Proposing Lane and Obstacle Detection Algorithm Using YOLO to Control Self-Driving Cars on Advanced Networks

Table 5

Result of comparison lane detection algorithm.

Method	Algorithm	Average accuracy (%)	Average processing time (second)	Environment system
Traditional method	Hough transform [33]	95.70	0.06540	Intel Core i7-6700K CPU@ 4 GHz
Traditional method	RANSAC + HSV [34]	86.21	0.50000	Intel Core i7-4700 CPU@ 2.40 GHz
Horizontal filter + Otsu	[35]	83.00	0.013	CPU Intel 3.30 GHz
Traditional method	Sliding window KITTI [4]	84.00	0.00318	Intel Core i5 5200U CPU@ 2.20 GHz
Deep learning	FastDraw ResNet [36]	95.00	0.06533	NVIDIA GeForce GTX 1080, GPU
Our proposal	HSL + Sobel filter + SWS KITTI [4]	85.13	0.08620	Intel Core i3-6100U, CPU@ 2.3 GHz
Our proposal	HSL + Sobel filter + SWS TuSimple	97.91	0.08500	Intel Core i5-9300H, CPU@ 2.4 GHz
Our proposal	HSL + Sobel filter + SWS TuSimple	97.91	0.0021	GPU GeForce GTX TITAN X