Research on Object Detection Algorithm Based on Multilayer Information Fusion

Table 4

Experimental results on the PASCAL VOC2007 dataset.

Model	Fast RCNN (VGG16)	Faster RCNN (RPN)	SSD (300)	Proposal
aeroplane	64.15	67.25	69.94	87.13
bicycle	73.14	75.51	77.34	85.65
bird	65.38	68.96	69.83	72.95
boat	55.25	56.94	59.41	70.71
bottle	51.34	52.71	51.46	66.05
bus	75.93	76.73	78.92	86.10
car	76.55	78.65	79.59	89.50
cat	86.77	89.95	88.62	84.93
chair	49.22	50.21	52.82	59.12
cow	74.34	75.96	77.35	76.33
diningtable	63.55	65.15	67.38	68.32
dog	80.05	81.47	82.49	75.11
horse	79.07	81.21	82.83	86.20
motorbike	73.11	74.23	76.71	82.86
person	78.34	79.14	79.85	85.12
potted plant	42.76	44.76	44.06	43.41
sheep	67.14	66.78	65.33	31.60
sofa	63.02	66.32	67.94	70.26
train	75.32	76.83	78.79	81.99
tvmonitor	70.34	71.43	73.14	73.96
MAP	68.24	70.01	71.19	73.87