On this page, you can download dataset and codes related to paper
Robust Audio-Based Vehicle Counting in Low-to-Moderate Traffic Flow Slobodan Djukanović, Jiři Matas, Tuomas Virtanen 31st IEEE Intelligent Vehicles Symposium, October 2020, Las Vegas, NV, United States
Paper abstract:
The paper presents a method for audio-based vehicle counting (VC) in low-to-moderate traffic using one-channel sound. We formulate VC as a regression problem, i.e., we predict the distance between a vehicle and the microphone. Minima of the proposed distance function correspond to vehicles passing by the microphone. VC is carried out via local minima detection in the predicted distance. We propose to set the minima detection threshold at a point where the probabilities of false positives and false negatives coincide so they statistically cancel each other in total vehicle number. The method is trained and tested on a traffic-monitoring dataset comprising 422 short, 20-second one-channel sound files with a total of 1421 vehicles passing by the microphone. Relative VC error in a traffic location not used in the training is below 2% within a wide range of detection threshold values. Experimental results show that the regression accuracy in noisy environments is improved by introducing a novel high-frequency power feature.
You can download the paper here.
The dataset (audio-video recordings of the road traffic) has been recorded using a GoPro Hero5 Session camera, installed on a sidewalk, at a safe distance of at least 0.5 m from the road and at the height of around 1.2 m. The dataset acquisition took place from September to November 2019.
The recorded video material is divided into 422 short, 20-second non-overlapping video clips. The total material includes 1421 vehicles passing by the microphone (on average 303 vehicles/hour/lane).
The dataset is split into two parts: VC-PRG-1:5 (first five recording locations, 250 sound files with 841 vehicles in total) and VC-PRG-6 (sixth location, 172 sound files with 580 vehicles in total).
Annotation data contain the pass-by-microphone times of vehicles (relative time from the beginning of the file, measured in seconds with a two-decimal precision). Five vehicle classes are considered in the annotation: motorcycles, cars, vans, buses and trucks.
The dataset is publicly released to promote comparison (links below). The use of the data is free in the context of scientific research (the use for research in the commercial domain is also allowed).
![]() |
![]() |
![]() |
![]() |
![]() |
![]() |
Sample images from the camera in six different two-lane one-direction roads in Prague, Czech Republic. Various camera positions (both road sides and different angles of the camera with respect to the road) were taken in order not to be sensitive to the actual camera position.
The project was implemented in the Python 3 programming language. In file Instructions.pdf, you can find a short description of each file and folder in the project. We have included .h5 files with features and annotations (folder datasets), as well as all results presented in Figs. 6-10 in the paper (folders SVR, Probs and RVCE).
In case you use this dataset in your research, please cite the following paper:
@inproceedings{AudioVehicleDataset,
author = "Slobodan Djukanovi\'{c} and Ji\v{r}i Matas and Tuomas Virtanen",
title = "Robust audio-based vehicle counting in low-to-moderate traffic flow",
booktitle = "IV2020 Conference",
year = "2020",
pages = "1337-1343"
}
For all inquiries and questions regarding the paper, the dataset and the codes, contact Slobodan Djukanović.
This material is presented to ensure timely dissemination of scholarly and technical work. Copyright and all rights therein are retained by authors or by other copyright holders. All persons copying this information are expected to adhere to the terms and constraints invoked by each author's copyright. In most cases, these works may not be reposted without the explicit permission of the copyright holder. Personal use of this material is permitted. However, permission to reprint/republish this material for advertising or promotional purposes or for creating new collective works for resale or redistribution to servers or lists, or to reuse any copyrighted component of the works published in IEEE publications in other works must be obtained from the IEEE.