A Block Object Detection Method Based on Feature Fusion Networks for Autonomous Vehicles

Table 4

Comparison of object detection effects on different models.

Detection model	Data set	Number of categories	Total number	Number of	Number of test sets	Detection accuracy of each category							Total accuracy	Detection rate
			of training samples	verification samples		Person	Cyclist	Motocycle	Car	MiniBus	Bus	Truck
SSD	VOC-2007	20	9963	5011	4952	0.5145	0.5379	0.5701	0.5883	0.5534	0.5499	0.5407	0.5507	28Fps
	KITTI	5	7982	7183	799	0.6145	0.6379	0.6701	0.6883	0.6534	0.6499	0.6407	0.6507	24Fps
	SSM-CAR	7	11550	10394	1156	0.7903	0.8057	0.8109	0.8044	0.8170	0.7968	0.8102	0.8050	27Fps
Our SSD	VOC-2007	20	9963	5011	4952	0.7741	0.7856	0.7812	0.7821	0.7830	0.7904	0.7923	0.7841	31Fps
	KITTI	5	7982	7183	799	0.8145	0.8279	0.8001	0.8083	0.8234	0.8499	0.8407	0.8107	29Fps
	SSM-CAR	7	11550	10394	1156	0.9805	0.8972	0.9196	0.9585	0.9043	0.8807	0.8984	0.9085	33Fps