Richtig enttäuschend war der Einstieg in Handschrift-#OCR mit #tesseract. Deutlich besser war #llama-vision mit #ollama auf dem Alpaca-Flatpak. Das Setup hat es allerdings nicht geschafft, meine CPU im Notebook richtig zu nutzen und war entsprechend langsam. Ein #LLM mit Bildverarbeitung scheint mir auch Overkill zu sein.
A pauta de hoje do #TerSoftware é sobre "gestão de papel". Recentemente, testei OCR para digitalização de tabelas e... não fiquei muito feliz com o resultado.
Acredito que #OCR funcione melhor quando fica bem amarrado com o documento digitalizado (por exemplo, tornando um arquivo PDF buscável), mas para extração de texto, ainda é um grande "depende".
Na minha curta jornada, testei #Tesseract e #Docling. Talvez funcione com código bem escrito, mas acabei me rendendo e indo "no muque" mesmo.
O Tesseract parece bem fácil de instalar no Linux (mesmo no #openSUSE Leap, que tem suas limitações por sair do SUSE empresarial, achei fácil), mas o Docling exigiu alguns malabarismos com ambientes em Python (usando conda e pip).
Para texto corrido, o Tesseract parece bem suficiente, já. Pode ser rodado via linha de comando e, pelo menos no openSUSE Leap, vários dicionários se encontram empacotados para facilitar.
Tonight I had the opportunity to attend the live performance of #Tesseract at the #RessurectionFest. I have known their music for years, but was in the least preapared for this incredible show. True bearers of the #ProgressiveRock ethos.
This concert felt like being in the driver's seat of a fast, loud race car. Scary, but enthralling. No clue where it is exactly taking you, but for sure into interesting places.
Then I was asked for #berkeley, and this was the first real inflection point for the project.
Berkeley didn't use Legistar. It didn't use any system I had seen before, and it had minutes going back to 1905.
This was going to be prohibitively expensive using AWS' OCR tool, so I had to get creative.
This was where I started exploring #tesseract, and building a pipeline for this project that could run entirely on one machine.
3/n
@inkorrupt Ich habe munkeln hören, dass gerade ein paar Leute daran sitzen, aber der Vorgang zieht sich ganz schön hin. Ist ja auch ein ordentlicher Batzen schlechter Scans. #Tesseract muss ganz schön durchatmen. Wenn das Ergebnis brauchbar ist, werden vermutlich noch eine Menge Fehler drin sein. Aber das wäre immer noch besser als gar keine Suchmöglichkeit.
This is the syntaxis of #Tesseract-OCR
tesseract IMAGEFILE TEXTFILE [OPTIONS]
It is that simple.
The resulting text file may not be perfect, and you may have to tweak it somewhat.
See the man page for more options.
Hat schon mal jemand mit #imagemagick ein Bild von einem digitalen Display für #tesseract OCR aufbereitet? Wie sähe eine näherungsweise sauber arbeitende Zeile für #convert hinsichtlich Schärfung, Kontrastverbesserung und Graustufenkonversion aus? Retoot gerne gesehen.
Every now & then, I give #ChatGPT a scan of my handwriting to test its skills in working with #handwrittentexts. Initially, it responded that it could not process the scans or gave me entirely fictional output, but today it got almost everything right. These results are better than those I achieved with #HWR models in #Tesseract & #OCR4all without additional training. I also asked ChatGPT what it "thought" about my writing & it called it "consistently shaped & large with stylistic strokes."
Looks like #TheDearHunter is coming back to Europe this year at #BeProgMyFriend in Spain and #Euroblast in Germany, both in September. Alongside #TesseracT!
I think "A Haven with Two Faces" from the new Spiritbox album "Tsunami Sea" is just awesome!
I've mentioned it before, but I really like the drift into atmospheric, progressive music and it reminds me of TesseracT, one of my favourite bands 
#OpenSource Programm I need.
1.some sort of an apple tags like variant for the open source world ( best is file manager from #elementaryos at this point but it only support tagging 8 colours no #
(Nice to look at automation like the Mac #hazel or Mac #defaultfolderx)
2.and #preview replacement ( pdf and other files reader with most of the pro features and some sort of working #ocr ( possibly a gui of #tesseract ? ) for #Linux and #android preferably (best I found was #pdfsambasic)
Hi #histodons,
I need your expertise. We want to integrate an #opensource #ocr tool into our #useGalaxy Platform so you can better analyse your texts, etc.
I worked with #tesseract some years ago, and I heard about #ocr4all.
Do you have experience with any of these - or other recommendations?
We are also integrating #tranksribus via API but want another ocr-specific option.
Looking forward to your experiences!
#TesseracT, #Leprous, #GreenLung and #Sungazer among others announced for this year's #ArcTanGent. Tempting 
@lavaeolus have worked with #tesseract but not via a GUI - Thanks for this.
My Album Of The Year is: "Lingua Ignota Pt. 1" by Persefone.
I chose the album as my AOTY because I was eagerly awaiting the release and not a single song on the EP disappoints! Every single one picks me up and combines the familiar Persefone sound without being stingy with refreshing new elements. I attended their live concert shortly after the releases and couldn't be happier! Great band, awesome show, nice crowd! Can't wait for Pt. 2!
There were 3 more candidates for my AOTY. In order of preference:
TesseracT - War of Being
DVNE - Voidkind
VOLA - Friend of a Phantom
Que programas #OCR #OpenSource recomendais para #Linux?
Já experimentei vários baseados em #tesseract e descobri recentemente o #Rescribe que faz um trabalho razoável.
Hey, #Android friends.
Can anyone recommend a recently updated Text Recognition / OCR app which is open source?
All the ones on F-Droid are outdated. I need something which works offline rather than posting to the cloud.
Any thoughts?
Greetings from #RadarFestival! #TesseracT
The #Zotero #OCR Plugin is now compatible with Zotero 7. 
Thank you @sw and A. Borel for all the work on this and releasing version 0.7.0 of the plugin. https://github.com/UB-Mannheim/zotero-ocr #tesseract
Update on my Peertube channel! Latest video is "Starburn", by #VOLA, from their album "Inmazes".
Previous videos are:
- "Survival", by #TesseracT
- "Eye Of Chaos", by #OnceHuman
- "People", by #KingCrimson
- "Ride" by #Cathedral
- "Generic Hostile Rigmarole" by #Nonoia
- "Dying Light", by #InfectedRain
- "On The Top", by #Jinjer
- "King" by #TesseracT
As always, boosts, comments & likes are very appreciated! 
