Baseline based model for Grantha script for Kraken.
Example Usage
for i in tam.Noto_Sans_Grantha.0001000.exp0.png ; do
kraken -i "$i" "${i%.*}"-grantha_best.txt binarize ocr -s -m grantha_best.mlmodel ;
done
INPUT
OUTPUT
𑌮𑍁𑌰𑍈𑌂𑌪𑌂𑌟 𑌵𑌥 𑌪𑌬 𑌦𑍍𑌰𑌾𑌵𑌿𑌡𑍍𑌯𑌾𑌂 𑌪𑌚𑌿
TRAINED ALPHABET
[523.8546] Training set 3123 lines, validation set 348 lines, alphabet 108 symbols
[523.8548] alphabet mismatch: chars in training set only: {'_', '5', '6', '0', '3', '7', '9', '1', '4', '*', '2'} (not included in accuracy test during training)
[523.8550] grapheme count
[523.8551] SPACE 27739
[523.8552] GRANTHA SIGN VIRAMA 19701
[523.8553] GRANTHA VOWEL SIGN AA 9398
[523.8554] 𑌤 7881
[523.8554] 𑌰 7489
[523.8555] GRANTHA VOWEL SIGN I 6179
[523.8556] 𑌮 5876
[523.8556] 𑌵 5837
[523.8557] 𑌨 5706
[523.8557] 𑌯 5587
[523.8558] 𑌸 4580
[523.8559] GRANTHA VOWEL SIGN EE 4017
[523.8559] DEVANAGARI STRESS SIGN ANUDATTA 4016
[523.8560] 𑌕 3607
[523.8561] 𑌪 3497
[523.8561] GRANTHA VOWEL SIGN U 3486
[523.8562] 𑌦 3362
[523.8563] VEDIC TONE CANDRA ABOVE 2669
[523.8563] GRANTHA SIGN ANUSVARA 2618
[523.8564] । 2524
[523.8564] 𑌶 2514
[523.8565] GRANTHA VOWEL SIGN OO 2467
[523.8566] GRANTHA SIGN VISARGA 2422
[523.8566] 𑌷 2274
[523.8567] 𑌚 2210
[523.8568] 𑌹 2022
[523.8568] 𑌣 1944
[523.8569] GRANTHA VOWEL SIGN II 1840
[523.8569] 𑌭 1817
[523.8570] 𑌜 1674
[523.8571] 𑌗 1619
[523.8571] 𑌲 1591
[523.8572] 𑌧 1540
[523.8573] ॥ 1419
[523.8573] 𑌅 1344
[523.8574] - 1227
[523.8575] GRANTHA VOWEL SIGN VOCALIC R 1226
[523.8575] 𑌬 1086
[523.8576] GRANTHA VOWEL SIGN UU 1054
[523.8576] . 1035
[523.8577] 𑌥 939
[523.8578] GRANTHA VOWEL SIGN AI 775
[523.8578] 𑌟 741
[523.8579] ௧ 694
[523.8580] 𑌞 621
[523.8580] , 575
[523.8581] 𑌙 510
[523.8582] ) 510
[523.8582] 𑌇 501
[523.8583] ௦ 465
[523.8583] ( 461
[523.8584] 𑌡 458
[523.8585] ௨ 457
[523.8585] 𑌆 453
[523.8586] 𑌖 446
[523.8587] GRANTHA AU LENGTH MARK 403
[523.8587] 𑌠 383
[523.8588] 𑌛 375
[523.8588] ௩ 374
[523.8589] 𑌽 361
[523.8590] 𑍐 352
[523.8590] 𑌉 351
[523.8591] 𑌏 325
[523.8591] ௪ 315
[523.8592] ௫ 307
[523.8593] DEVANAGARI STRESS SIGN UDATTA 274
[523.8593] 𑌘 270
[523.8594] GRANTHA SIGN NUKTA 268
[523.8595] ௮ 265
[523.8595] 𑌫 264
[523.8596] ௭ 262
[523.8596] / 258
[523.8597] ௬ 249
[523.8598] ௯ 248
[523.8598] : 232
[523.8599] = 220
[523.8600] ; 217
[523.8600] 𑌢 202
[523.8601] ' 189
[523.8601] 𑌳 183
[523.8602] GRANTHA SIGN COMBINING ANUSVARA ABOVE 139
[523.8603] 𑌝 89
[523.8603] 𑌓 85
[523.8604] 𑌋 82
[523.8605] 𑌈 75
[523.8605] 𑌊 66
[523.8606] 𑍡 63
[523.8606] GRANTHA VOWEL SIGN AU 59
[523.8607] GRANTHA SIGN CANDRABINDU 49
[523.8608] 𑍠 49
[523.8608] 𑍝 36
[523.8609] GRANTHA VOWEL SIGN VOCALIC RR 34
[523.8610] ^ 32
[523.8610] GRANTHA VOWEL SIGN VOCALIC L 31
[523.8611] 1 23
[523.8611] * 22
[523.8612] 2 16
[523.8613] _ 11
[523.8613] 𑌐 10
[523.8614] 𑌔 7
[523.8615] 3 5
[523.8615] 4 4
[523.8616] 6 4
[523.8616] 5 4
[523.8617] 7 4
[523.8618] 9 3
[523.8618] 8 3
[523.8619] 0 3
BEST MODEL
[47906.5020] Moving best model grantha_9.mlmodel (0.9393690912598881) to grantha_best.mlmodel