Face Recognition Algorithm Based on Attribute-Driven Loss Function

Shen Li; Hansong Su; Gaohua Liu; Huihua Wu; Meng Wang

doi:10.3788/LOP56.241505

Journals >Laser & Optoelectronics Progress >Volume 56 >Issue 24 >Page 241505 > Article

Laser & Optoelectronics Progress
Vol. 56, Issue 24, 241505 (2019)

Face Recognition Algorithm Based on Attribute-Driven Loss Function

Shen Li, Hansong Su, Gaohua Liu^*, Huihua Wu, and Meng Wang

Author Affiliations

School of Electrical and Information Engineering, Tianjin University, Tianjin 300072, China

show less

DOI: 10.3788/LOP56.241505 Cite this Article Set citation alerts

Shen Li, Hansong Su, Gaohua Liu, Huihua Wu, Meng Wang. Face Recognition Algorithm Based on Attribute-Driven Loss Function[J]. Laser & Optoelectronics Progress, 2019, 56(24): 241505 Copy Citation Text

show less

Fig. 1. Comparison of feature distributions trained by A-Softmax loss function and attribute-driven loss function

Download full size

Influences of super parameters λ and η on verification accuracy. (a) Verification accuracy with same η and different λ; (b) verification accuracy with different η and same λ

Fig. 2. Influences of super parameters λ and η on verification accuracy. (a) Verification accuracy with same η and different λ; (b) verification accuracy with different η and same λ

Download full size

Loss function	Decision boundary
Softmax loss	(W₁-W₂)x+b₁-b₂=0
Modified Softmax loss	‖x‖cosθ₁-‖x‖cosθ₂=0
A-Softmax	class 1: ‖x‖[cos(mθ₁)-cosθ₂]=0class 2: ‖x‖[cosθ₁-cos(mθ₂)]=0

Table 1. Comparison of decision boundaries of different loss functions in binary case

Network structure	LFW /%	CFP-FP /%	AgeDB-30 /%
ResNet50	99.27	91.34	94.31
ResNet101	99.35	92.15	95.82
MobileNet	99.13	90.10	93.88
Inception-ResNet v2	99.67	93.00	97.42
DenseNet	99.54	92.39	96.40
SE-ResNet101	99.48	92.78	96.67

Table 2. Comparison of verification accuracy of different network structures

Loss function type	LFW /%	CFP-FP /%	AgeDB-30 /%
Softmax	97.78	89.64	93.04
Triplet loss	98.65	90.22	95.88
Center loss	99.02	91.10	96.12
L-Softmax loss	99.15	91.90	96.20
A-Softmax loss	99.42	92.80	96.83
Modified A-Softmax loss	99.67	93.00	97.42

Table 3. Verification accuracy of different loss functions

Method	Protocol	Identificationaccuracy /%	Verificationaccuracy /%
Vocord-DeepVo1	Large	75.127	67.318
Google-FaceNet V8	Large	70.496	86.493
Softmax loss	Small	54.628	65.732
Triplet loss	Small	64.698	78.030
Center loss	Small	65.334	80.106
L-Softmax	Small	67.035	80.185
A-Softmax	Small	72.729	85.561
Modified A-Softmax loss	Small	74.531	87.134

Table 4. Accuracy of different loss functions in MegaFace dataset

Shen Li, Hansong Su, Gaohua Liu, Huihua Wu, Meng Wang. Face Recognition Algorithm Based on Attribute-Driven Loss Function[J]. Laser & Optoelectronics Progress, 2019, 56(24): 241505

Download Citation

Set citation alerts for the article

Tools

Set citation alerts for the article

Save the article for my favorites

Paper Information