A Review of YOLO-Based Real-Time Fire Detection Models and Applications

View at publisher Summarise with VIRAT
Authors: Kishor Kumar, Manish Sharma, Richa Sharma
Cited by: 0
References: 13

Summary

No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.

Summarise with VIRAT Free to try with your Science Catalogs account.

Abstract

Fire detection is essential for intelligent safety and monitoring systems as it can ensure timely and accurate detections that prevent destruction and loss of life. Traditional sensor-based solutions (i.e. smoke detectors and heat detectors), often have slow response times and high false-positive rates due to changes in the environment. Recent developments in deep learning, in particular YOLO-based object detection models, allow for camera-based fire detection to be both accurate and near-real-time. In this paper, we detail a systematic review of YOLO fire detection studies, including hybrid CNN–YOLO pipelines, edge-enhanced models, attention modules, and embedded-optimized lightweight versions of YOLO. Each study is reviewed in terms of its methodology, dataset, key performance metrics, advantages and disadvantages. Across studies, a side-by-side comparison shows the trade off between accuracy, computational complexity, and contextual specificity (i.e. substation monitoring, UAV-based forest monitoring, urban CCTV). Finally, the survey ends with an overview of open challenges in terms of dataset variability, generalizability to new contextual applications, and multi-modal data integration, which offers a pathway for future development of advanced fire detection systems.

Fields of Study & Subjects

Fire Detection and Safety Systems Advanced Neural Network Applications Fire dynamics and safety research

Publication details

Journal / source
International Conference on Signal Processing and Electronics Design (ICSPED)
Pages
96–101
Publication date
23 January 2026
Language
EN