[HN Gopher] PaddleOCR: Lightweight, 80 Langauge OCR
       ___________________________________________________________________
        
       PaddleOCR: Lightweight, 80 Langauge OCR
        
       Author : zuhayeer
       Score  : 65 points
       Date   : 2021-09-09 19:14 UTC (3 hours ago)
        
 (HTM) web link (huggingface.co)
 (TXT) w3m dump (huggingface.co)
        
       | aliabd wrote:
       | Thanks for sharing! For context this is a demo of PaddleOCR V2
       | [0] which was released yesterday. You can find their original
       | repo here [1]. We built this demo using Gradio [2] and deployed
       | it on HuggingFace's Spaces [3].
       | 
       | [0]: https://arxiv.org/abs/2109.03144 [1]:
       | https://github.com/PaddlePaddle/PaddleOCR [2]:
       | https://gradio.app/ [3]: https://huggingface.co/spaces
        
       | jalopy wrote:
       | How does this compare to Tesseract?
        
         | aidenn0 wrote:
         | Worse for the corpus of english text I tried on it; it doesn't
         | seem to recognize punctuation at all, and it's marginally worse
         | at I/1/l on sans-serif text (which, to be fair, trips up humans
         | too).
         | 
         | Those were the only two relative deficiencies I noticed.
         | 
         | It does seem to beat tesseract on samples with mixed dark-on-
         | light and light-on-dark text, but that was the only big win I
         | saw in my brief look at it.
        
       | BugsJustFindMe wrote:
       | When I submit an image it just starts counting up until it
       | reaches 60.0/4.9s (whatever _that_ means) and then says ERROR.
       | -\\_(tsu)_ /-
       | 
       | Edit: I finally got it to work. The result looks good!
       | https://i.imgur.com/hoS4oMP.png
       | 
       | Though it looks like yet another OCR program that doesn't
       | understand archaic lexical paradigms like the long S or
       | ligatures.
        
         | aliabd wrote:
         | Sorry, this is because of the traffic right now. what you're
         | seeing is a counter for how long the current prediction time is
         | vs avg prediction time.
        
       | mdp2021 wrote:
       | Very nice.
       | 
       | But I see that (like tesseract) it cannot recognize different
       | styles (italics, bold, monospace...) - not only it seems to just
       | translate into pure (UTF) characters, it shows confusion on terms
       | in alternative styling.
        
       | danuker wrote:
       | I apologize for clicking "Flag" in case it doesn't flag errors.
        
         | aliabd wrote:
         | Yes the flag button is for wrong or unusual outputs. No worries
         | though
        
       | pacman2 wrote:
       | How well does it perform compared to Abbyy?
       | 
       | This packages the tesseract very nicely:
       | https://kebekus.gitlab.io/scantools/
       | 
       | I actually would buy Abbyy OCR but pricing for Linux ist just
       | insane for private use. I just saw, the CLI is even discontinued:
       | https://www.ocr4linux.com/
        
         | Pamar wrote:
         | ABBY is really great... but I was sorry to see they stopped
         | offering it as a subscription based webapp :(
        
           | pacman2 wrote:
           | That is really odd https://pdf.abbyy.com/finereader-online-
           | end-of-life/
           | 
           | What may drive this decisions?
        
       | vahid4m wrote:
       | I'm wondering if there is any image that results in accuracy of
       | 1? Or 0.999 is always the best.
        
       | magedqwani wrote:
       | does not recognize Arabic letter system so no use in arabic
       | .urdu.farsi etc..
        
       | antisthenes wrote:
       | Not sure what 80 languages are supported (couldn't find a list
       | anywhere), but I guess Russian wasn't one of them
       | 
       | Tried to OCR some russian text from an image and got absolute
       | nonsense.
        
       | ShakataGaNai wrote:
       | Demo is cool, but it tells us nothing about this particular OCR.
       | 
       | * Github: https://github.com/PaddlePaddle/PaddleOCR
       | 
       | * PyPi: https://pypi.org/project/paddleocr/
        
         | godelski wrote:
         | I've been using it is a current project. The OCR is pretty good
         | but I've been learning that OCR isn't as good as many of us
         | think it is. Specifically with handwritten text and more "in
         | the wild" type text. But website text? No problems.
        
       | aidenn0 wrote:
       | It seems to completely ignore punctuation for the corpus of
       | English text I tried on it; punctuation came through either not
       | at all (e.g. "Id" for "I'd") or as letters (e.g. "P" for "?").
        
         | timClicks wrote:
         | Welcome to OCR. It's often possibly to overlay the raw results
         | with a language model to improve them, but ultimately it's a
         | probabilistic process.
        
       ___________________________________________________________________
       (page generated 2021-09-09 23:00 UTC)