Accurate machine learning-based germination detection, prediction and quality assessment of three grain crops

Table 2 mAP for the validation and test sets for different model architectures

	ResNet50		ResNet101		Inception v2		Inception-ResNet v2
	val	test	val	test	val	test	val	test
Pennisetum glaucum	85.80	93.66	87.62	93.66	86.16	93.06	88.91	94.25
Secale cereale	89.99	91.83	91.58	92.70	90.07	91.48	92.67	94.21
Zea mays	96.21	96.29	96.54	96.69	95.81	95.62	97.48	97.90

Results (in %) on the validation set are indicated by val and results on the test set are indicated by test. Italic values indicate the model with the highest mAP for each seed type

ISSN: 1746-4811