Table 3.

The average performance of models with different evaluation metrics on images with soil background

Model	Accuracy	Error	Sensitivity	Specificity	Precision	False positive rate
Multi-class CNN	0.63 ± 0.20	0.37 ± 0.20	0.63 ± 0.2	0.94 ± 0.05	0.88 ± 0.1	0.06 ± 0.05
2-class CNN’s	0.72 ± 0.25	0.28 ± 0.26	0.72 ± 0.24	0.95 ± 0.06	0.90 ± 0.11	0.08 ± 0.05
CNN-LSTM	0.83 ± 0.10	0.15 ± 0.10	0.82 ± 0.10	0.93 ± 0.06	0.85 ± 0.10	0.06 ± 0.06
ConvLSTM	0.62 ± 0.2	0.33 ± 0.2	0.68 ± 0.2	0.93 ± 0.07	0.84 ± 0.1	0.06 ± 0.06