Since their simple structure, broad range of applications, and ease of transport, ton bags are widely used by many enterprises as a packaging solution for cargo transportation. However, the handling of ton bags mainly relies on manually operated cranes, which suffer from low efficiency and compromised safety. To address these challenges, the YOLOv8 model in the YOLO series was selected for the ton bag detection experiments, and the lightweight YOLOv8n model in YOLOv8 was comprehensively considered to be used for the subsequent experiments. The EIoU loss function was introduced to improve the loss calculation of YOLOv8, thereby improving its accuracy and speed in the ton bag detection tasks. Comparison experiments between the improved YOLOv8n-E model and the original YOLOv8n model were conducted. The results demonstrate that the YOLOv8n-E model, utilizing the EIoU loss function, increases precision by 1%, improves mean Average Precision (mAP) by 0.5%, and achieves a recall rate of 0.98 across the ton bag dataset.