Collection of German Handwriting Samples

To train AI models for our Zibian Labs Handwriting Examination software, we rely on real-world handwriting samples from native speakers. Depending on the language, we require contributions from several hundred individuals, with each participant providing approximately 10 short sentences. Each sample contains a different text to ensure linguistic variety and robustness of the model.

We collect only minimal background information from participants, limited to age, gender, and preferred writing hand. No further personal data is requested. All physical paper samples are securely disposed of after digitization and model training. The data is used exclusively in the context of human rights and criminal investigations, and for no other purpose.

We are grateful to the Zurich Institute of Forensic Science for their collaboration and for providing a substantial number of handwriting samples collected in their forensic context. Many thanks to the team around Jonathan Heckeroth and colleagues for their valuable support.