A Review of Capsule Network Limitations, Modifications, and Applications in Object Recognition

Abstract

Modern computer vision and machine learning technologies have enabled numerous advances in a variety of domains, including pattern recognition and image classification. One of the most powerful machine learning methods is the capsule network, which encodes features based on their hierarchical relationships. A capsule network is a sort of neural network that uses inverted graphics to represent an item in distinct sections and see the existing link between these pieces, as opposed to CNNs, which lose most of the evidence relating to spatial placement and require a large amount of training data. As a result, the authors give a comparison of various capsule network designs utilized in diverse applications. The fundamental contribution of this study is that it summarizes and discusses the major current published capsule network topologies, including their advantages, limits, modifications, and applications.