[Retracted] Hybrid Depth-Separable Residual Networks for Hyperspectral Image Classification

Table 9

The comparison of experiments in the UP dataset.

No. of classes	Train/test	SVM-RBF	2D CNN	3D CNN	SSRN	HybridSN	HDSRN
1	663/5968	93.68	97.37	97.4	99.75	99.76	99.78
2	1865/16784	97.02	99.26	94.73	99.79	99.78	99.8
3	210/1889	82.41	80.73	95.05	98.29	99.01	99.82
4	306/2758	96.51	95.54	98.04	99.52	99.53	99.56
5	135/1211	98.38	99.75	99.01	99.82	99.87	99.9
6	503/4526	90.01	93.14	98.62	99.77	99.77	99.81
7	133/1197	85.92	91.65	97.02	99.65	99.69	100
8	368/3314	88.08	92.39	98.23	99.05	99.21	99.8
9	95/852	99.85	99.09	99.29	99.78	99.80	100
OA (%)		82.67	96.89	99.07	99.62	99.72	99.86
AA (%)		80.84	95.79	98.75	99.49	99.60	99.83
Kappa × 100		81.21	96.56	98.87	99.50	99.64	99.81