Skip to main content

Table 3 Performance comparison of ICANet with other algorithms using the FERET database fb and fc represent expression and light changes, and dup1 and dup2 represent age changes

From: ICANet: a simple cascade linear convolution network for face recognition

Method fb fc dup1 dup2
LBP [17] 93.00 51.00 61.00 50.00
DMMA [19] 98.10 98.50 81.60 83.20
G-LBP [20] 98.00 98.00 90.00 85.00
WPCA-POEM [21] 99.60 99.50 88.80 85.00
sPOEM+POD [22] 99.70 100 94.90 94.00
KDCV [20] 98.00 98.00 90.00 85.00
LGXP [23] 99.00 99.00 94.00 93.00
G-LQP [24] 99.90 100.0 93.20 91.00
GOM [25] 99.90 100 95.70 93.10
DFD [26] 99.40 100.0 91.80 92.30
MDML-DCPs [27] 99.75 100.0 96.12 95.73
PCANet [6] 99.58 100 95.43 94.02
ICANet 99.67 100 96.26 94.44
  1. The ICANet recognition performance in all four subsets reached the level of the current algorithm, with the best recognition performance achieved in dup1