IgNet. A Super-precise Convolutional Neural Network

TL;DR

This paper introduces IgNet, a novel CNN architecture capable of highly precise analysis of unique and irregular images, such as children's drawings, achieving perfect classification and minimal age estimation error.

Contribution

IgNet is a new family of networks designed to accurately analyze irregular images, outperforming traditional CNNs in recognizing unique features and details.

Findings

IgNet achieved 100% accuracy in categorical classification of drawings.

In regression, IgNet estimated ages with an error of no more than 0.4%.

The design principles of IgNet enable high performance with simple topology.

Abstract

Convolutional neural networks (CNN) are known to be an effective means to detect and analyze images. Their power is essentially based on the ability to extract out images common features. There exist, however, images involving unique, irregular features or details. Such is a collection of unusual children drawings reflecting the kids imagination and individuality. These drawings were analyzed by means of a CNN constructed by means of Keras-TensorFlow. The same problem - on a significantly higher level - was solved with newly developed family of networks called IgNet that is described in this paper. It proved able to learn by 100 % all the categorical characteristics of the drawings. In the case of a regression task (learning the young artists ages) IgNet performed with an error of no more than 0.4 %. The principles are discussed of IgNet design that made it possible to reach such substantial results with rather simple network topology.