Table 3.

Results using YOLO models trained with different datasets (D₁ is the original dataset, D₂ is D₁ augmented using flips; D₃ is D₁ augmented using blurring, gamma correction and equalisation; and D₄ is the combination of D₂ and D₃) for the stomata dataset

	Precision	Recall	F1-score	TP	FP	FN	IoU
D 1	0.97	0.61	0.75	591	21	374	0.75
D 2	0.97	0.88	0.92	852	26	113	0.81
D 3	0.95	1.00	0.97	961	52	4	0.79
D 4	0.99	0.90	0.94	869	12	96	0.83