Heuristic Generation of Multispectral Labeled Point Cloud Datasets for Deep Learning Models

Abstract

Abstract. Deep Learning (DL) models need big enough datasets for training, especially those that deal with point clouds. Artificial generation of these datasets can complement the real ones by improving the learning rate of DL architectures. Also, Light Detection and Ranging (LiDAR) scanners can be studied by comparing its performing with artificial point clouds. A methodology for simulate LiDAR-based artificial point clouds is presented in this work in order to get train datasets already labelled for DL models. In addition to the geometry design, a spectral simulation will be also performed so that all points in each cloud will have its 3 dimensional coordinates (x, y, z), a label designing which category it belongs to (vegetation, traffic sign, road pavement, …) and an intensity estimator based on physical properties as reflectance.

Transport Infrastructure Management Based on LiDAR Synthetic Data: A Deep Learning Approach with a ROADSENSE Simulator

“…Furthermore, ROADSENSE is an improved version of the work presented in [36], which consists of three steps: (i) the generation of a digital terrain model (DTM), (ii) the simulation of roadways and their assets according to official road design norms, and, finally, (iii) the positioning of trees along suitable regions of the DTM. Regardless of the type of point cloud, the initial stages involving the creation of the DTM and tree generation remain consistent.…”

“…Innovatively, our simulator enhances realism by incorporating road cross section definitions, addressing a limitation observed in a prior work [36]. The previous model placed all road stretch points at the same height, hindering DL model learning due to difficulties in distinguishing equivalent areas.…”