Table 5. The comparative experimental results on the Tusimple dataset.

Bold numbers are the best results.

Method	Backbone	F1 (%)	Accuracy (%)	Proportion of FP (%)	Proportion of FN (%)
SCNN (Pan et al., 2018)	VGG16	95.57	96.53	6.17	1.8
RESA (Zheng et al., 2021)	Resnet18	96.93	96.84	3.63	2.48
LaneATT (Tabelini et al., 2021a)	Resnet18	96.71	95.57	3.56	3.01
LaneATT (Tabelini et al., 2021a)	Resnet34	96.77	95.63	3.53	2.92
LaneATT (Tabelini et al., 2021a)	Resnet122	96.06	96.10	5.64	2.17
PolyLaneNet (Tabelini et al., 2021b)	EfficientNetB0	90.02	93.36	9.42	9.33
LSTR (Liu et al., 2021b)	Resnet18	−	96.18	2.91	3.38
CondLane (Liu et al., 2021a)	Resnet18	97.01	95.48	2.18	3.80
CondLane (Liu et al., 2021a)	Resnet34	96.98	95.37	2.20	3.82
CondLane (Liu et al., 2021a)	Resnet101	97.24	96.54	2.01	3.50
CLRNet (Zheng et al., 2022)	Resnet18	97.89	96.84	2.28	1.92
CLRNet (Zheng et al., 2022)	Resnet34	97.82	96.87	2.27	2.08
CLRNet (Zheng et al., 2022)	Resnet101	97.62	96.83	2.37	2.38
Our method	Resnet18	98.01	96.91	2.31	2.12
Our method	Resnet34	97.89	96.93	2.52	2.41
Our method	Resnet101	97.68	96.89	2.97	3.09