Spectral-Spatial Attention Transformer with Dense Connection for Hyperspectral Image Classification

Table 7

Classification results of different methods for the Kennedy Space Center dataset.

Class	1D-CNN	M3D-DCNN	SC-FR	pResNet	DBDA	Proposed
1	87.81	98.75	99.45	99.84	99.91	99.84
2	84.65	98.60	100.00	100.00	100.00	100.00
3	90.71	98.39	99.82	100.00	100.00	100.00
4	57.69	93.85	100.00	98.65	99.81	99.81
5	49.69	95.62	100.00	100.00	99.69	100.00
6	60.00	98.97	100.00	100.00	100.00	100.00
7	81.90	100.00	100.00	100.00	100.00	99.52
8	88.18	99.18	99.96	100.00	100.00	100.00
9	94.12	99.84	99.88	100.00	100.00	100.00
10	95.00	99.46	100.00	100.00	100.00	100.00
11	94.79	99.91	99.95	100.00	100.00	100.00
12	88.32	99.70	100.00	100.00	100.00	100.00
13	100.00	100.00	90.00	99.92	100.00	100.00
OA (%)	91.54 ± 0.64	99.36 ± 0.25	97.27 ± 7.75	99.92 ± 0.08	99.97 ± 0.05	99.96 ± 0.06
Variance (OA)	0.4113	0.0631	60.1004	0.0061	0.0021	0.0032
AA (%)	82.53 ± 1.63	98.64 ± 0.67	99.16 ± 2.29	99.88 ± 0.17	99.95 ± 0.09	99.94 ± 0.12
Variance (AA)	2.6728	0.4429	5.2215	0.0288	0.0073	0.0133
KAPPA × 100	90.07 ± 0.75	99.24 ± 0.30	96.87 ± 8.87	99.91 ± 0.09	99.97 ± 0.05	99.95 ± 0.07