Conference Agenda

This work presents a methodological approach for developing a 3D Virtual Reality model of a landslide in the urban area of ​​the municipality of Petrópolis, State of Rio de Janeiro, Brazil. The 3D virtual reality model was generated from data obtained by Unmanned Aerial Vehicle, 3D modeling, and data integration in an immersive and interactive geovisualization environment. Some images from before the event were used to reconstruct the buildings on the hillside, and 3D modeling was carried out using parametric rules. The generated model can be used in a virtual reality room (team experience) or with virtual reality glasses (individual experience). Although the Virtual Reality models do not replace fieldwork, they aid in understanding the landslide triggering process and analyzing the impact caused, and have the potential to be used in educational projects for risk and disaster reduction.

The existing LoD frameworks for 3D city models are based around the typical data sources used to construct these models in GIS: 2D topographic data, semantic data and/or 3D measurements. In recent years, Building Information Modelling (BIM) has become a viable alternative source. However, a BIM model is a very different source from those that are typically used in GIS, one with access to smaller features and installations that are typically hidden from view. At the same time, geometric errors are prevalent in BIM models and often prevent LoD3 models from being made successfully-limiting BIM-derived models to LoD2. In this extended abstract, we make the case for four new LoDs that we believe are especially valuable for BIM-derived models. In the full paper, we will provide details on the proposed LoDs, show how these LoDs can be derived from BIM models using an automated BIM-to-GIS conversion method, and evaluate the BIM-derived 3D building models at different LoDs on a set of performance criteria.

This study addresses the challenges of accessibility and laborious intensity in visual inspections of public metropolitan mobility infrastructure, such as elevated Metro systems. It explores an experimental 3D-Mapping Inspection and Classification Evaluation method (3D-MICE) utilizing UAV imagery and geometric mensuration from 3D point clouds. The method introduces two classification techniques: Condition Classification by Intensity (CCI) and Geometry Classification by RGB color (GCC), applied to orthomosaics. 3D-MICE enables semi-automatic detection, segmentation, and measurement of cracks and stains in reinforced concrete by selecting areas of interest based on intensity and geometric features. This approach offers a promising, efficient, and precise alternative to traditional inspection methods. 3D-MICE can detect, segment and measure, semi-automatically, cracks and stains of reinforced concrete structures by selecting areas of interest based on intensity and geometry.