A Public Ground-Truth Dataset for Handwritten Circuit Diagram Images

CGHD

This dataset contains images of hand-drawn electrical circuit diagrams as well as accompanying annotation and segmentation ground-truth files. It is intended to train (e.g. ANN) models for extracting electrical graphs from raster graphics.

Content

• 3.269 Annotated Raw Images
  
  • 31 Main Drafters
    • 12 Circuits per Drafter
    • 2 Drawings per Circuit
    • 4 Photos per Drawing
  • Additional Circuit Images provided by TU Dresden (from Real-World Examinations, Drafter 0)
  • Additional Circuit Images provided by RPTU Kaiserslautern-Landau (Drafter -1)
  • 248.020 Bounding Box Annotations
  • 40.711 Rotation Annotations
  • 1.437 Mirror Annotations
  • 85.417 Text String Annotations (equals 93.74% completeness)
    
    • 289.850 Text Characters
    • 98 Character Types (Upper/Lower Case Latin, Numbers, Special Characters)
• 320 Binary Segmentation Maps
  
  • Strokes vs. Background
  • Accompanying Polygon Annotation Files
  • 22.929 Polygon Annotations
• 59 Object Classes
• Scripts for Data Loading, Statistics, Consistency Check and Training Preparation