Research on lightweight gas station target detection algorithm based on improved Yolov8n

In recent times, as society progresses, urban safety has become a subject of extensive concern. Gas stations, crucial for oil and gas supply within a city, can cause incalculable losses and damage to cities and enterprises in case of a safety mishap. To address the issues of sluggish detection speed and large model size in target detection algorithms for gas station settings, this article presents a lightweight target detection algorithm founded on the enhanced Yolov8n. The algorithm takes the Yolov8n model as its fundamental network, substitutes the traditional convolution module with SPDConv spatial depth conversion convolution, incorporates the concept of arbitrary kernel convolution (AKConv) into the C2f module, and replaces the CIOU loss function with the SIOU loss function, ultimately accomplishing target detection in gas station scenarios. Experimental outcomes demonstrate that the overall mAP of the refined algorithm has increased by 0.1% compared to the original one, while the model size has been diminished by approximately 14.3%, rendering it more proficiently applicable for target detection at gas stations.