Neural Network-based Acoustic Vehicle Counting

TL;DR

This paper presents a neural network approach for acoustic vehicle counting that predicts pass-by instants and introduces a deep learning counting method that operates without local minima detection, improving robustness and accuracy.

Contribution

It introduces a neural network-based two-stage regression for distance prediction and a novel deep learning counting method independent of minima detection.

Findings

Neural network regression outperforms support vector regression.

The pass-by counting error has a 95% confidence interval within [0.28%, -0.55%].

Removing low frequencies improves counting performance.

Abstract

This paper addresses acoustic vehicle counting using one-channel audio. We predict the pass-by instants of vehicles from local minima of clipped vehicle-to-microphone distance. This distance is predicted from audio using a two-stage (coarse-fine) regression, with both stages realised via neural networks (NNs). Experiments show that the NN-based distance regression outperforms by far the previously proposed support vector regression. The $ 95\% $ confidence interval for the mean of vehicle counting error is within $[0.28\%, -0.55\%]$. Besides the minima-based counting, we propose a deep learning counting that operates on the predicted distance without detecting local minima. Although outperformed in accuracy by the former approach, deep counting has a significant advantage in that it does not depend on minima detection parameters. Results also show that removing low frequencies in features improves the counting performance.