Observations.txt

Missing classes:
S (pluta) (nn_3_cropped.bmp)
|| (nn_3_cropped.bmp)

kPart (d_9_cropped.bmp) : kPart + aaDheergha = kaa
_nPart (d_9_cropped.bmp) : _nPart + aaDheergha = _naa
vPart (d_9_cropped.bmp) : vPart + aaDheergha = vaa (currently "vPart" is recognized as "e")
_dPart (d_9_cropped.bmp) : _dPart + aaDheergha = _daa (currently "_dPart" is recognized as "yePart")
LPart (d_9_cropped.bmp) : LPart + aaDheergha = Laa (currently "LPart" is recognized as "naalku")
_tPart (d_9_cropped.bmp) : _tPart + aaDheergha = _taa (currently "_tPart" is recognized as "_taa")

31 more classes:
kPart, k1Part, gPart, g1Part
chPart, ch1Part, jPart
TPart, T1Part, DPart, D1Part, NPart(?)
_tPart, _t1Part, _dPart, _d1Part, _nPart
pPart(?), p1Part(?), bPart, b1Part
rPart, _lPart, vPart, _shPart, ShPart(?), sPart(?), hPart, LPart (==naalku), rOldPart, LOldPart

Confusion Pairs:
(anuswara, sonne, zero) : Taken care at character-class level itself
ombattu and arkaaOttu : Resolved in OCR_Word::verify()
entu and _la : Resolved in OCR_Word::verify()
eeLu/seven and question : Resolved in OCR_Word::verify()

che -> Te (d_9_cropped.bmp)
he -> Ne (d_9_cropped.bmp)
La -> pipe (d_9_cropped.bmp) (AspectRatio should help here)
mOttu -> TrOttu (d_9_cropped.bmp: line-9), paadam (d_9_cropped.bmp: line-11, 13, 15), T1Ottu (d_9_cropped.bmp: line-13)
_ni -> si (d_9_cropped.bmp: line-10)
_duu -> _do (d_9_cropped.bmp: line-12)

dot -> comma (d_9_cropped.bmp)
dot -> pipe (d_9_cropped.bmp: line-6)
comma -> paadam (d_9_cropped.bmp: line-7)