Table 5.

Accuracy performance with augmented dataset (30,000 images).

Models	Training Input Size	Training (AP50)	Testing Input Size	Testing (AP50)	Training Time	Iteration Number
YOLOv4 [17]	416 × 416	72.8%	608 × 608	70.3%	102 h	900
YOLOv4-tiny [17]		51.5%		48.4%	12 h
YOLOv3 [48]		68.2%		63.1%	110 h
YOLOv3-tiny [48]		43.9%		40.6%	15 h
Improved YOLOv4		73.6%		71.5%	97 h