The Internet of Things has found extensive applications in the medical field, where the utilization of the You Only Look Once version 5 (YOLOv5) object detection network has played a crucial role. One notable application of the Internet of Things in healthcare is the detection of Femoral Neck Fractures (FNF) in pathology. Because of its subtle fissures, it is difficult to observe and is considerably hindered by treatment. Therefore, this paper proposes a new network model CSFF-YOLOv5 based on YOLOv5s network for the detection task of FNF. The network integrates the attention mechanism and channel split idea to make the network pay more attention to the parts of interest, better learn the feature information of fracture, and ensure the integrity of information through feature fusion to improve the detection accuracy of the network. Through a large number of ablation experiments to prove the performance of different modules, the experimental results show that compared with YOLOv5s, the detection effect of the method proposed in this paper is significantly improved, and there is a certain increase in precision, recall, F1, mAP@0.5 and mAP@0.5:0.95, especially in the mAP@50:95 index, which is increased by 4.985%, which is enough to show that CSFF-YOLOv5 has a good detection effect on femoral neck data.