Empirical Analysis of Feature Reduction in Deep Learning and Conventional Methods for Foot Image Classification

Deep learning algorithms are employed in many applications, especially medical fields such as gait analysis and human pose detection for rehabilitation. However, creating the desired model with deep requires high memory computing costs, which is problematic because technologies must be run on low-power devices edge equipment. To deal these problems, feature reduction methods reduce energy costs. This paper presents an empirical of reduction. The method classifies foot images knee rehabilitation using convolutional dense autoencoders. obtained results compared those conventional (histograms oriented gradients local binary pattern algorithms). features were classified support vector machine, k-nearest neighbor, multilayer perceptron methods. experimental demonstrate that uses fewer than higher accuracy its algorithm projects pixels onto histogram. In addition, layers maintains accuracy, beneficial implementations.