Lane detection based on dual attention mechanism

Feng-lei REN; Hai-bo ZHOU; Lu YANG; Xin HE

doi:10.37188/CO.2022-0033

Chinese Optics, Volume. 16, Issue 3, 645(2023)

Lane detection based on dual attention mechanism

Feng-lei REN^1,2, Hai-bo ZHOU^1,2、*, Lu YANG^1,2, and Xin HE³

¹Tianjin Key Laboratory for Advanced Mechatronic System Design and Intelligent Control, School of Mechanical Engineering, Tianjin University of Technology, Tianjin 300384, China

²National Demonstration Center for Experimental Mechanical and Electrical Engineering Education, Tianjin University of Technology, Tianjin 300384, China

³Changchun Institute of Optics, Fine Mechanics and Physics, Chinese Academy of Sciences, Changchun 130033, China

show less

Abstract Get PDF(in Chinese)

Figures & Tables(10)

Fig. 1. Schematic diagram of lane detection

Download full size

View in Article

Fig. 2. Schematic diagram of semantic segmentation of image

Download full size

View in Article

Fig. 3. Schematic diagram of proposed lane detection algorithm

Download full size

View in Article

Fig. 4. Diagram of atrous convolution. (r=1, 2, 4 from left to right)

Download full size

View in Article

Fig. 5. Schematic diagram of the position attention module

Download full size

View in Article

Fig. 6. Schematic diagram of the channel attention module

Download full size

View in Article

Fig. 7. Lane detection results of proposed algorithm on Tusimple

Download full size

View in Article

Fig. 8. Lane detection results of our algorithm on CULane

Download full size

View in Article

Table 1. Quantitative experiment results of proposed algorithm on Tusimple
View table
View in Article
Table 1. Quantitative experiment results of proposed algorithm on Tusimple
Method acc(%) FP(%) FN(%) FPS
SCNN^[18] 96.53 6.17 1.80 7.5
LaneNet^[13] 96.38 7.80 2.44 52.6
PolylaneNet^[19] 93.36 9.42 9.33 115
FastDraw^[20] 95.20 7.60 4.50 90.3
R-50-E2E^[21] 96.04 3.11 4.09 −
Ours 96.63 6.02 2.03 134

Table 2. Quantitative experiment results of proposed algorithm on CULane

View table

View in Article

Table 2. Quantitative experiment results of proposed algorithm on CULane

Method	Normal	Crowd	Dazzle	Shadow	Noline
SCNN^[18]	90.60	69.70	58.50	66.90	43.40
FastDraw^[20]	85.90	63.60	57.00	69.90	40.60
UFSD-18^[1]	87.70	66.00	58.40	62.80	40.20
UFSD-34^[1]	90.70	70.20	59.50	69.30	44.40
LaneATT^[22]	91.17	72.71	65.82	68.03	49.13
Ours	91.21	76.33	69.51	73.25	50.16
Method	Arrow	Curve	Cross	Night	Total
SCNN^[18]	84.10	64.40	1990	66.10	71.60
FastDraw^[20]	79.40	65.20	7013	57.80	-
UFSD-18^[1]	81.00	57.90	1743	62.10	68.40
UFSD-34^[1]	85.70	69.50	2037	66.70	72.30
LaneATT^[22]	87.82	63.75	1020	68.58	75.13
Ours	88.72	71.25	1265	70.73	77.32