| DOI | Resolve DOI: https://doi.org/10.1109/SAS58821.2023.10254050 |
|---|
| Author | Search for: Zhang, Zhenyu; Search for: Shen, Yichun1; Search for: Valdes, Julio J.1; Search for: Huq, Saiful; Search for: Wallace, Bruce; Search for: Green, James; Search for: Xi, Pengcheng1; Search for: Goubran, Rafik |
|---|
| Affiliation | - National Research Council of Canada. Digital Technologies
|
|---|
| Format | Text, Article |
|---|
| Conference | 2023 IEEE Sensors Applications Symposium (SAS), July 18-20, 2023, Ottawa, Ontario, Canada |
|---|
| Subject | aging in place; sound event detection; acoustic scene classification; convolutional neural network; pooling |
|---|
| Abstract | Smart sensing and AI technologies are playing an increasingly important role in supporting aging in place, including embedded microphone devices that can contribute to safety by recognizing home events that need attention. In this work, we study sound event detection through classifying different sources of sounds. Specifically, we study the performance of models applied to a dataset of recordings in a domestic environment. We propose a light-weight, VGG-like model with a customized loss function. Relative to existing methods, the proposed model has significantly reduced the number of parameters from 2.6M to 75K and is therefore effective for use with embedded devices. Our proposed model can be deployed on low-cost edge-computing devices and still achieves comparable performance as those from complex models. |
|---|
| Publication date | 2023-09-22 |
|---|
| Publisher | IEEE |
|---|
| In | |
|---|
| Language | English |
|---|
| Peer reviewed | Yes |
|---|
| Export citation | Export as RIS |
|---|
| Report a correction | Report a correction (opens in a new tab) |
|---|
| Record identifier | 4b3f4d1d-d90d-4758-a534-7ce1d77d7f28 |
|---|
| Record created | 2023-09-26 |
|---|
| Record modified | 2023-09-26 |
|---|
