Abstract
Gross Vehicle Weight Rating (GVWR)-based vehicle activity data is widely used in freight planning, fuel efficiency evaluation, and on-road emission estimation. However, a vehicle’s GVWR remains challenging to obtain using existing highway sensor infrastructure. This paper describes a novel approach to acquire GVWR-based classification data through the fusion of two complementary infrastructure-based sensing technologies: inductive loop sensors and side-fire video cameras. While inductive loops are widely deployed in the U.S., they only provide single-dimensional data with limited information. Side-fire cameras can offer richer details to enhance vehicle classification. Accordingly, an open-source intelligence (OSINT) method was used to establish a GVWR-based vehicle dictionary, linking vehicle specifications from online data sources to GVWR classes. A dataset comprising 9,154 vehicle inductive loop signatures paired with images was then collected and annotated according to the pre-defined dictionary. Next, signature-based and image-based classification models were developed for GVWR classification. Each model was designed to function independently. A signature-based GVWR classification model was trained with a multi-layer perceptron (MLP) neural net architecture and optimized through the implementation of a weighted cross-entropy loss function. A two-stage image-based GVWR classification framework was designed to extract vehicle objects and classify them based on the GVWR scheme. Finally, a linear fusion model was implemented to combine the output of the signature- and image-based models to achieve an improvement over each standalone classification model. The sensor fusion framework significantly outperformed each individual sensing technology, achieving an average correct classification rate of 0.97 and an score of 0.96, which surpasses state-of-the-art methods.