Login| Sign Up| Help| Contact|

Patent Searching and Data


Title:
IMAGE PROCESSING DEVICE, IMAGE PROCESSING METHOD, AND IMAGE PROCESSING PROGRAM
Document Type and Number:
WIPO Patent Application WO/2023/032360
Kind Code:
A1
Abstract:
In the present invention, a basic class selection unit (20) selects a basic class for input data on the basis of: an embedding vector output by a basic neural network trained on a basic class; and a center of gravity vector for the basic class. A continuous learning unit (50) uses an additional neural network trained on the basic class to continuously learn an additional class. An additional class selection unit (40) selects an additional class for the input data on the basis of: an embedding vector output by the continuously trained additional neural network; and a center of gravity vector for the basic class and the additional class. A classification determination unit (80) subjects the input data to class classification on the basis of: the basic class selected by the basic class selection unit (20); and the additional class selected by the additional class selection unit (40).

Inventors:
TAKEHARA HIDEKI (JP)
KIDA SHINGO (JP)
YANG YINCHENG (JP)
TAKAMI MAKI (JP)
Application Number:
PCT/JP2022/021174
Publication Date:
March 09, 2023
Filing Date:
May 24, 2022
Export Citation:
Click for automatic bibliography generation   Help
Assignee:
JVCKENWOOD CORP (JP)
International Classes:
G06T7/00; G06N3/08; G06N20/00
Foreign References:
US20210124993A12021-04-29
Other References:
THOMAS MENSINKJAKOB VERBEEKFLORENT PERRONNINGABRIELA CSURKA: "Distance-Based Image Classification: Generalizing to new classes at near-zero cost", IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE, INSTITUTE OF ELECTRICAL AND ELECTRONICS ENGINEERS, vol. 35, no. 11, 2013, pages 2624 - 2637, XP011527040, DOI: 10.1109/TPAMI.2013.83
LU YUBARTLOMIEJ TWARDOWSKIXIALEI LIULUIS HERRANZKAI WANGYONGMEI CHENGSHANGLING JUIJOOST VAN DE WEIJER: "Semantic Drift Compensation for Class-Incremental Learning", COMPUTER VISION AND PATTERN RECOGNITION, 2020, pages 6982 - 6991
HANBIN ZHAOYONGJIAN FUMINTONG KANGQI TIANFEI WUXI LI: "MgSvF: Multi-Grained Slow vs. Fast Framework for Few-Shot Class-Incremental Learning", ARXIV, vol. 2006, 2021, pages 15524
Attorney, Agent or Firm:
MORISHITA Sakaki (JP)
Download PDF: