Research Article

Fine-Grained Point Cloud Semantic Segmentation of Complex Railway Bridge Scenes from UAVs Using Improved DGCNN

Table 3

The percentage of different classification elements present in each of the three scenes.

ScenesCategory (%)Total points
CableClusterGuardrailMastProtective wallRailTrack bed

(a)5.1231.2816.113.598.597.9427.013253843
(b)3.5927.0715.994.4511.7112.6124.584179984
(c)4.2339.0611.778.097.2814.9614.612720482