DA-ActNN-YOLOV5: Hybrid YOLO v5 Model with Data Augmentation and Activation of Compression Mechanism for Potato Disease Identification

Table 3

Test performance of the DA-ActNN-YOLOV5m model on the PBD-IM dataset using data augmentation methods.

Method	Mertics (%)	Earlyblight	Healthy	Lateblight	AP (%)
With data augmentation		99.38	99.58	99.88	99.61
		99.01	100.00	100.00	—
		98.04	100.00	99.03	—
		98.05	98.06	99.09	—
Without data augmentation		93.57	85.17	92.45	90.39
		88.01	93.05	90.00	—
		91.03	86.04	87.06	—
		89.06	89.00	88.04	—