Can Machine Automatically Discover Text Image from Overall Perspective

Jiang Wei,Wu Jiayi,Yao Chao

Table 2 Evaluation in ICDAR2003 dataset

Methods Text (%) Error (%) Proposed Method 82.21 17.79 Bai et al.[1] 89.2 10.8 Shivakumara et al.[12] 80.97 19.03 Shivakumara et al.[5] 81.12 18.88