An Effective Method for Sensing Power Safety Distance Based on Monocular Vision Depth Estimation

Table 3

Comparison between six kinds of computer vision methods on accuracy and speed.

Method	Lower is better					Higher is better
	AbsRel	RMSE	RMSE-log	SqRel	Speed (per picture)	Correct (%)
Binocular camera	0.3398	7.9764	0.4283	1.8364	2.0 s	64.46
SFMlearner	0.3081	7.6983	0.4887	2.8249	1.0 s	68.42
MonoDepth	0.1898	5.9764	0.2283	0.8364	89 ms	77.88
DenseDepth	0.1550	5.8630	0.2150	0.9051	1.0 s	79.98
Power density depth model	0.1378	5.4645	0.1932	0.8363	69 ms	85.36