Name
Abstract | ||
|---|---|---|
Fire is a dangerous disaster that takes many lives and human property. Fire happens everywhere, especially in areas with high temperatures or hot sun. Fires can be caused by humans or by nature. Therefore, an early warning of fire is necessary to reduce the damage. Research in many different fields has long been focused on fire alerts. This paper proposes a fire alarm system based on a lightweight convolutional neural network. The design takes the advantage of convolution layers, depthwise separable convolution layers, inception module, and softmax function to optimize network parameters while ensuring feature extraction and classification. This network is trained and evaluated on FireNet dataset with an accuracy of 97.14%. In addition, this work also builds and implements the fire video testing systems on low-computation devices such as CPU-Based personal computer and embedded devices. |
Year | DOI | Venue |
|---|---|---|
2022 | 10.1109/MAPR56351.2022.9924654 | 2022 International Conference on Multimedia Analysis and Pattern Recognition (MAPR) |
Keywords | DocType | ISSN |
Convolutional neural network,fire classification,inception network,fire warning system | Conference | 2770-6842 |
ISBN | Citations | PageRank |
978-1-6654-7411-5 | 0 | 0.34 |
References | Authors | |
6 | 5 | |
Name | Order | Citations | PageRank |
|---|---|---|---|
| Duy-Linh Nguyen | 1 | 3 | 1.41 |
| Muhamad Dwisnanto Putro | 2 | 8 | 3.91 |
| Xuan-Thuy Vo | 3 | 0 | 0.34 |
| Tien-Dat Tran | 4 | 0 | 0.68 |
| Kang-Hyun Jo | 5 | 0 | 0.34 |