Large-Scale Video Retrieval via Deep Local Convolutional Features

Table 5

Comparison with existing visual search approaches on the NTU.

Method	Acc	Time (s)	Dimension
Ours	0.9691	1.67	64, 512
DFSP	0.9516	1.613	64
DF [16]	0.9198	1.892	25088
DFSE [27]	0.8441	1.606	64
DFTA [28]	0.8254	1.739	512
LBTT [28]	0.9096	1.693	256