Fast and accurate fire detection is of great significance to the sustainable development of human society and the earth ecology. The existence of objects with similar characteristics to the fire increases the difficulty of fire detection based on vision. Improving the accuracy of fire detection by digging deeper visual features of the fire has always been a concerned problem.

Recently, researchers from Hainan Acoustics Laboratoty of the Institute of Acoustics of the Chinese Academy of Sciences (IACAS) proposed an efficient deep learning model for vision-based fire detection based on multiscale feature extraction, implicit deep supervision, and channel attention mechanism. This method realized fast and accurate fire detection, showing great practical value potential.

It is worth noting that the advantages of this model are more remarkable in complex scenes. Moreover, the size of this model is relatively small, making it easy to implement on resource-constrained devices.

Researchers utilized real-time acquired image as the input of the model and normalized the image. At the low-level feature extraction stage, they introduced the multiscale feature extraction mechanism to enrich spatial detail information, which enhanced the discriminative ability of fire-like objects. Then, the implicit deep supervision mechanism was employed to enhance the interaction among information flows. Finally, researchers used the channel attention mechanism to selectively emphasize the features contributing to the task, and effectively suppressed the interference of image noise.

The experimental results demonstrated that the accuracy of this efficient deep learning model for fire detection achieved 95.3%, but the model size was only 4.80 MB. The model could process 63.5 frames per second on NVIDIA GTX 2080TI, meaning that it is able to detect fire in real-time. Compared with the current deep-learning based methods, this model showed great improvement not only in detection accuracy but also in model size and detection speed.

This research provides a feasible solution for realizing fast and accurate fire detection and makes it possible for vision-based fire detection to become practical.

This research, published in the IEEE Transactions on Image Processing, was supported by the Important Science and Technology Project of Hainan Province.

Figure 1. Deep learning model for fire detection. (Image by IACAS)

Figure 2. The comparison results with the current deep-learning based fire detection methods. (Image by IACAS) 

Reference:

LI Songbin, YAN Qiandong, LIU Peng, An Efficient Fire Detection Method Based on Multiscale Feature Extraction, Implicit Deep Supervision and Channel Attention Mechanism. IEEE Transactions on Image Processing, vol. 29, pp. 8467-8475, 2020. DOI: 10.1109/TIP.2020.3016431.

Contact:

ZHOU Wenjia

Institute of Acoustics, Chinese Academy of Sciences, 100190 Beijing, China

E-mail: media@mail.ioa.ac.cn