Cross-Age Face Recognition Based on Deep Learning

Wenbin Sun; Rong Wang; Lianzhu Sun; Yuansong Lin

doi:10.3788/LOP202259.0215001

Journals >Laser & Optoelectronics Progress >Volume 59 >Issue 2 >Page 0215001 > Article

Laser & Optoelectronics Progress
Vol. 59, Issue 2, 0215001 (2022)

Cross-Age Face Recognition Based on Deep Learning

Wenbin Sun¹, Rong Wang^2、3、*, Lianzhu Sun⁴, and Yuansong Lin¹

Author Affiliations

¹School of Information Engineering, Guangdong University of Technology, Guangzhou , Guangdong 510006, China

²College of Information Engineering, Northwest A&F University, Xianyang , Shaanxi 712100, China

³National Engineering Research Center for Information Technology in Agriculture, Beijing 100097, China

⁴School of Traffic and Transportation, Beijing Jiaotong University, Beijing 100044, China

show less

DOI: 10.3788/LOP202259.0215001 Cite this Article Set citation alerts

Wenbin Sun, Rong Wang, Lianzhu Sun, Yuansong Lin. Cross-Age Face Recognition Based on Deep Learning[J]. Laser & Optoelectronics Progress, 2022, 59(2): 0215001 Copy Citation Text

show less

Fig. 1. Structure of CA-CNN model

Download full size

Fig. 2. Structure of CBAM and ECBAM. (a) CBAM; (b) ECBAM

Download full size

Fig. 3. BKCCA module

Download full size

Fig. 4. Face alignment effect

Download full size

Fig. 5. Variation curve of loss value and rank-1 accuracy during training

Download full size

Fig. 6. Rank-1 curves of different models during training

Download full size

Fig. 7. Some examples of failed retrievals

Download full size

$λ_{1}$	$λ_{2}$	Accuracy of Rank-1 /%
0.01	1	99.03
0.005	1	99.00
0.015	1	98.80
0.01	0.5	98.56
0.02	1	98.67

Table 1. Recognition rate of Rank-1 for different values of

λ_{1}

and

λ_{2}

Algorithm	Accuracy of Rank-1 /%
HFA（2013）^［10］	91.14
MEFA（2015）^［11］	93.80
LF-CNN（2016）^［12］	97.51
AE-CNN（2017）^［14］	98.13
AFJT-CNN（2018）^［34］	97.85
JMCNN（2018）^［26］	98.36
OE-CNN（2018）^［15］	98.67
DAL（2019）^［16］	98.97
FSDS-CNN（2020）^［32］	98.41
Baseline	95.67
Baseline+Age	97.73
Proposed algorithm	99.03

Table 2. Rank-1 results of different algorithms on Morph Album 2

Algorithm	Training image size /Million	EER /%	$F N M R @ F M R (0.1)$ /%
VGG-Face（2015）^［35］	2.6	13.5	17.6
Nosiy softmax（2017）^［36］	0.5	17.5	29.2
AFJT-CNN（2018）^［34］	0.7	14.8	21.8
FSDS-CNN（2020）^［32］	0.5	10.1	10.2
Proposed algorithm	0.5	9.8	10.0

Table 3. Evaluation results of different algorithms on CALFW dataset

Wenbin Sun, Rong Wang, Lianzhu Sun, Yuansong Lin. Cross-Age Face Recognition Based on Deep Learning[J]. Laser & Optoelectronics Progress, 2022, 59(2): 0215001

Download Citation

Set citation alerts for the article

Tools

Set citation alerts for the article

Save the article for my favorites

Paper Information