Light predictions from dashboard cameras

Abstract

© 2017 IEEE. The objective of this paper is to develop a predictive model for classification of lights in images from the dashboard camera video through the use of machine learning algorithms. In this study, the authors used a DS dataset, comprising of 300 images. The dataset was categorized into three different types of lightness: day light, low light, and night light. The dataset was analyzed using four features, which include mean, minimum, maximum, and summation of histogram of the images. The four machine learning algorithms that were used as a classifier include Decision Tree, Naïve Bayes, Neural Network and Sequential minimal optimization. The results obtained from this study indicated that Neural Network algorithm generated the most desirable results with respect to other algorithms. The accuracy rate of the prediction model is 98.518 percent.