TINTO — Transform Tabular Data into Synthetic Images for Deep Neural Networks

Name: TINTO
Author: Manuel Castillo-Cara

Important Notice

This repository contains the original TINTO engine for converting tabular data into synthetic images.

📦 We now recommend using TINTOlib, which includes:

Original TINTO + extra methods (IGTD, REFINED, BarGraph, DistanceMatrix, FeatureWrap, SuperTML, BIE)
More flexible interface and better documentation
Free course with examples and video tutorials

🔄 TINTOlib is actively developed with regular improvements.

👉 For new projects, switch to TINTOlib.

TINTO — Tabular Data → Synthetic Images

Open-source Python framework that turns tidy tabular data into images using PCA, t-SNE and blurring — ready for CNN pipelines.

⭐ GitHub Ver características 🎓 Curso gratuito (Udemy)

🎉 Free Course: TINTOlib & Hybrid Neural Networks

Learn how to convert tabular data into images and train CNNs, ViTs and hybrid architectures.

👉 Access on Udemy

🔎 Abstract

TINTO is an open-source, user-extendable framework to convert tidy data into images via 2-D projection (PCA, t-SNE) and a blurring technique that adds ordered information, often improving CNN classification.

TINTO framework diagram

📺 VideoTutorial Course (English/Spanish)

Prefer GitHub over Udemy? Follow the full bilingual course with notebooks:

📄 Documentation

All documentation and source code are available in the OEG GitHub repository.

🎬 Video Example

🔍 Main Features

Works with CSV files in Tidy Data format.
Input: numeric features; target in the last column.
Projection methods: PCA and t-SNE.
Output: black-and-white synthetic images.
Blurring technique for pixel blending.
Python 3.7+, Linux/Windows/macOS.

📥 Input

Example (IRIS dataset format):

sepal length	sepal width	petal length	petal width	target
4.9	3.0	1.4	0.2	1
7.0	3.2	4.7	1.4	2
6.3	3.3	6.0	2.5	3

🖼️ Output

TINTO output example

📖 Citation

If you used TINTOlib with Hybrid Neural Networks, cite the IEEE JSTSP paper:

@ARTICLE{10946146,
  author={Castillo-Cara, Manuel and Martínez-Gómez, Jesus and Ballesteros-Jerez, Javier and García-Varea, Ismael and García-Castro, Raúl and Orozco-Barbosa, Luis},
  journal={IEEE Journal of Selected Topics in Signal Processing},
  title={MIMO-Based Indoor Localisation with Hybrid Neural Networks: Leveraging Synthetic Images from Tidy Data for Enhanced Deep Learning},
  year={2025},
  pages={1-13},
  doi={10.1109/JSTSP.2025.3555067}
}

If you used TINTO, cite the Information Fusion paper:

@article{inffus_TINTO,
  title   = {A novel deep learning approach using blurring image techniques for Bluetooth-based indoor localisation},
  journal = {Information Fusion},
  author  = {Reewos Talla-Chumpitaz and Manuel Castillo-Cara and Luis Orozco-Barbosa and Raúl García-Castro},
  volume  = {91},
  pages   = {173-186},
  year    = {2023},
  issn    = {1566-2535},
  doi     = {10.1016/j.inffus.2022.10.011}
}

And the SoftwareX paper:

@article{softwarex_TINTO,
  title   = {TINTO: Converting Tidy Data into Image for Classification with 2-Dimensional Convolutional Neural Networks},
  journal = {SoftwareX},
  author  = {Manuel Castillo-Cara and Reewos Talla-Chumpitaz and Raúl García-Castro and Luis Orozco-Barbosa},
  year    = {2023},
  volume  = {22},
  pages   = {101391},
  issn    = {2352-7110},
  doi     = {10.1016/j.softx.2023.101391}
}

⭐ Ver repositorio en GitHub

Ph.D. Manuel Castillo-Cara