Multilevel weather detection based on images: a machine learning approach with histogram of oriented gradient and local binary pattern-based features

The primary objective of this study was to develop a trajectory-level weather detection system capable of providing real-time weather information at the road surface level using only a single video camera. Two texture-based features, including histogram of oriented gradient (HOG) and local binary pattern (LBP), were extracted from images and used as classification parameters to train the weather detection models using several machine learning classifiers, such as gradient boosting (GB), random forest (RF), and support vector machine (SVM). In addition, a unique multilevel model, based on a hierarchical structure, was also proposed to increase detection accuracy. Evaluation results revealed that the multilevel model provided an overall accuracy of 89.2%, which is 3.2%, 7.5%, and 7.9% higher compared to the SVM, RF, and GB model, respectively, using the HOG features. Considering the LBP features, the multilevel model also produced the best performance with an overall accuracy of 91%, which is 1.6%, 8.6%, and 9% higher compared to the SVM, RF, and GB models, respectively. A sensitivity analysis using the proposed multilevel model revealed that the classification accuracy improved with the increasing number of HOG and LBP features at the expense of more computational powers. The proposed weather detection method is cost-efficient and can be made widely available mainly due to the recent booming of smartphone cameras and can be used to expand and update the current weather-based variable speed limit (VSL) systems in a connected vehicle (CV) environment.