Skip to content

This is a simple convolutional neural network for recognizing the artist of several old style paintings by just by feeding the network only with the images itself.

Notifications You must be signed in to change notification settings

HanSur94/ARTNet

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

8 Commits
 
 
 
 
 
 
 
 
 
 

Repository files navigation

ARTNet

This is a simple convolutional neural network for recognizing the artist of several old style paintings by just by feeding the network only with the images itself.

Table of contents

Introduction

The purpose of this project was to learn something about artificial neura networks. Therefore, I decided to program one for my self using the Matlab inbuild neural network tools. Furthermore, I decided to do some image classification and found a decend dataset containing paintings from thousands of artists including images of paintings, the artist names and painting titles. This sparked a idea inside me to create a CNN which classifies all the art pieces to their corresponding artists.

Neural Net Architecture

For this Image classification task I used a convolutional neural network (CNN). I will not explain the working principle of a CNN, since there are alreday man good explanations on CNN's out there. Important is probably the the used architechture, which can be found seen in the list below.

Layers:

  • imageInputLayer(inputSize)
  • convolution2dLayer(10, 30)
  • batchNormalizationLayer
  • reluLayer
  • convolution2dLayer(3, 30)
  • batchNormalizationLayer
  • reluLayer
  • convolution2dLayer(3, 30)
  • batchNormalizationLayer
  • reluLayer
  • convolution2dLayer(2, 30)
  • batchNormalizationLayer
  • reluLayer
  • fullyConnectedLayer(height(labelCount))
  • reluLayer
  • fullyConnectedLayer(height(labelCount))
  • softmaxLayer
  • classificationLayer

inputSize = [100, 100, 3]; labelCount --> Number of different Labels

As you can see, the architecture is quite straight forward. As i tried out several Architectures, I came to the conlcusion that most importantly the CNN training time shouldnt take years. Therefore the CNN architecture was kept simple so that multiple epochs of training could be performed over night.

Training of the Network

.....

Examples

The CNN in action can be seen here, where random paintings on canvas from the artwork dataset are used and the corresponding artists are identified. The CNN is only fed with the image data itself.

alt text

Built With

Authors

About

This is a simple convolutional neural network for recognizing the artist of several old style paintings by just by feeding the network only with the images itself.

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages