature data is a second feature map type;

obtaining encoding information of the feature data from the bitstream based on the input source being the feature data; and

decoding the feature data based on the feature map type and the encoding information.

2 . The feature decoding method of claim 1 , wherein the encoding information includes supplementary information related to a method for constructing a feature map of the feature data and meaning information related to a method for extracting and reconstructing the feature data.

3 . The feature decoding method of claim 1 , wherein the encoding information includes data property information indicating a privacy protection level.

4 . A feature encoding method performed by a feature encoding apparatus, the feature encoding method comprising:

generating input format identification information indicating at least one of whether or not an input source is feature data;

or a feature map type of the feature data,

wherein the input format identification information equal to a first value indicates that the input source is image data, the input format identification information equal to a second value indicates that the input source is the feature data and the feature map type of the feature data is a first feature map type, and the input format identification information equal to a third value indicates that the input source is the feature data and the feature map type of the feature data is a second feature map type;

generating encoding information of the feature data based on the input source being the feature data; and

encoding the input format identification information and the encoding information into a bitstream.

5 . The feature encoding method of claim 4 , wherein the encoding information includes supplementary information related to a method for constructing a feature map of the feature data and meaning information related to a method for extracting and reconstructing the feature data.

6 . A computer-readable recording medium storing a bitstream generated by the feature encoding method of claim 4 .

7 . A method for transmitting a bitstream, comprising:

generating input format identification information indicating at least one of whether or not an input source is feature data;

or a feature map type of the feature data,

wherein the input format identification information equal to a first value indicates that the input source is image data, the input format identification information equal to a second value indicates that the input source is the feature data and the feature map type of the feature data is a first feature map type, and the input format identification information equal to a third value indicates that the input source is the feature data and the feature map type of the feature data is a second feature map type;

generating encoding information of the feature data based on the input source being the feature data; and

encoding the input format identification information and the encoding information into the bitstream.

Assignments (1)
ASSIGNMENT OF ASSIGNOR'S INTEREST Recorded Sep 26, 2024
From: KIM, CHULKEUN; LIM, JAEHYUN; GWAK, DONGGYU
To: LG ELECTRONICS INC.
Reel/Frame 068704/0681 →
Priority Claims (1)
KR 10-2022-0028829 · Mar 7, 2022 · national
Continuity (1)
Related Publication 20250234005A1 · Jul 17, 2025
References Cited (23)
US 11838551B2 · Goswami · 2023 [cited by examiner]
US 20030108099A1 · Nagumo · 2003 [cited by examiner]
US 20210182111A1 · Jakobsson · 2021 [cited by examiner]
US 20210203997A1 · Veselov et al. · 2021 [cited by applicant]
US 20210297685A1 · Karczewicz · 2021 [cited by examiner]
US 20220286459A1 · Dohare · 2022 [cited by examiner]
US 20220292730A1 · Gao · 2022 [cited by examiner]
US 20230085554A1 · Kim et al. · 2023 [cited by applicant]
US 20230370087A1 · Henry · 2023 [cited by examiner]
AU 2021202140A1 · 2022 [cited by examiner]
CN 113223101A · 2021 [cited by applicant]
KR 1020210147830A · 2021 [cited by applicant]
WO WO2012124461A1 · 2012 [cited by examiner]
WO 2021050007A1 · 2021 [cited by applicant]
WO 2021172956A1 · 2021 [cited by applicant]
WO WO2021177652A1 · 2021 [cited by applicant]
WO WO2021205065A1 · 2021 [cited by applicant]
WO 2022226087A1 · 2022 [cited by applicant]
International Search Report and Written Opinion in International Appln. No. PCT/KR2023/003073, mailed on May 31, 2023, 8 pages (with partial English translation). [cited by applicant]
Choi et al., “Deep Feature Compression for Collaborative Object Detection,” Paper, Presented at the Proceedings of the 25th IEEE International Conference on Image Processing (ICIP), Athens, Greece, Oct. 7-10, 2018, pp. … [cited by applicant]
Extended European Search Report in European Appln. No. 23767118.5, mailed on Dec. 8, 2025, 17 pages. [cited by applicant]
Han et al., “[VCM] Investigation on feature map channel reordering and compression for object detection,” ISO/IEC JTC 1/SC 29/WG 2, MPEG Technical Requirements, Online, m56653, Apr. 2021, 5 pages. [cited by applicant]
Zhang et al., “IN/CM] Updates to use cases and requirements for video coding for machines,” ISO/IEC JTC 1/SC 29/WG 2, MPEG Technical Requirements, Online, m57648, Jul. 2021, 18 pages. [cited by applicant]