FD-Net: Fire detection from CC videos with frame inter correlated pattern analysis and Bi-directional SVM pyramid networks
Year: 2025
Type: article
Abstract: Fire detection from closed-circuit (CC) camera videos is crucial for real-time security applications in industries, urban areas, and public spaces. Statistically, fire incidents cause over 1.3 million injuries and approximately $10 billion in property damage annually. Due to environmental noise and light reflections, fire detection methods often struggle with high false positive rates. The proposed Fire Detection Network (FD-Net) enhances fire detection accuracy by incorporating a robust data preprocessing stage that filters out background noise and stabilises video frames. The system also utilises the Frame Inter Correlated Pattern (FICP) feature analysis, improving the identification of unique frame movement patterns. The Bi-directional Support Vector Machine-based Pyramid Network (BSVM-PN) is then applied to classify fire incidents with enhanced precision, leveraging a multi-layer architecture that processes video data bidirectionally to account for frame variations over time. This approach reduces false positives and increases detection speed, making it highly suitable for real-time fire detection across diverse environments.
Source: Systems and Soft Computing
Authors: Nivedita Das, G.S.N. Murthy, Kapu Akhila, Ailada Uma Maheshwar Rao, M. Ruchitha, Nagender
Institution: Aditya University
Language: English
Subfield: Safety, Risk, Reliability and Quality
Field: Engineering
Domain: Physical Sciences
Open Access status: gold
Estimated APC paid: $1,870
OpenAlex ID: W4414282965