Abstract:As a natural disaster with high suddenness and great destructiveness, fire has long posed a major threat to human society and ecological environment. In recent years, with the rapid development of smart city and Internet of Things (IoT) technologies, fire detection systems based on deep learning have gradually become a key means to cope with fire hazards. However, existing fire detection models still have many challenges in terms of detection accuracy and real-time performance in complex contexts. To address these issues, we propose two key modules: EAConv (Efficient Attention Convolution) and EADown (Efficient Attention Downsampling). The EAConv module significantly improves the feature extraction efficiency by combining an efficient attention mechanism with depth-separable convolution, while the EADown module enhances the accuracy and efficiency of feature downsampling by utilizing spatial and channel attention mechanisms in combination with pooling operations. Based on these two modules, we design an efficient and lightweight flame detection model, EFA-YOLO (Efficient Feature Attention YOLO). Experimental results show that EFA-YOLO has a model parameter quantity of only 1.4M, GFLOPs of 4.6, and the inference time per image on the CPU is only 22.19 ms. Compared with existing mainstream models (e.g., YOLOv5, YOLOv8, YOLOv9, and YOLOv10), EFA-YOLO exhibits a significant enhancement in detection accuracy (mAP) and inference speed, with model parameter amount is reduced by 94.6 and the inference speed is improved by 88 times.

What problem does this paper attempt to address?

### Problems Addressed by the Paper This paper aims to address the challenges of detection accuracy and real-time performance of existing fire detection models in complex backgrounds. Specifically, fire, as a highly sudden and destructive natural disaster, has long posed a significant threat to human society and the ecological environment. With the rapid development of smart cities and IoT technology, fire detection systems based on deep learning have gradually become a key means of addressing fire hazards. However, existing fire detection models still face many issues in terms of detection accuracy and real-time performance in complex backgrounds. To tackle these problems, the authors propose two key modules: EAConv (Efficient Attention Convolution) and EADown (Efficient Attention Downsampling). These two modules significantly improve feature extraction efficiency by combining efficient attention mechanisms and depthwise separable convolutions, and enhance the accuracy and efficiency of feature downsampling by combining spatial and channel attention mechanisms with pooling operations. Based on these two modules, the authors designed an efficient and lightweight fire detection model—EFA-YOLO (Efficient Feature Attention YOLO). ### Main Contributions 1. **Proposed Two Key Modules**: - **EAConv Module**: Significantly improves the efficiency and accuracy of feature extraction by combining efficient attention mechanisms and depthwise separable convolutions. - **EADown Module**: Enhances the accuracy and efficiency of feature downsampling by combining spatial and channel attention mechanisms with pooling operations. 2. **Designed and Implemented an Efficient and Lightweight Fire Detection Model EFA-YOLO**: - This model significantly reduces the number of model parameters and computational complexity while maintaining high detection accuracy, making it particularly suitable for real-time fire detection applications. - EFA-YOLO improves detection capability in complex backgrounds through effective multi-scale feature fusion. - Experimental results show that EFA-YOLO exhibits significant improvements in detection accuracy (mAP) and inference speed, with a 94.6% reduction in model parameters and an 88-fold increase in inference speed. 3. **Provided an Effective Lightweight Solution for Embedded Devices and Smart City Scenarios**: - Due to its lightweight design, EFA-YOLO can run efficiently on resource-constrained devices, making it suitable for fire monitoring systems in smart cities and providing a flexible and scalable model architecture for future fire detection technology. ### Experimental Results Experimental results show that EFA-YOLO outperforms existing mainstream models (such as YOLOv5, YOLOv8, YOLOv9, and YOLOv10) across multiple evaluation metrics, particularly in detection accuracy (mAP) and inference speed. Specifically: - **Precision and Recall**: EFA-YOLO shows significant improvements in precision and recall, especially in handling small target flames, where its detection boxes are closer to the true labels and more compact. - **mAP50 and mAP50-95**: EFA-YOLO performs exceptionally well on the mAP50 metric, showing significant improvements over other models, indicating stronger object detection capabilities in complex scenes. - **Model Parameters (Params), Computational Complexity (GFLOPs), and Model Size**: EFA-YOLO also excels in these metrics, with a substantial reduction in model parameters, computational complexity, and model size. In summary, this paper effectively addresses the detection accuracy and real-time performance issues of existing fire detection models in complex backgrounds by proposing the EFA-YOLO model, providing new ideas and methods for the development of fire detection technology.

EFA-YOLO: An Efficient Feature Attention Model for Fire and Flame Detection

Ea-yolo: efficient extraction and aggregation mechanism of YOLO for fire detection

Flame Detection Using Deep Learning

DG-YOLO: A Novel Efficient Early Fire Detection Algorithm Under Complex Scenarios

Conflagration-YOLO: a lightweight object detection architecture for conflagration

Fire-PPYOLOE: An Efficient Forest Fire Detector for Real-Time Wild Forest Fire Monitoring

YOLOv5-CSF: an improved deep convolutional neural network for flame detection

Fire-YOLO: A Small Target Object Detection Method for Fire Inspection

A Novel Intelligent Online Fire Monitoring Approach for Underground Pipe Galleries

FL-YOLOv7: A Lightweight Small Object Detection Algorithm in Forest Fire Detection

A modified YOLOv5 architecture for efficient fire detection in smart cities

An Explainable AI-Based Modified YOLOv8 Model for Efficient Fire Detection

LUFFD-YOLO: A Lightweight Model for UAV Remote Sensing Forest Fire Detection Based on Attention Mechanism and Multi-Level Feature Fusion

An Improved YOLOv5s Fire Detection Model

Research on Fire Smoke Detection Algorithm Based on Improved YOLOv8

CF-YOLO: a capable forest fire identification algorithm founded on YOLOv7 improvement

Fs-yolo: fire-smoke detection based on improved YOLOv7

Fire Safety Detection Based on CAGSA-YOLO Network

Multi-Scale Video Flame Detection for Early Fire Warning Based on Deep Learning

Forest Fire Detection Based on Lightweight Yolo

YOLOv7-FIRE: A tiny-fire identification and detection method applied on UAV