Art Pictograms Dataset

Empowering Neural Networks to Explore Artistic Creation

Overview

The Art dataset is designed for advanced experiments in deep learning and transfer learning. It aims to facilitate the generation of new images by neural networks, mimicking human artistic styles.

Dataset Status

The dataset is currently under development, and a detailed description will be provided upon completion. The available images are meticulously labeled to support varied learning scenarios.

Structure and Labels

Each image file is prefixed with labels that categorize the image by style and subject:

• Style Labels:
  • 01 – Pictogram
  • 02 – Contour
  • 03 – Sketch
• Subject Labels:
  • 001 – Flower
  • 002 – Bird
  • 003 – Butterfly
  • 004 – Tree
  • 005 – Plane
  • 006 – Crane
  • 007 – Dog
  • 008 – Horse
  • 009 – Deer
  • 010 – Truck
  • 011 – Car
  • 012 – Cat
  • 013 – Frog
  • 014 – Ship
  • 015 – Fish
  • 016 – House

Vision

Can neural networks learn to replicate the creative process of human artists? This dataset not only challenges AI with the task of artistic replication but also explores the potential of algorithms in achieving proficiency similar to that of professional artists.