Table 4. Comparison of model efficiency (data in bold are ranked first, those underlined are ranked second).

Method	Parameters	FLOPs	FPS
UNet	31 M	16.415	161.325
TransUNet	67.1 M	32.506	89.840
CGNet	491 K	0.854	152.966
SegFormer	7.7 M	3.272	174.886
GCNnet	61.4M	2.425	237.387
ContextNet	911 K	0.212	318.380
DABNet	752 K	1.283	263.743
FPENet	114 K	0.183	184.508
Ours	466 K	0.202	280.417