Towards Proactive Surveillance through CCTV Cameras under Edge-Computing and Deep Learning

Table 2

Results comparison of all trained models.

Evaluation matrices	Faster RCNN	R-FCN	YoloV4	Hybrid method
True positive	551	581	576	586
False negative	99	77	105	98
False positive	89	33	65	35
True negative	590	638	593	610
Accuracy	0.85	0.91	0.87	0.90
Precision	0.86	0.93	0.90	0.94
Recall	0.84	0.88	0.84	0.86
F1 score	0.85	0.91	0.87	0.89
Avg frames/sec	5–17	7	55	(i) 55 -> (ii) 31