A fire incident can cause significant damage that is difficult to estimate. It can happen anywhere and in any situation, there is no limit to the damage caused by it, and it can be seen widely at the social and economic levels. The state-of-the-art algorithms VGG16, VGG19, Inception v3, and Xception are widely used benchmarks for image classification tasks. We proposed a fire detection model on custom datasets. Our model uses a combination of 1x1, 1x3, 3x1, and 3x3 filters in its multipath convolutional layers to learn both local and global features. The proposed model achieved better performance than the some state-of-the-art algorithms when tested on custom datasets while detecting fires in images.
Paper
Full text
An Intelligent Image Fire Detection Approach Based On Deep Convolutional Neural Network
Semantic Scholar · Computer Science · 2023
Abstract
A fire incident can cause significant damage that is difficult to estimate. It can happen anywhere and in any situation, there is no limit to the damage caused by it, and it can be seen widely at the social and economic levels. The state-of-the-art algorithms VGG16, VGG19, Inception v3, and Xception are widely used benchmarks for image classification tasks. We proposed a fire detection model on custom datasets. Our model uses a combination of 1x1, 1x3, 3x1, and 3x3 filters in its multipath convolutional layers to learn both local and global features. The proposed model achieved better performance than the some state-of-the-art algorithms when tested on custom datasets while detecting fires in images.