Well, first of all, since this is an OpenCV question, I'd suggest you use an OpenCV mailing list or forum to ask it.
You don't mention re-training the model. If you try and recognize a completely different style of digit using a model trained on the images in the example, I wouldn't expect it to work at all. Tom On Monday, September 14, 2015 at 4:24:07 AM UTC-4, Srinivasa TN wrote: > > I used an image in the original link (attached for reference - > 2015014132000.png) and it is recognising correctly: > > 131601 ------- > > I used the image posted by OP (attached for reference - 2015014132100.png) > and it is not recognizing the 3 in it: > > 4578 ------- > > I used another image of digital display (attached for reference - > 2015014133400.png) and whole image is recognized as single digit 0: > > 0 ------- > > Any suggestions on how to make it recognize digital displays? > > Regards, > Seenu. > > On Thursday, 10 September 2015 22:19:08 UTC+5:30, Tom Morris wrote: >> >> On Thursday, September 10, 2015 at 7:18:05 AM UTC-4, Srinivasa TN wrote: >>> >>> Can I know where can I get the images required for training emeocv? >>> (using *./emeocv -i ../../image/ -l -v DEBUG* but I am trying to find >>> out what images I should have in image directory) >>> >> >> Sorry, I'm not the author of the article. I just found it and posted a >> pointer. >> >> Having said that, I would have thought you'd want to use representative >> images of the style of meters that you'll be targeting, rather than the >> examples that were used in the original experiments. >> >> Tom >> >> On Saturday, 4 July 2015 22:55:25 UTC+5:30, Tom Morris wrote: >>>> >>>> I'd suggest looking at OpenCV. It looks more like a computer vision >>>> task than an OCR task. Some of the specific issues like dials not fully >>>> aligned in the window are things the OCR systems aren't designed to deal >>>> with, but you could use domain knowledge to deal with in a system like >>>> OpenCV. >>>> >>>> Here's a description of implementing a meter reader in OpenCV: >>>> http://www.thekompf.com/cplus/emeocv.html >>>> >>>> Tom >>>> >>> -- You received this message because you are subscribed to the Google Groups "tesseract-ocr" group. To unsubscribe from this group and stop receiving emails from it, send an email to [email protected]. To post to this group, send email to [email protected]. Visit this group at http://groups.google.com/group/tesseract-ocr. To view this discussion on the web visit https://groups.google.com/d/msgid/tesseract-ocr/de0009e1-36d9-4412-992e-a4f49e96b4d5%40googlegroups.com. For more options, visit https://groups.google.com/d/optout.
