Improvement and Validation of a Smart Road Traffic Noise Model Based on Vehicles Tracking Using Image Recognition: EAgLE 3.0.

Noise coming from road traffic represents a major contributor to the high levels of noise to which people are continuously exposed-especially in urban areas-throughout all of Europe. Since it represents a very detrimental pollutant, the assessment of such noise is an important procedure. Noise levels can be measured or simulated, and, in this second case, for the building of a valid model, a proper collection of input data cannot be left out of consideration. In this paper, the authors present the development of a methodology for the collection of the main inputs for a road traffic noise model, i.e., vehicle number, category, and speed, from a video recording of traffic on an Italian highway. Starting from a counting and recognition tool already available in the literature, a self-written Python routine based on image inference has been developed for the instantaneous detection of the position and speed of vehicles, together with the categorization of vehicles (light or heavy). The obtained data are coupled with the CNOSSOS-EU model to estimate the noise power level of a single vehicle and, ultimately, the noise impact of traffic on the selected road. The results indicate good performance from the proposed model, with a mean error of -1.0 dBA and a mean absolute error (MAE) of 3.6 dBA.