Multilingual Text Detection with Nonlinear Neural Network

Table 2

Performance comparison on different benchmarks.

Method	Dataset	Precision	Recall	-measure
Pan [15]	ICDAR 2003	0.645	0.659	0.652
Zhou [16]	ICDAR 2003	0.37	0.79	0.53
Our method	ICDAR 2003	0.45	0.80	0.57
Lee [17]	KAIST	0.69	0.60	0.64
Our method	KAIST	0.59	0.79	0.67