Classification of multi-year and multi-variety pumpkin seeds using hyperspectral imaging technology and three-dimensional convolutional neural network

Table 4 Results of pumpkin seed classification models based on spectral images

Model		Parameters (MB)	Training set accuracy (%)	Test set
Model		Parameters (MB)	Training set accuracy (%)	Accuracy (%)	Precision (%)	Recall (%)	F1
Transfer pretrained models	Vgg16	254	86.84	83.42	83.58	83.42	83.34
	Vgg19	240	81.24	80.66	80.67	80.67	80.46
	ResNet50	849	76.67	75.89	75.86	75.89	75.58
	ResNet101	849	75.46	70.83	71.02	70.83	70.57
	InceptionV3	409	65.18	61.38	61.08	61.38	60.80
	Xception	1608	71.85	71.10	71.08	71.10	70.65
Ours	3DCNN	49	95.69	94.16	94.19	94.16	94.13
Ours	PA-3DCNN	49	99.14	95.20	95.24	95.20	95.19

ISSN: 1746-4811