Table 2

SensatUrban dataset experiment

MethodOA(%)mIoU (%)mAcc (%)(IoU %)class
GroundVeg-BuildingWallBridgeParkingRailTrafficStreetCarFootpathBikeWater
PointNet80.7822.7523.7167.9689.5280.05003.95031.55035.14000
PointNet++84.335.0632.9272.4694.2484.772.822.0925.79031.5411.4238.847.12056.93
SegCloud85.2737.2937.2969.9394.5588.8732.8312.5815.7715.4830.6322.9656.420.54044.24
SPG88.6640.9342.6674.1097.994.263.37.524.2030.103474.40054.8
RandLA-Net90.256.4357.5887.198.9195.3374.428.6941.38055.9954.4385.6750.39071.30
DSF-Net(ours)89.7856.958.6989.1198.0796.5888.4050.4561.62066.6753.2386.1439.63071.31

Note(s): SPG: super point graph

Source(s): Authors’ own creation

or Create an Account

Close Modal
Close Modal