FERD-YOLO:A Lightweight Fire Detection Network for Mobile Robots

With social development and the continuous advancement of urban modernization, fire safety issues receive increasing attention. However, in recent years, fire incidents continue to occur frequently, posing serious threats to people's lives and property. Traditional fire detection methods are prone to missed detections and false alarms. Therefore, this paper employs a visual object detection algorithm for fire detection. Considering that many existing object detection algorithms have high computational complexity, which makes them unsuitable for operation on resource-constrained devices, this paper proposes a lightweight model, FERD-YOLO, based on YOLO11. A novel feature detection module, RLEF, is designed and combined with the lightweight backbone RevCol to reconstruct a new lightweight backbone, RLEFRevCol. Based on the original detection head, a DGE detection head is designed to further reduce computational complexity and parameters, and ECA attention is introduced before the backbone output and at each detection head layer to enhance the model's detection capability.