Table 1 A summary of notable contributions on recognition of Urdu text
From: Segmentation-free optical character recognition for printed Urdu text
Study | Database | Classifier | Results | Recognition unit | Approach |
|---|---|---|---|---|---|
Pal and Sarkar [6] | Custom | – | 97.8% | Isolated characters | Analytical |
Shamsher et al. [7] | Custom | Neural networks | 98.3% | Isolated characters | Analytical |
Tariq et al. [8] | Custom | Neural networks | 97.43% | Isolated characters | Analytical |
Sardar and Wahab [9] | Custom | – | 97.12% | Isolated characters | Analytical |
Nawaz et al. [12] | Custom | – | 89% | Isolated characters | Analytical |
Ahmed et al. [11] | Custom | Neural networks | 93.4% | Segmented characters | Analytical |
Hussain et al. [21] | CLE Urdu | HMM | 87.76% | 250 graphemes | Analytical |
Hassan et al. [16] | UPTI | BLSTM | 86.4%/95.8% | Characters | Analytical |
Ahmed et al. [22] | UPTI | BLSTM | 89% | Characters | Analytical |
Naz et al. [18] | UPTI | MDLSTM | 96.40% | Characters | Analytical |
Hussain et al. [32] | Custom | – | 95% | Spotting ligatures | Holistic |
Sabbour and Shafait [5] | UPTI | KNN | 91% | 10,000 primary ligatures | Holistic |
Javed and Hussain [10] | CLE Urdu | HMM | 92% | 1282 unique primary ligatures | Holistic |
Akram et al. [13] | CLE Urdu | Modified tesseract | 97.87% | 1475 unique primary ligatures | Holistic |
Akram et al. [28] | CLE Urdu | Modified tesseract | 86.15% | Unique ligatures | Holistic |
Javed et al. [19] | CLE Urdu | HMM | 92.73% | 1692 Unique ligatures | Holistic |
Khattak et al. [29] | CLE Urdu | HMM | 97.93% | 2028 Unique ligatures | Holistic |