G-YOLOX: A Lightweight Network for Detecting Vehicle Types

Table 3

Comparison of models with the combined VOC2007+2012 dataset.

Model	Parameters (M)	GFLOPs	Weight file (MB)	mAP50	mAP50_95
YOLOX_s	8.95	26.68	70.207	0.8053	0.5964
Ghost+YOLOX_s	4.52↓49.5%	13.63↓48.9%	46.387↓34%	0.8029↓0.0024	0.5902↓0.0062
G-YOLOX (ours)	2.95↓67.1%	7.00↓73.8%	41.878↓40.4%	0.7764↓0.0289	0.5437↓0.0427