Design of Audio based Accident and Crime Detection using Simple Architecture of Neural Network

Afis Pratama, Sritrusta Sukaridhoto, Mauridhi Purnomo, Achmad Basuki, Rizqi Budiarti

2021

Abstract

The accident and crime happened on the road still increasing nowadays. Those two events were considered as emergency event that need a quick response. In this research, a method to detect accident and crime were proposed. The proposed method uses audio data as input and extracting the Mel spectrogram as the feature, which later be fed to our simple neural network architectures. We classify our dataset into engine_idling, car_crash, and gun_shot classes to represent normal, accident, and crime condition on the road. Our simple CNN architecture obtains accuracy of 95.31% and 93.75% with 200ms and 1000ms segment duration respectively, and our simple RNN architecture obtains 86.67% and 58.67% by using 200ms and 1000ms segment duration respectively. We can conclude that the best simple architecture was performed by CNN architecture with 200ms segment duration.

Paper Citation

in Harvard Style

Pratama A., Sukaridhoto S., Purnomo M., Basuki A. and Budiarti R. (2021). Design of Audio based Accident and Crime Detection using Simple Architecture of Neural Network. In Proceedings of the 4th International Conference on Applied Science and Technology on Engineering Science - Volume 1: iCAST-ES, ISBN 978-989-758-615-6, pages 76-81. DOI: 10.5220/0010940300003260

in Bibtex Style

@conference{icast-es21,
author={Afis Pratama and Sritrusta Sukaridhoto and Mauridhi Purnomo and Achmad Basuki and Rizqi Budiarti},
title={Design of Audio based Accident and Crime Detection using Simple Architecture of Neural Network},
booktitle={Proceedings of the 4th International Conference on Applied Science and Technology on Engineering Science - Volume 1: iCAST-ES,},
year={2021},
pages={76-81},
publisher={SciTePress},
organization={INSTICC},
doi={10.5220/0010940300003260},
isbn={978-989-758-615-6},
}

in EndNote Style

TY - CONF

JO - Proceedings of the 4th International Conference on Applied Science and Technology on Engineering Science - Volume 1: iCAST-ES,
TI - Design of Audio based Accident and Crime Detection using Simple Architecture of Neural Network
SN - 978-989-758-615-6
AU - Pratama A.
AU - Sukaridhoto S.
AU - Purnomo M.
AU - Basuki A.
AU - Budiarti R.
PY - 2021
SP - 76
EP - 81
DO - 10.5220/0010940300003260