The original paper is in English. Non-English content has been machine-translated and may contain typographical errors or mistranslations. ex. Some numerals are expressed as "XNUMX".
Copyrights notice
The original paper is in English. Non-English content has been machine-translated and may contain typographical errors or mistranslations. Copyrights notice
본 논문에서는 얼굴 인식을 위한 보다 설명적이고 강력한 얼굴 표현을 학습하기 위한 JMM-CNN(Joint Multi-Patch and Multi-Task Convolutional Neural Networks) 프레임워크를 제시합니다. 제안된 JMM-CNN에서는 전역 및 국소 얼굴 특징을 학습하고 융합하기 위해 다중 패치 CNN 세트와 특징 융합 네트워크를 구성한 후 얼굴 인식 작업, 포즈 추정 작업을 포함한 다중 작업 학습 알고리즘이 작동됩니다. 얼굴 인식 작업을 위한 포즈 불변 얼굴 표현을 얻기 위해 융합된 특징을 사용합니다. 학습된 얼굴 표현의 포즈 무감각성을 더욱 향상시키기 위해 두 작업의 특징에 대한 유사성 정규화 용어를 도입하여 정규화 손실을 제안합니다. 또한 JMM-CNN이 엔드투엔드 네트워크 아키텍처를 갖도록 간단하지만 효과적인 패치 샘플링 전략이 적용됩니다. Multi-PIE 데이터 세트에 대한 실험을 통해 제안된 방법의 효율성을 입증했으며, LFW(Labeled Face in the Wild), YTF(YouTube Faces) 및 MegaFace Challenge에서 최첨단 방법과 비교하여 경쟁력 있는 성능을 달성했습니다.
Yanfei LIU
Chongqing University of Technology
Junhua CHEN
Chongqing University of Posts and Telecommunications
Yu QIU
Chongqing Industry Polytechnic College
The copyright of the original papers published on this site belongs to IEICE. Unauthorized use of the original or translated papers is prohibited. See IEICE Provisions on Copyright for details.
부
Yanfei LIU, Junhua CHEN, Yu QIU, "Joint Multi-Patch and Multi-Task CNNs for Robust Face Recognition" in IEICE TRANSACTIONS on Information,
vol. E103-D, no. 10, pp. 2178-2187, October 2020, doi: 10.1587/transinf.2020EDP7059.
Abstract: In this paper, we present a joint multi-patch and multi-task convolutional neural networks (JMM-CNNs) framework to learn more descriptive and robust face representation for face recognition. In the proposed JMM-CNNs, a set of multi-patch CNNs and a feature fusion network are constructed to learn and fuse global and local facial features, then a multi-task learning algorithm, including face recognition task and pose estimation task, is operated on the fused feature to obtain a pose-invariant face representation for the face recognition task. To further enhance the pose insensitiveness of the learned face representation, we also introduce a similarity regularization term on features of the two tasks to propose a regularization loss. Moreover, a simple but effective patch sampling strategy is applied to make the JMM-CNNs have an end-to-end network architecture. Experiments on Multi-PIE dataset demonstrate the effectiveness of the proposed method, and we achieve a competitive performance compared with state-of-the-art methods on Labeled Face in the Wild (LFW), YouTube Faces (YTF) and MegaFace Challenge.
URL: https://global.ieice.org/en_transactions/information/10.1587/transinf.2020EDP7059/_p
부
@ARTICLE{e103-d_10_2178,
author={Yanfei LIU, Junhua CHEN, Yu QIU, },
journal={IEICE TRANSACTIONS on Information},
title={Joint Multi-Patch and Multi-Task CNNs for Robust Face Recognition},
year={2020},
volume={E103-D},
number={10},
pages={2178-2187},
abstract={In this paper, we present a joint multi-patch and multi-task convolutional neural networks (JMM-CNNs) framework to learn more descriptive and robust face representation for face recognition. In the proposed JMM-CNNs, a set of multi-patch CNNs and a feature fusion network are constructed to learn and fuse global and local facial features, then a multi-task learning algorithm, including face recognition task and pose estimation task, is operated on the fused feature to obtain a pose-invariant face representation for the face recognition task. To further enhance the pose insensitiveness of the learned face representation, we also introduce a similarity regularization term on features of the two tasks to propose a regularization loss. Moreover, a simple but effective patch sampling strategy is applied to make the JMM-CNNs have an end-to-end network architecture. Experiments on Multi-PIE dataset demonstrate the effectiveness of the proposed method, and we achieve a competitive performance compared with state-of-the-art methods on Labeled Face in the Wild (LFW), YouTube Faces (YTF) and MegaFace Challenge.},
keywords={},
doi={10.1587/transinf.2020EDP7059},
ISSN={1745-1361},
month={October},}
부
TY - JOUR
TI - Joint Multi-Patch and Multi-Task CNNs for Robust Face Recognition
T2 - IEICE TRANSACTIONS on Information
SP - 2178
EP - 2187
AU - Yanfei LIU
AU - Junhua CHEN
AU - Yu QIU
PY - 2020
DO - 10.1587/transinf.2020EDP7059
JO - IEICE TRANSACTIONS on Information
SN - 1745-1361
VL - E103-D
IS - 10
JA - IEICE TRANSACTIONS on Information
Y1 - October 2020
AB - In this paper, we present a joint multi-patch and multi-task convolutional neural networks (JMM-CNNs) framework to learn more descriptive and robust face representation for face recognition. In the proposed JMM-CNNs, a set of multi-patch CNNs and a feature fusion network are constructed to learn and fuse global and local facial features, then a multi-task learning algorithm, including face recognition task and pose estimation task, is operated on the fused feature to obtain a pose-invariant face representation for the face recognition task. To further enhance the pose insensitiveness of the learned face representation, we also introduce a similarity regularization term on features of the two tasks to propose a regularization loss. Moreover, a simple but effective patch sampling strategy is applied to make the JMM-CNNs have an end-to-end network architecture. Experiments on Multi-PIE dataset demonstrate the effectiveness of the proposed method, and we achieve a competitive performance compared with state-of-the-art methods on Labeled Face in the Wild (LFW), YouTube Faces (YTF) and MegaFace Challenge.
ER -