Abstract:Fires can cause severe damage to heritage buildings, making timely fire detection essential. Traditional dense cabling and drilling can harm these structures, so reducing the number of cameras to minimize such impact is challenging. Additionally, avoiding false alarms due to noise sensitivity and preserving the expertise of managers in fire-prone areas is crucial. To address these needs, we propose a fire detection method based on indirect vision, called Mirror Target YOLO (MITA-YOLO). MITA-YOLO integrates indirect vision deployment and an enhanced detection module. It uses mirror angles to achieve indirect views, solving issues with limited visibility in irregular spaces and aligning each indirect view with the target monitoring area. The Target-Mask module is designed to automatically identify and isolate the indirect vision areas in each image, filtering out non-target areas. This enables the model to inherit managers' expertise in assessing fire-risk zones, improving focus and resistance to interference in fire <a class="link-external link-http" href="http://detection.In" rel="external noopener nofollow">this http URL</a> our experiments, we created an 800-image fire dataset with indirect vision. Results show that MITA-YOLO significantly reduces camera requirements while achieving superior detection performance compared to other mainstream models.

What problem does this paper attempt to address?

The problem that this paper attempts to solve is to achieve efficient and low - invasive fire detection in cultural heritage buildings. Specifically, the paper focuses on the following problems: 1. **Limitations of traditional fire detection methods**: - Although traditional smoke alarms, temperature sensors, and manual fire alarm buttons are effective, installing these devices requires a large number of drillings and wirings, which may damage the structure and appearance of cultural heritage buildings. - The coverage of these traditional methods in irregular spaces is limited, and a large number of cameras are required to achieve comprehensive monitoring, increasing cost and complexity. 2. **Challenges of deep - learning methods**: - Existing deep - learning - based fire detection methods mainly rely on direct vision, which will be affected by occlusion in irregular spaces, resulting in limited vision. - These methods cannot effectively filter out low - risk areas, are prone to false alarms, and affect the accuracy of detection. 3. **Requirements for cultural heritage protection**: - Cultural heritage buildings have important historical and cultural values, and any damage to the building structure will cause irreversible losses. - Therefore, a method that can effectively detect fires without causing damage to the building is required. To address the above challenges, the paper proposes a fire detection method based on indirect vision - Mirror Target YOLO (MITA - YOLO). This method expands the camera's field of view by using mirrors, reduces the number of cameras, and automatically identifies and isolates indirect vision areas through the Target - Mask module, improving detection accuracy and anti - interference ability. Specific improvements include: - **Integrating indirect vision technology**: Use mirrors to expand the camera's field of view, reduce the number of cameras deployed in irregular spaces, and align the indirect vision with the target detection area. - **Proposing the Target - Mask module**: Automatically identify the boundaries of indirect vision, filter out irrelevant areas, and enhance detection accuracy and robustness. - **Constructing a unique data set**: Simulate real - life scenarios in cultural heritage buildings and construct a fire detection data set containing 800 indirect vision images. - **Experimental verification**: Conduct comparative experiments with mainstream models such as YOLOv8. The results show that MITA - YOLO has improved by 3.7% and 3% in mAP50 and recall rate respectively, demonstrating superior performance. Through these improvements, MITA - YOLO can achieve efficient fire detection without damaging cultural heritage buildings.

Mirror Target YOLO: An Improved YOLOv8 Method with Indirect Vision for Heritage Buildings Fire Detection

A Novel Intelligent Online Fire Monitoring Approach for Underground Pipe Galleries

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

Mirror-Yolo: An attention-based instance segmentation and detection model for mirrors

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

An Improved YOLOv5s Fire Detection Model

Improving real-time object detection in Internet-of-Things smart city traffic with YOLOv8-DSAF method

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

Dual-YOLO Architecture from Infrared and Visible Images for Object Detection

A Small Target Forest Fire Detection Model Based on YOLOv5 Improvement

A modified YOLOv5 architecture for efficient fire detection in smart cities

Image-Based Fire Detection in Industrial Environments with YOLOv4

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

Site Safety Target Detection With Improved YOLOv5

Fire Detection From Image and Video Using YOLOv5

GPAC-YOLOv8: Lightweight Target detection for fire scenarios

Industrial and Mining Fire Detection Algorithm Based on Improved YOLO

Research on Infrared Dim Target Detection Based on Improved YOLOv8

Two-stage deep learning-based video image recognition of early fires in heritage buildings

Mirror-Yolo: A Novel Attention Focus, Instance Segmentation and Mirror Detection Model

A Robust Fire Detection Model via Convolution Neural Networks for Intelligent Robot Vision Sensing