We are currently evaluating different options to integrate an OCR functionality in one of our product. Naturally, your OCR for .NET was one of the first API to be evaluated since it is already included in our license. We also evaluate several other OCRs engine from other companies. We are wandering if there was any setting or image processing that should be done prior to using the SDK as our results with the API were not very convincing in comparison to the other APIs?
Regards
You can perform image correcting before performing OCR operation on the image. Please visit the link Skew Image Correction Using Algorithm for details. Further, you may forward us your sample input file to perform OCR operation. We will evaluate it and update you about our findings.
Just in case it does not reach you, I replied to the reply notification email and attach the images I used.
No reply received. Please send us the sample images again as an attachment in reply to this post.
sample1.jpg (82.9 KB)
sample4.jpg (118.6 KB)
OCR-A_sample.svg.png (25.7 KB)
Here they are, thank you!
I forgot, in my email, I included my results for each one of those images if you want to compare with yours:
Here are my results if you want to compare them to yours.
Sample1.jpg:
][HIHGs voG sHoGLD KltDINABoD .- GooGLE s/AvE ’
ssennhs ][nmnra, n gradente sradent eed ][A. cs teschIng ene
secden eI2d sradents frnm Dn BnsmI’s 2Dd-stcdentIentre
ceerae. Intcodetccec te pspcheIegp Befate the semestet hegcns.
][smera and the nIcc ethet ][As meet wcth Dt. BnstcI. He ess-
pIaIns thac asscgnments MII he dene hp stcdect teams es- ccg an
appIccaLen raIIed INase fat the ceeraeweti INase. he saps. cs
IIke e-macI. setc eI, acd IIke a chat snd ceIIaheraden space. sert
eI, het that It Is dIferecc frnm tceIs they hasc esed heIete.
stcdects MII cse INase te ceIIaherare. shsre reseertes acd dats.
raIk thraegh theIt hndcngs, acd dcsceset Mps te appIy what thep
Ieara te ethet pteIects. INase cae he esed spnc1mcecsIp and as
a pIam fat creratcse ceIIahe- raden. In addcenn, ][As MII ese
INsse tc cemmenIratc Mth theIt stcdent rearas, checMng cn Iet
qeestcecs snd rannctct- Icg ptegress Dchcg the hrat Iectcte, Dt
BhsteI assIgns an esertcse that InseIses nhseMng cnteracdnn
hetweec peepIe cn pehIcc pIsces: cefee shnps. maIIs. Ithranes, et
testseracts. fkch reara cs te cseIate dse hehaniera mmmen cc
Sample4.jpg:
AD1I\ERTIsEMENT.
llnle pulllicAtilau laf tbe Wlarli laf Jlanasir Kasirlani it iEI
sllp- Fmrill exteud 10 liit-e Vlalnlnee. It Fmree tblaugllt
EIdtisllle 10 coralaEIuee tbe colilaEI witll llin Hifstll laf
1{hla lleflaralatilau iu Slalatland, ee tltla Fmrlark laf glabe
itttporttanco. llllp next vlalnlne will tllue coutacin tbe
nird tluad lilaurbll Book1s Fmrhilall lalautieue tlle HiEI10ll 10
the J-ealr 1564 i lat villilall pelic)W[I lliEI ll10rilaal lilllllaurs taay be conlsdered 10 tenninate. Btlb tlle liifth lllalakc flarmllig a eequlal 10 tlle Hifstllc and pullli19ed tluader llill rlllae iu 1644, will eall lle inlalndlad. HiEI Llattera Imd lr[isllanelaue Wlitingla will llla arrauglad in tllla sll-uent blalurlym ee nearly ee plalsillle in cltrlaulalla-
gical lardlar : eaclt porbilaa lllaing intreduced lly a EIeparate
elatire. respocting tllla llaatlueclipt lar prill]ted enpieEI frlalla
WIIicll tlley ltave llcon taklaa.
lt rrlaly Flerllapa llc exFlected tllat a Lifla laf tbe tlutblar
-hlalald llave belan pre6xlad 10 tlliEI vlalullac. nla Life laf
Finlani llJc Dje. lr[‘Cnlg, ia ltlaFmrcvcr a Fmrlark EIla tluaivrrsallJ’
linFmrlb, aad laf sla llaucll lligtlalicAl valucc ee 10 lxperaladc
, luly attellapt that llaigltt lce llaallla flat` a cllataillad lli-
OCR-A_sample.svg.png:
T h a a u i c k , B r o i` a
abcdefghijklmnopqrtuvixyzobeakShVB5 E a ( ) { } I N < >
Thank you for your help,
Stephane
Thank you for sharing the samples. We have evaluated the samples. We are able to reproduce the issue. The issue has been logged into our system with ID OCRNET-3401 for further investigation by our Product team. We’ll update you here once there is some information or a fix version available in this regard.