AEID-YOLO: A Safety Equipment Detection Model Based on Multi-Scale Feature Fusion and Dynamic Focusing Mechanism

To address the problem of low detection accuracy of safety equipment in actual construction site scenarios, caused by factors such as small object sizes and occlusion, this paper proposes AEID-YOLO, a safety equipment detection model based on YOLOv8. In the AEID-YOLO model, we introduce the efficient multi-scale feature fusion AEMC module, which enhances the multi-scale feature extraction capability of the C2f structure. The ELA attention mechanism is employed to strengthen the network's ability to capture global contextual information. We also design an auxiliary bounding box regression loss function, Inner-Wise-MPDIoU, with a dynamic focusing mechanism to address the inaccuracy of CIoU in regression results. Additionally, we adopt the DyHead dynamic detection head, which integrates scale-aware, spatial-aware, and task-aware mechanisms to improve long-range modeling capabilities and further enhance detection performance. Experimental results demonstrate that our detection model improves mAP@0.5 by 2.1% and mAP@0.5:0.95 by 1.9% on the SHD dataset, and mAP@0.5 by 1.3% and mAP@0.5:0.95 by 1.9% on the SHWD dataset. The proposed detection model outperforms other mainstream algorithms, enabling accurate detection of safety equipment in various complex construction site scenarios.