Published February 3, 2026 | Version v1.1.0
Software Open

TINTOlib: A Python library for transforming tabular data into synthetic images for deep neural networks

Authors/Creators

  • 1. Universidad Politécnica de Madrid
  • 2. Universidad Nacional de Educación a Distancia (UNED), Instituto de Investigación Científica - Universidad de Lima

Description

TINTOlib v1.1.0 Release Notes

WHAT'S NEW

• Two new synthetic image methods: Fotomics and DeepInsight • Customizable transformer system for flexible data preprocessing • New three-level class hierarchy (AbstractImageMethod → MappingMethod → ParamImageMethod) • Feature-to-pixel mapping with explicit CSV export • Support for multiple pixel assignment strategies and relevance scoring

BREAKING CHANGES

Problem parameter updated: OLD: problem="supervised" NEW: problem="classification"

(Deprecated value still works with FutureWarning for backward compatibility)

BUG FIXES

• Fix LogScaler Class (#18) • Fix abstractImageMethod transformer reference in fit_transform() • Fix SuperTML uint8 rendering (#16) • Fix Random State for reproducibility • Fix CSV file generation

DEPENDENCY CHANGES

• numpy: 2.0.2 → 1.26.4 • mpi4py: NOW OPTIONAL (only needed for REFINED method) • Added: numba==0.62.0

VERSION INFO

• Version: 1.0.6 → 1.1.0 • Status: Stable Release

RESOURCES

GitHub: https://github.com/oeg-upm/TINTOlib Documentation: https://tintolib.readthedocs.io/ PyPI: https://pypi.org/project/TINTOlib/ Issues: https://github.com/oeg-upm/TINTOlib/issues

Notes

If you use TINTOlib or this article, please cite it as below.

Files

oeg-upm/TINTOlib-v1.1.0.zip

Files (50.1 MB)

Name Size Download all
md5:03719ac54c1ca8bfc7e1a5a1e83220fa
50.1 MB Preview Download

Additional details

Related works

Is supplement to
Software: https://github.com/oeg-upm/TINTOlib/tree/v1.1.0 (URL)

Software

Repository URL
https://github.com/oeg-upm/TINTOlib