ETRI-Knowledge Sharing Plaform

KOREAN
논문 검색
Type SCI
Year ~ Keyword

Detail

Conference Paper 영상 기반 광용적맥파에서 데이터 불균형을 고려하는 심층 심박수 추정기
Cited - time in scopus Share share facebook twitter linkedin kakaostory
Authors
이준성
Issue Date
2021-07
Citation
대한전자공학회 학술 대회 (하계) 2021, pp.2557-2560
Publisher
대한전자공학회
Language
Korean
Type
Conference Paper
Abstract
Deep heart rate estimator considering data imbalance is proposed in this paper. The proposed estimator is mainly made up of three parts: 1) video-based photoplethysmography extractor, 2) 1-dimensional heart rate estimator with ResNet-50 backbone, and 3) heart rate focal loss. By applying a different loss value between a heart rate that is easy to predict and otherwise heart rate, the proposed estimator focuses on a heart rate that is hard to predict. The proposed estimator was validated showing 6.61 mean absolute error on VIPL-HR database.
KSP Keywords
Data imbalance, Heart rate, Mean Absolute Error, video based