Table 2.

The overall accuracy of different models in Indian Pines ¹.

Model\Amount of Training Samples	20%	10%	8%	6%	4%	2%
2D-CNN	91.23 ± 0.21	83.86 ± 1.00	82.43 ± 0.62	78.59 ± 0.79	74.92 ± 0.98	67.13 ± 1.12
M3D-DCNN	90.03 ± 2.18	80.10 ± 4.56	78.04 ± 2.13	74.57 ± 2.63	70.48 ± 3.21	62.28 ± 3.18
HybridSN	99.30 ± 0.18	97.66 ± 0.23	96.37 ± 1.19	93.64 ± 4.40	91.88 ± 1.39	83.14 ± 1.60
Model A	99.21 ± 0.24	96.94 ± 0.59	97.03 ± 0.43	93.80 ± 0.97	91.20 ± 1.61	81.97 ± 2.28
Model B	99.12 ± 0.20	96.47 ± 0.59	95.09 ± 0.52	91.46 ± 1.40	87.25 ± 1.07	73.81 ± 3.33
R-HybridSN	99.52 ± 0.16	98.44 ± 0.44	98.12 ± 0.35	96.69 ± 0.70	94.99 ± 0.39	86.67 ± 1.02

¹ The standard deviation is shown after ±.