Table 6.

Results of comparison for inference time in different environments. All models were inferred by converting them to ONNX form, and the different computing environments are as follows: “@GPU” refers to the Nvidia Tesla T4, “@Nano” refers to the Nvidia Jetson Nano, and “@CPU” refers to the Intel Xeon processor.

Model	Inference Time (ms) ONNX @Nano	ONNX @GPU	ONNX @CPU
YOLOv8n	12.04	133.30	353.13
YOLOv8s	12.16	217.20	891.93
YOLOv8m	20.20	471.59	1792.31
YOLOv8l	29.09	733.27	3340.26
YOLOv8x	41.87	1208.69	4222.87