CAW: A Remote-Sensing Scene Classification Network Aided by Local Window Attention

Table 6

Overall accuracy and other parameters of the method on the RSSCN7 dataset.

Model	Accuracy (%)	Precision (%)	Recall (%)	Specificity (%)	F1-score (%)
ResNet50 [31]	94.46	94.59	94.09	99.09	94.49
Vgg16 [32]	93.75	93.79	93.76	98.99	93.71
GoogleNet [27]	93.57	93.61	93.57	98.93	93.56
DenseNet [33]	93.21	93.34	93.21	98.89	93.21
ViT-Ti [14]	90.89	90.89	90.89	98.49	90.89
Swin transformer [20]	93.93	93.96	93.91	99.00	93.93
VAN [17]	94.11	94.17	94.11	99.03	94.11
Conformer-Ti [15]	95.00	95.06	95.00	99.20	95.00
CMT [26]	94.82	95.06	94.83	99.14	94.81
MPViT [25]	95.00	95.03	95.00	99.19	95.00
CAW	96.25	96.27	96.24	99.40	96.24