Skip to main content

Table 8 Text detection performance comparison with other techniques

From: Detection and recognition of cursive text from video frames

Image type Study Language Data size Recognition rate (%)
Document Ahmed et al. [146] Urdu 56 character clusters 93.40
  Akram et al. [147] Urdu 224 images 86.15
  Hussain et al. [148] Urdu 5249 ligatures 87.44
  Ahmed et al. [149] Urdu 15,251 lines 96.00
  Naz et al. [123] Urdu 10,000 lines 98.12
  Hassan et al. [150] Urdu 10,000 lines 94.85
Videos Zayene et al. [114] Arabic 7843 lines 96.85
  Tayyab et al. [134] Urdu 19,824 lines 93.02
  Hayat et al. [133] Urdu 290 uniqe ligatures 99.50
  Proposed Urdu 40,470 lines 87.00