Unrestricted Face Recognition Algorithm Based on Transfer Learning on Self-Pickup Cabinet

Table 1

Comparison of recognition rate on the CASIA-WebFace dataset.

Methods	Training set size/MB	Test of time/s	Recognition rate/%
FaceNet [8]	199.8	0.61	99.18
DeepFace [22]	3.99	0.29	97.27
DDML [23]	2.53	0.38	98.81
Ouamane [24]	0.68	0.52	98.99
DeepID2 [6]	0.27	0.32	98.87
Faster RCNN [15]	0.49	0.185	97.42
Proposed	0.49	0.049	99.52