No abstract

90% of the time needed for the conversion from point clouds to 3D models of industrial facilities is spent on geometric modelling due to the sheer number of Industrial Objects (IOs) of each plant. Hence, cost reduction is only possible by automating modelling. Our previous work has successfully identified the most frequent industrial objects which are in descending order: electrical conduit, straight pipes, circular hollow sections, elbows, channels, solid bars, I-beams, angles, flanges and valves. We modelled those on a state-of-theart software, EdgeWise and then evaluated the performance of this software for pipeline and structural modelling. The modelling of pipelines is summarized in three basic steps: (a) automated extraction of cylinders, (b) their semantic classification and (c) manual extraction and editing of pipes. The results showed that cylinders are modelled with 75 % recall and 62 % precision on average. We discovered that pipes, electrical conduit and circular hollow sections require 80 % of the Total Modelling Hours (TMH) of the 10 most frequent IOs to build the plant model. TMH was then compared to modelling hours in Revit and showed that 67 % of pipe modelling time is saved by EdgeWise. This paper is the first to evaluate state-of-the-art industrial modelling software. These findings help in better understanding the problem and serve as the foundation for researchers who are interested in solving it.

Generation of digital models of existing industrial facilities is labor intensive and expensive. The use of state-of-the-art deep learning algorithms can assist to reduce the modelling time and cost. However large databases of labelled, laser-scanned industrial facilities do not exist to date, henceforth training of deep learning models is not possible. Our paper solves this problem by proposing a new benchmark dataset, which consists of five labelled industrial plants. The labelling schema that we followed for the generation of this dataset is based on the frequency of appearance of industrial object types. We labelled the ten most frequent industrial object shapes as identified in previous work. We present CLOI (Channels, L-shapes, circular sections, I-shapes): a richly annotated large-scale repository of shapes represented by labelled point clusters. CLOI has more than 140 million hand labelled points and serves as the foundation for researchers who are interested in automated modelling of industrial assets using deep learning algorithms.

No abstract

No abstract