Multiscale Efficient Channel Attention for Fusion Lane Line Segmentation

Table 3

Comparision of different methods in KITTI and KITTI-aug. F-MS-ECA denotes fusion with our method(multiscale efficient channel attention).

Dataset	Mode	REC	PRE	F1	F2	Acc
	LiDAR	85.43	25.82	37.84	54.41	95.19
KITTI	Camera	82.30	45.53	55.59	67.06	97.84
	Fusion	93.34	45.89	57.89	71.22	98.16
	F-MS-ECA	93.66	48.16	62.81	77.68	98.50
	LiDAR	85.19	29.40	42.57	60.83	96.22
KITTI-aug	Camera	82.44	47.65	57.31	71.19	97.95
	Fusion	92.24	48.57	62.89	77.12	98.44
	F-MS-ECA	92.23	51.38	65.25	78.57	98.65