This AI service extracts the bridge gradients from pixel-based drawings. The gradient describes the elevation profile of the bridge axis and offers essential information for an accurate reconstruction of the geometry of the superstructure. The course of the gradient is explicitly described in the longitudinal view of the bridge by the gradient symbols and associated parameters. This includes essential information about the elevation and slope of specific points along the bridge axis.

This service combines Deep Learning-based methods with optical text recognition (OCR) to extract the required information. To achieve the best possible results, good-quality drawings are of particular importance. This means they should be high in contrast and free of age artifacts such as crease lines, discolorations, and equalizations.

Data

  •     Quality: The scanned plans should be of good quality, i.e., high in contrast and free of age artifacts such as crease lines, discoloration, and fading. Processing poor-quality plans is possible but tends to produce poorer results.
  •     Input data format (e.g. dxf/pdf/doc/jpg/xls):
    •         PNG
  •     Training data:
    •         Construction plans
  •     Output data format (e.g.):
    •         PNG
    •         Knowledge graph (probably textual)

Contact Persons:
Benedikt Faltin, M. Sc., Lehrstuhl für Informatik im Bauwesen, Ruhr-Universität Bochum
Phillip Schönfelder, M. Sc., Lehrstuhl für Informatik im Bauwesen, Ruhr-Universität Bochum

Felix Kretschmann, elevait GmbH & Co. KG
Bianca Preißler, elevait GmbH & Co. KG