A Two-Stream Deep Fusion Framework for High-Resolution Aerial Scene Classification

Table 4

Comparison with the state-of-the-art methods on the WHU-RS dataset.

Methods	Training ratios
	40%	60%
Bag of SIFT [45]	-	85.52 ± 1.23
MS-CLBP + BoVW [40]	-	89.29 ± 1.30
GoogLeNet [41]	93.12 ± 0.82	94.71 ± 1.33
VGG-VD-16 [41]	95.44 ± 0.60	96.05 ± 0.91
CaffeNet [41]	95.11 ± 1.20	96.24 ± 0.56
salM^3LBP-CLM [42]	95.35 ± 0.76	96.38 ± 0.82
TEX-Net-LF [43]	97.61 ± 0.36	98.00 ± 0.52
DCA by addition [44]	-	98.70 ± 0.22
Ours	98.23 ± 0.56	98.92 ± 0.52