Nowadays, cars host an increasing number of sensors to improve safety, efficiency and comfort. Acoustic sensors have been proposed, in recent works, to acquire information related to the road conditions. Thanks to effectiveness of Deep Learning techniques in analyzing audio data, new scenarios can be envisioned. Based on previous works employing Convolutional Neural Networks (CNN) trained specifically for either of the two tasks, we compare the performance of a CNN trained to jointly classify both wetness and roughness and a transfer learning approach where two CNN, specialized for one task singularly, are joined in a single network. Then we investigate several issues related to the deployment of a classification system able to detect road wetness and roughness on an embedded processor. The first approach seems to score better in our tests and is, thus, selected for the deployment to an ARM-based embedded processor. The computational cost, Real-Time Factor and memory requirements are discussed, as well as the degradation related to the extraction of the features and the weights quantization. Results are promising and show that such an application can be readily deployed on off-the-shelf hardware.

Road Type Classification Using Acoustic Signals: Deep Learning Models and Real-Time Implementation / Pepe, G.; Gabrielli, L.; Principi, E.; Squartini, S.; Cattani, L.. - 184:(2021), pp. 33-43. [10.1007/978-981-15-5093-5_4]

Road Type Classification Using Acoustic Signals: Deep Learning Models and Real-Time Implementation

Pepe G.;Gabrielli L.;Principi E.;Squartini S.;
2021-01-01

Abstract

Nowadays, cars host an increasing number of sensors to improve safety, efficiency and comfort. Acoustic sensors have been proposed, in recent works, to acquire information related to the road conditions. Thanks to effectiveness of Deep Learning techniques in analyzing audio data, new scenarios can be envisioned. Based on previous works employing Convolutional Neural Networks (CNN) trained specifically for either of the two tasks, we compare the performance of a CNN trained to jointly classify both wetness and roughness and a transfer learning approach where two CNN, specialized for one task singularly, are joined in a single network. Then we investigate several issues related to the deployment of a classification system able to detect road wetness and roughness on an embedded processor. The first approach seems to score better in our tests and is, thus, selected for the deployment to an ARM-based embedded processor. The computational cost, Real-Time Factor and memory requirements are discussed, as well as the degradation related to the extraction of the features and the weights quantization. Results are promising and show that such an application can be readily deployed on off-the-shelf hardware.
2021
Smart Innovation, Systems and Technologies
978-981-15-5092-8
978-981-15-5093-5
File in questo prodotto:
Non ci sono file associati a questo prodotto.

I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.

Utilizza questo identificativo per citare o creare un link a questo documento: https://hdl.handle.net/11566/286029
 Attenzione

Attenzione! I dati visualizzati non sono stati sottoposti a validazione da parte dell'ateneo

Citazioni
  • ???jsp.display-item.citation.pmc??? ND
  • Scopus 0
  • ???jsp.display-item.citation.isi??? ND
social impact