typed_digital_signatures
VerifiedSynthetic signatures from 30 handwriting-style Google Fonts for image tasks.
What is typed_digital_signatures?
This dataset consists of synthetic digital signatures rendered using 30 Google Fonts with handwriting and signature-style features, yielding 3,000 images per font for a total of roughly 90,000 samples.
It supports training and evaluation of vision models on signature analysis and font recognition through image classification and feature extraction workflows.
What you can build with typed_digital_signatures
Train font-based signature classifiers
Build and evaluate image classification models that identify which of the 30 handwriting-style fonts produced a given signature image.
Develop zero-shot signature models
Create and benchmark zero-shot classifiers that recognize signature styles from fonts never seen during training.
Pre-train feature extractors
Use the 90k images to train backbone networks for downstream tasks such as real signature verification or similarity search.
Load typed_digital_signatures
from datasets import load_dataset
ds = load_dataset("Benjy/typed_digital_signatures")- 1pip install datasets
- 2from datasets import load_dataset
- 3ds = load_dataset('benjy/typed_digital_signatures')
- 4Access splits with ds['train'] and ds['test']
- 5Load images via the 'image' column and labels via the 'font' column
typed_digital_signatures: pros & cons
Pros
- +90k images across 30 fonts
- +Consistent 3k samples per class
- +Ready for HF datasets loader
- +Supports classification and feature tasks
Cons
- –Entirely synthetic, no real handwriting
- –Only 30 font styles available
- –May not generalize to scanned paper signatures
Frequently asked questions
A collection of ~90,000 synthetic signature images generated from 30 Google Fonts chosen for handwriting traits, with 3,000 images per font.
User reviews
Verified reviews from the community shape this listing's rating.
Loading reviews…
Similar datasets
Other images & vision options worth comparing.