ContourCNN: convolutional neural network for contour data classification

TL;DR

This paper introduces ContourCNN, a neural network designed for classifying contour data by leveraging circular convolution and priority pooling to handle cyclical and sparse shape representations, achieving high accuracy.

Contribution

The paper presents a novel CNN architecture with circular convolution and priority pooling tailored for contour shape classification, addressing cyclical data and sparsity issues.

Findings

Achieved high classification accuracy on EMNIST shape data

Effectively handled cyclical properties of contour data

Demonstrated robustness to sparse contour representations

Abstract

This paper proposes a novel Convolutional Neural Network model for contour data analysis (ContourCNN) and shape classification. A contour is a circular sequence of points representing a closed shape. For handling the cyclical property of the contour representation, we employ circular convolution layers. Contours are often represented sparsely. To address information sparsity, we introduce priority pooling layers that select features based on their magnitudes. Priority pooling layers pool features with low magnitudes while leaving the rest unchanged. We evaluated the proposed model using letters and digits shapes extracted from the EMNIST dataset and obtained a high classification accuracy.