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

Table 4

Performances of different network depths for the 3D-CNN and 3D-FCNN models.

Model	Dataset	3	5	7	9	11
3D-CNN	IP	87.78	99.07	77.69	75.76	73.04
	PC	95.62	98.93	97.03	96.22	95.00
	UP	93.79	94.01	95.72	95.11	94.25
	BS	88.04	90.69	88.96	87.13	85.64
	SV	93.08	94.40	94.05	93.33	92.57
3D-FCNN	IP	89.60	99.25	98.51	96.35	95.63
	PC	99.33	99.63	99.68	98.72	97.77
	UP	94.80	98.25	98.49	98.55	98.41
	BS	88.44	96.13	97.02	95.45	94.28
	SV	93.76	96.38	96.97	95.87	95.44