Efficient-YOLO: A Research on Lightweight Safety Equipment Detection Based on Improved Yolov8

In the construction industry, safety protective equipment is crucial in protecting workers' lives and property. However, due to the particularity of its application scenarios, the real-time performance of the algorithm has a high demand. To solve this problem, we improved on YOLOv8 and designed a lightweight model, Efficient-YOLO. First of all, we use EfiicientN etv2 to replace the backbone network of the original YOLOv8, which greatly reduces the number of parameters and calculation of the model while taking into account the accuracy of the model. Secondly, due to the large proportion of small targets in security protective equipment, we introduce a smaller target detection layer(SMDL) into the original network structure, which strengthens the algorithm's fusion of shallow features and improves the accuracy of the model. Finally, experiments show that the Efficient-YOLO model achieves 86.8% mAP and 555.55 FPS on the CHV dataset. Compared to the original YOLOv8, the number of parameters decreased by 26.1 %, the calculation amount decreased by 49.4%, and the FPS increased by 1044.53%.