ara-Amiri traineddata at char train=1.127%

Shreeshrii · Feb 25, 2019 · 749c723 · 749c723
1 parent ba03b08
commit 749c723
Show file tree

Hide file tree

Showing 3 changed files with 30 additions and 4 deletions.
diff --git a/Arabic-TOC-ara-Amiri.txt b/Arabic-TOC-ara-Amiri.txt
@@ -0,0 +1,26 @@
+الجفا ...... ............................................. ‎٨٧٢‏
+‏غرام مشُميت ............................................... ‎٨٧٣‏
+‏الفؤاد الكسير .............................................. ‎٨٧٤‏
+‏عَقيقَ في عقيق في عقيقي ....................................... ‎٨٧٥‏
+‏الباب الحادي عشر: مُتفرزقات .................................... ‎٨٧٧‏
+‏الشاي ................................................... ‎٨٧٩‏
+‏مَديح الشاي ............................................... ‎٨٨٠٩‏
+‏مسك الشاي ............................................... ‎٨٨١‏
+‏ليلة الشاي ................................................ ‎٨٨٢‏
+‏رجال السر ...... ......................................... ‎٨٨٣‏
+‏في فضل الاجتماع ...... ..................................... ‎٨٨٤‏
+‏شججة ................................................... ‎٨٨٥‏
+‏لله دَبني رَوَاحة ............................................ ‎٨٨٦‏
+‏خطة عَبِْيّة ..... .......................................... ‎٨٨٧‏
+‏مزايا الزمان ............................................... ‎٨٨٨‏
+‏عَشرَاء ................................................... ‎٨٨٩‏
+‏قطع علاقة في عتاب ......................................... ‎٨٩١‏
+‏مُعاتبة ...... ............................................. ‎٨٩٢‏
+‏السمكة .................................................. ‎٨٩٤‏
+‏نظرة ...... .............................................. ‎٨٩٥‏
+‏القطار ................................................... ‎٨٩٦‏
+‏المعالي .................................................. ‎٨٩٨‏
+‏المصادر والمراجع ............................................ ‎٩٠٠‏
+‏الفهرس ................................................... ‎٩٠٢‏
+‎٩١٢‏
+
diff --git a/ara-Amiri.traineddata b/ara-Amiri.traineddata
diff --git a/build/tesstrain_amiri.sh b/build/tesstrain_amiri.sh
@@ -6,8 +6,8 @@
 ################################################################
 # variables to set tasks performed
 ################################################################
-MakeEval=yes
-MakePlusMinus=yes
+MakeEval=NO
+MakePlusMinus=NO
 RunPlusTraining=yes
 RunEval=yes
 ################################################################l
@@ -128,7 +128,7 @@ fi
 
 if [ $RunPlusTraining = "yes" ]; then
 
-for ((PlusMinusIterations=4300; PlusMinusIterations<=7000; PlusMinusIterations+=1000)); do
+for ((PlusMinusIterations=6300; PlusMinusIterations<=7300; PlusMinusIterations+=500)); do
 
     echo "#### Plus-Minus training Using Amiri text #####"
     lstmtraining \
@@ -173,7 +173,7 @@ tesseract $eval_output_dir/ara.Amiri.exp0.tif ara.Amiri.exp0-$ModelName-eval --t
 
 wdiff --no-common --statistics ./ara.Amiri.exp0-$ModelName-eval.txt $langdata_dir/$Lang/$Lang.evalnew.training_text 
 
-tesseract /home/ubuntu/tessdata_arabic/Arabic-TOC.tif /home/ubuntu/tessdata_arabic/Arabic-TOC-ara-Amiri  --tessdata-dir ./ara-Amiri-from-Arabic  --oem 1 --psm 6 -l ara-Amiri
+tesseract /home/ubuntu/tessdata_arabic/Arabic-TOC.png /home/ubuntu/tessdata_arabic/Arabic-TOC-ara-Amiri  --tessdata-dir ./ara-Amiri-from-Arabic  --oem 1 --psm 6 -l ara-Amiri
 
 tesseract /home/ubuntu/tessdata_arabic/Arabic-TOC-numbers.png /home/ubuntu/tessdata_arabic/Arabic-TOC-numbers-ara-Amiri  --tessdata-dir ./ara-Amiri-from-Arabic   --oem 1 --psm 6 -l ara-Amiri