May Thaw Tar
Art Classifier by May Thaw Tar
Parami University
A study in machine vision

ART CLASSIFIER

Five categories. 8,685 training images. One convolutional neural network. Upload an artwork and watch the model think.

Start classifying
Scroll
A challenge

Can you beat the model?

Three art images. You guess the category. We'll show you what the model picked.

Question 1 of 3
art image to classify

Which of the five categories is this?

Image Classification

Classify an artwork into one of five categories.

A convolutional neural network trained on 8,685 art images. Upload an image and the model returns a category along with the full probability distribution.

01

Choose an image

PNG or JPG of a drawing, engraving, icon, painting, or sculpture.

02

Run the model

The image is resized to 128ร—128 and passed through a trained CNN.

03

Read the result

A predicted class, a confidence score, and probabilities for all five categories.

Note. The model only recognises five categories: drawings, engraving, iconography, painting, sculpture. Inputs outside this set (a photograph of a person, a logo, a symbol) will still receive a prediction, but the result page will flag it as low confidence.

Upload

PNG, JPG, JPEG

The five categories

Output classes
Drawings Engraving Iconography Painting Sculpture

Trained on the Art Images dataset (Kaggle), 8,685 images split into 7,819 training and 866 validation samples.