Classification Modeling Method for Near-Infrared Spectroscopy of Tobacco Based on Multimodal Convolution Neural Networks

Table 6

Classification performance of different models.

Model	Main parameters of model	Test set classification accuracy (%)
1D_CNN_11	C: 1 ∗ 15; S: 1 ∗ 2;M = 24	93.15
2D_LeNet-5	Input size: 32 ∗ 32	93.05
PLS_DA	Characteristic spectrum: 4000 cm^−1∼7740 cm^−1; PC number: 7	81.25