On Computational Performances of the Actual Image Classification Methods in C# and Python

Convolutional Neural Networks (CNNs) built in Python, have become the methodology of choice in image classification. On the other hand, recently formulated Capsule Networks (CapsNets) show potential of deeper understanding of the objects' relations within the images, thereby promising better classification accuracy. The primary aim of this research is to analyze the computational aspects when the two algorithms are implemented in two different programming platforms such as Python and C#. As compared to the equivalent implementations in Python, the results show that C#-based CNN implementation can provide better computational efficiency. Also, C# implementation of CapsNets results in smaller number of epochs for the network to get stabilized, as compared to the Python-based implementation. Finally, at the price of higher computational costs, the results confirm the thesis that, in terms of the handwritten digit classification accuracy, the CapsNets will outperform CNNs.