SpikeSegNet-a deep learning approach utilizing encoder-decoder network with hourglass for spike segmentation and counting in wheat plant from visual imaging

Table 5 Performance analysis of SpikeSegNet approach on illuminated dataset

	Gamma 0.1	Gamma 0.3	Gamma 0.5	Gamma 1 (original image)	Gamma 1.5	Gamma 2.0	Gamma 2.5
Classification error rate (E₁)	0.005349223	0.003051249	0.002396139	0.001693726	0.00177917	0.0020579	0.002359009
Classification error rate (E₂)	0.083304137	0.042303843	0.040107991	0.04874738	0.06317883	0.08736489	0.108331881
Average_Precision	0.998094286	0.999318633	0.999408703	0.999325313	0.99911935	0.99879521	0.998452183
Average_Recall	0.996521902	0.997607927	0.998178134	0.99896944	0.99908922	0.99913291	0.998812372
Average_F_1_measure	0.997302422	0.998461999	0.998792781	0.999147219	0.99910416	0.99896389	0.998812369
Average_Accuracy	0.997302421	0.998462003	0.998792786	0.999147223	0.99910416	0.9989639	0.998812372
Average_Jaccard_Index_for_Spike_detection:	0.994627456	0.99693175	0.997591056	0.998298182	0.99821199	0.99793177	0.997629168

ISSN: 1746-4811