[Retracted] Advances in Hyperspectral Image Classification with a Bottleneck Attention Mechanism Based on 3D-FCNN Model and Imaging Spectrometer Sensor

Table 6

Classification effects of different modules on the SV dataset.

Class	3D-FCNN	SE+3D-FCNN	BandAM+3D-FCNN	BAM+3D-FCNN
1	100	98.99	100	100
2	100	100	100	100
3	100	99.90	100	100
4	100	100	99.76	98.49
5	94.19	95.44	99.75	99.96
6	100	98.55	100	100
7	100	100	100	99.76
8	99.93	97.65	100	99.08
9	100	100	100	100
10	99.97	99.32	100	100
11	100	99.62	100	100
12	100	96.59	100	99.78
13	100	98.90	100	100
14	99.90	99.72	99.79	100
15	79.80	93.39	91.48	99.96
16	99.94	99.94	100	100
OA (%)	96.88	98.05	98.83	99.73
AA (%)	98.27	98.59	99.39	99.81
Kappa	96.52	97.83	98.70	99.70