Table 5.

Performance comparison of different YOLO models on online exam cheat detection tasks.

Models	Precision	Recall	mAP50	mAP50-95	Parameters	GFLOPs
YOLOv5n	0.93855	0.95840	0.98088	0.75466	2,503,334	7.1
YOLOv8n	0.91304	0.97604	0.98513	0.79401	3,006,038	8.1
YOLOv9t	0.92883	0.94797	0.98123	0.73676	1,971,174	7.6
YOLOv10n	0.92774	0.94099	0.97988	0.77188	2,695,196	8.2
YOLOv11n	0.92852	0.96561	0.98199	0.76616	2,582,542	6.3
YOLOv12n	0.93823	0.94978	0.98156	0.75631	2,557,118	6.3
YOLOv12NoAttn	0.93019	0.95647	0.98208	0.75436	1,840,350	5.5