FD-YOLOv8:An Improved Fire Detection Model Base on YOLOv8n

Fire detection technology plays a crucial role in the field of fire safety. However, traditional fire alarms driven by smoke detectors often fail to report fires in a timely and effective manner. To cope with these challenges, We proposed an improved fire detection algorithm FD-YOLOv8 based on YOLOv8n. In this study, we introduced the OCA positive traffic attention mechanism to optimize the Bottleneck layer, enhancing the model's accuracy in object detection. Additionally, we integrated GSConv, an efficient lightweight convolution, to design an improved C2f-OGBlock module, further enhancing the model's detection capability while reducing parameters and computational costs. Furthermore, we incorporated a highly efficient feature enhancement algorithm, HIFA, into the model, significantly improving its feature extraction capabilities. We compared the improved YOLOv8 model with existing models. Experimental results demonstrate that our improved model achieves superior detection performance, with mAP@50 increasing by 3.2% and mAP@50:90 improving by 1.2% compared to the original model.