Research on Detection Method of Fuel Tank Cover Based on Lightweight YOLO

Automobile fuel tank cover recognition is one of the key technologies of intelligent unmanned fueling machine station system. A lightweight YOLOv8-based fuel cap detection algorithm is proposed to address the problem of recognizing and locating car fuel caps in unstructured environments. Considering the accuracy and real-time performance of the refueling task, the feature fusion network is firstly improved to two-scale feature detection, and then the detection accuracy and recognition speed are improved by combining with a weighted bidirectional feature fusion network (BiFPN); in order to further optimize the efficiency of the model’s feature extraction and to obtain feature information with stronger correlation, the C2f-Faster module is introduced to optimize the feature extraction module. The effectiveness of the lightweight improvement of the YOLOv8 model in fuel cap detection is verified through ablation experiments. The results show that the number of parameters of the method is reduced by 56.9% compared with the original YOLOv8 model, and the model complexity GFLOPs is reduced by 37%, while the detection speed and the detection accuracy are also improved, which demonstrates a good performance on the fuel cap detection task.