Comparative Analysis of Thermogram and Pre-Processed HoG Images Using Machine Learning Classifiers

Recently, machine learning models have been widely used to measure the performance of classification tasks. As per the specific applications, different classifiers are compared and summarized. In this paper, a set of thermal images is collected during nighttime hours and utilized for classification purposes. Machine learning models like Gaussian Naive Bayes, Decision Tree Algorithm, Random Forest, Linear Discriminant Analysis, Logistic Regression, Support Vector Machine, and K-Nearest Neighbor are used in this paper. Because of the complexity of the features in the thermal image, image processing is introduced to pre-processing before classification. The transformation of thermal images into HOG images is the conversion for reducing the intricacy of the thermal images. This paper exposes the optimal relative study between thermal and HOG images with the above types of machine learning classifiers. The most populated and spotted animal, deer, is the subject of this classification. The results gaudily conclude the classification of machine learning classifiers for thermal and HOG images with the highest accuracy of 90% for the random forest classifier.