Row 1: it has removed ","after 5 = 365,9
Row 2: it replace comma “,” to punctuation = 253,82
Row 3: it has removed comma and have added empty space = 6295,9
Row 4: it couldt read comma = 2764,32
And if the scrap value is 0 the its not reading that 0.
Now here is the result, it has read correctly …
Help me… that how can i be sure that it has read the correct value… and i need the correct value.
I have tried with Google Ocr also but it’s worse result, also tried wih low or high scale. Same with Microsoft.
So just to get one thing straight, per my experience, the most accurate OCR is ABBYY Flexy capture, and that one needs to be payed. It is payed per request and even so… it is 99% accurate, and that means… your comma can be lost, 1 can be 7 or vice versa, etc etc
The best way, if you need exact value is to try to get it with get text and selector, if it is possible.
Now, to do this with OCR…
Image that you provided was pretty ok, so OCR Google read it without problems
So I have used Tessaract, that one is Google but I dont think its cloud, Google Cloud is something else, can you test it with tessaract, and in properties make it to be Legacy.
I did it…
here is result of Tessaract
of the same ID…
Tessaract is far better but problem is, some time it’s not reading right number…
Like reading 1 as 4. and if its 4 then its reading as 1. and writing that value as output. and some time it’s writing 3099,S6 where s is a 5…
So when i used Tessaract I felt its perfect as its putting comma right place and reading comma as no space… but when i analyzed it with the real data then i felt it’s not prefect…
So if i use Tessaract then i can not be sure that value 4541,61 is correct or its 1541,61… and oppesit also,
Per my knowledge, not. Even with ABBYY there is chance of reading false values.
But lets take step back, from where you are reading those values, is that picture, pdf, web page? what is the selector, I can see that is table, maybe there is another way of accessing there not to have OCR.
