IAPR-TC11: Association for Pattern Recognition Technical Committee Number 11

There is a newer version of the record available.

Published October 31, 2023 | Version 12
Dataset Open

A Public Ground-Truth Dataset for Handwritten Circuit Diagram Images

Authors/Creators

  • 1. DFKI

Description

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

  • 2.424 Raw Images (Annotated)
    • 24 Drafters (plus Images provided by TU Dresden)
    • 12 Circuits per Drafter
    • 2 Drawings per Circuit
    • 4 Photos per Drawing
    • 201.142 Bounding Box Annotations
    • 26.898 Rotation Annotations
    • 61.595 Text String Annotations
  • 257 Binary Segmentation Maps (Annotated)
    • Strokes vs. Background
    • Accompanying Polygon Annotation Files
    • 19.562 Polygon Annotations
  • 59 Object Classes
  • Scripts for Data Loading, Statistics, Consistency Check and Training Preparation

Notes

Added and corrected text and rotation annotations across the entire dataset. Added new circuit images in examination context from TU Dresden including annotation preview (Drafter 0).

Files

cghd-zenodo-12.zip

Files (3.4 GB)

Name Size Download all
md5:1d7783647edcd7c73f38d79eb785f901
3.4 GB Preview Download

Additional details

Related works

Is described by
Preprint: https://arxiv.org/abs/2107.10373 (URL)
Conference paper: 10.1007/978-3-030-86198-8_2 (DOI)
Preprint: https://arxiv.org/abs/2301.03155 (URL)
Poster: 10.5281/zenodo.7652836 (DOI)