What question did this study set out to answer?

The aim is to enhance smoke and fire detection using a decision-level framework that integrates various sensor data and machine learning models.

February 12, 2026Open Access

On the Performance of YOLO and ML/DL Models for Lightweight, Real-Time Smoke and Fire Detection on an x86-Based Edge Platform: An Explainable Decision-Level Sensor Fusion Framework

Key Points

The aim is to enhance smoke and fire detection using a decision-level framework that integrates various sensor data and machine learning models.
Developed a framework integrating temperature, humidity, and gas readings with ML and DL models.
Evaluated 16 ML and DL models on an IoT sensor dataset and various YOLO models on an image-label dataset.
Prioritized lightweight architectures suitable for edge devices and evaluated using standard metrics.
Incorporated explainable AI techniques like SHAP for ML models and LIME for YOLO detectors.
The framework shows a high level of internal consistency and reliability.
Demonstrated effectiveness in reducing false alarms.
Evaluated models achieved desirable performance metrics like F1-score and mean average precision.

Abstract

In this study, a decision-level detection framework is presented and evaluated; it integrates sensor data (e.g., temperature, humidity, gas readings) with machine learning (ML) models and computer vision-based smoke and fire detection systems, in an effort to increase overall robustness, as well as false-alarm reduction. To this end, sixteen (16) ML and deep learning (DL) models are employed on an Internet of Things (IoT) sensor dataset. Moreover, a range of YOLO (You Only Look Once) models, such as older versions (YOLOv5n, YOLOv8n), as well as newer versions (YOLOv10n, YOLOv11n, YOLOv12n), are employed on an image-label-based dataset. Model selection initially prioritizes lightweight architectures that are suitable for resource-constrained edge devices. Afterwards, the selected models are evaluated via well-known metrics, such as parameter count, F1-score/mean average precision (mAP) and real-time inference latency. In the same context, explainable AI (XAI) techniques, such as SHAP (SHapley Additive exPlanations) for ML models and LIME (Local Interpretable Model-agnostic Explanations) for the YOLO detectors, are integrated into the platform as well. According to the presented results, the Explainable Sensor Fusion (ESF) framework demonstrates a high level of internal consistency and logical reliability through its decision-level fusion paradigm in a controlled environment.

Bookmark

View Full Paper

Cite This Study

Dibra et al. (Tue,) studied this question.

synapsesocial.com/papers/698d6edc5be6419ac0d54bb2 https://doi.org/https://doi.org/10.3390/electronics15040747

Bookmark

View Full Paper