---
license: mit
task_categories:
  - text-generation
language:
  - de
tags:
  - handwriting
  - stroke-data
  - stylus
  - s-pen
size_categories:
  - n<1K
---

# handwriting-test

This dataset contains handwriting stroke data collected using a stylus (S Pen) on a tablet device.

## Dataset Description

- **Schema Version:** 1.0.0
- **Total Samples:** 67
- **Language:** German

## Data Format

Each sample in `data/train.jsonl` contains:

| Field | Description |
|-------|-------------|
| `id` | Unique identifier (UUID) |
| `text` | The prompt text that was written |
| `created_at` | ISO timestamp of when the sample was created |
| `device` | Device information (user agent, platform, pixel ratio) |
| `canvas` | Canvas dimensions (width, height) |
| `strokes` | Array of strokes, each containing points |
| `svg_path` | Path to SVG visualization |
| `image_path` | Path to PNG visualization |

### Stroke Point Format

Each point in a stroke contains:

| Field | Description |
|-------|-------------|
| `x`, `y` | Raw coordinates (pixels) |
| `xn`, `yn` | Normalized coordinates (0-1) |
| `t` | Timestamp (ms since sample start) |
| `p` | Pressure (0-1) |
| `tiltX`, `tiltY` | Pen tilt angles (if available) |

## Visualization

The dataset includes SVG and PNG renders of each handwriting sample in the `renders/` directory.

## Usage

```python
from datasets import load_dataset

dataset = load_dataset("finnbusse/handwriting-test")
```

## Collection Method

Data was collected using a web application with Pointer Events API, capturing stylus input including pressure and tilt when available.