Mean Hand is a unique typeface created by Anna Zhang using 814,255 handwritten character samples from the 1990s NIST database, which later became the EMNIST dataset for training handwriting recognition systems. The typeface demonstrates how legibility emerges only within a specific threshold of character variation, while also highlighting how such datasets shape what is considered 'legible' handwriting in technology. The project makes visible the biases inherent in handwriting recognition systems and is available as a free download.
Background
The EMNIST dataset, derived from NIST Special Database 19, is a standard benchmark for training and evaluating handwriting recognition systems, influencing what is considered legible in digital handwriting processing.
- Source
- Lobsters
- Published
- Jun 3, 2026 at 09:20 PM
- Score
- 6.0 / 10