Real-Time Forest Fire Detection Framework Based on Artificial Intelligence Using Color Probability Model and Motion Feature Analysis

• Published in: Fire (Basel, Switzerland) Vol. 5; no. 1; p. 23
• Main Authors: Wahyono, Harjoko, Agus, Dharmawan, Andi, Adhinata, Faisal Dharma, Kosala, Gamma, Jo, Kang-Hyun
• Format: Journal Article
• Language: English
• Published: Basel MDPI AG 01.02.2022
• Subjects: Accuracy; Artificial intelligence; Classification; Color; color probability; Damage prevention; Datasets; Deep learning; early warning system; Early warning systems; Fire damage; fire detection; Forest & brush fires; Forest ecosystems; forest fire; Forest fire detection; Forest fires; Forests; Histograms; Image classification; Image segmentation; Learning algorithms; Machine learning; motion feature analysis; Neural networks; Performance evaluation; real-time process; Smoke detectors; Support vector machines; Terrestrial ecosystems; Indonesia
• ISSN: 2571-6255; 2571-6255
• DOI: 10.3390/fire5010023

As part of the early warning system, forest fire detection has a critical role in detecting fire in a forest area to prevent damage to forest ecosystems. In this case, the speed of the detection process is the most critical factor to support a fast response by the authorities. Thus, this article proposes a new framework for fire detection based on combining color-motion-shape features with machine learning technology. The characteristics of the fire are not only red but also from their irregular shape and movement that tends to be constant at specific locations. These characteristics are represented by color probabilities in the segmentation stage, color histograms in the classification stage, and image moments in the verification stage. A frame-based evaluation and an intersection over union (IoU) ratio was applied to evaluate the proposed framework. Frame-based evaluation measures the performance in detecting fires. In contrast, the IoU ratio measures the performance in localizing the fires. The experiment found that the proposed framework produced 89.97% and 10.03% in the true-positive rate and the false-negative rate, respectively, using the VisiFire dataset. Meanwhile, the proposed method can obtain an average of 21.70 FPS in processing time. These results proved that the proposed method is fast in the detection process and can maintain performance accuracy. Thus, the proposed method is suitable and reliable for integrating into the early warning system.