Obsidian OCR allows you to search for text in your images and PDFs
tesseract
for OCRimagemagick
for pdf to png conversion
❗Make sure the executables are in your path. If you don't know how look here: https://www.architectryan.com/2018/03/17/add-to-the-path-on-windows-10/❗
- Install tesseract
- Install imagemagick
- If you want to use another language that english, you have to install it manually. (It's really simple)
- Go to https://github.com/tesseract-ocr/tessdata_fast/
- Download the file for your language (i.e. for German download ger.traineddata)
- Move to file to
C:\Program Files\Tesseract-OCR\tessdata
❗The automatic installation features is still in very early development. Expect bugs❗
ℹ️ Obsidian OCR uses chocolatey for automatic installation ℹ
- Install & enable the plugin
- Under
settings
, press theautomatic install
button
- Install tesseract:
brew install tesseract
- Install tesseract-lang :
brew install tesseract-lang
- Install imagemagick :
brew install imagemagick
For the path: check where the binaries are located and add these to "/private/etc/paths" (I also added them to ~/.zshrc, not sure if that is needed)
brew list tesseract
in my case:/opt/homebrew/Cellar/tesseract/5.2.0/bin/
brew list tesseract-lang
in my case:/opt/homebrew/Cellar/tesseract/5.2.0/bin/
brew list imagemagick
in my case:/opt/homebrew/Cellar/imagemagick/7.1.0-43/bin/
- Run
sudo apt install -y tesseract-ocr imagemagick
- Install any languages you need by installing the appropriate package (usually named
tesseract-ocr-<lang>
)
❗The automatic installation features is still in very early development. Expect bugs❗
- Install & enable the plugin
- Under
settings
, press theautomatic install
button
- Run
sudo pacman -S tesseract imagemagick
- Install any languages you need by installing the appropriate package (usually named
tesseract-data-<lang>
)
❗The automatic installation features is still in very early development. Expect bugs❗
- Install & enable the plugin
- Under
settings
, press theautomatic install
button
If Obsidian is running via the Flatpak installation (such as provided by default in Pop!_OS) then this plugin will not operate. Flatpak sandboxing will change the filepaths so even providing host access will still be problematic. If you have a Flatpak installation you will need to reinstall via a different method to successfully use this plugin.
- Simply install the plugin from the community plugins
- Upon enabling the plugin, you will see a progress bar in the bottom right, indicating that your vault is currently
being processed.
- Upon completion, you can either use the
Search OCR
command or themagnifying-glass icon
in the ribbon to open the search menu. - When adding a new PDF / PNG to your vault, the file is automatically being searched for text.
- Upon starting Obsidian, you will now see another progress bar, indicating that all transcripts are being cached.
Obsidian OCR offers a variety of settings you can configure yourself.
Name | Description | Default |
---|---|---|
Max OCR Processes | The maximum amount of ocr processes running at the same time. Increasing this speeds up indexing but also increases cpu usage | 1 |
Max caching processes | The maximum amount of caching processes running at the same time. Increaing this speeds up caching but also increases cpu usage | 10 |
OCR Image | Decides whether or not images (.png, .jpg, .jpeg) are OCRed | true |
OCR PDF | Decides whether or not PDFs (.pdf) are OCRed | true |
Image density | The density of generated PNGs, in dpi. Increasing this helps to OCR smaller text, but increases cpu usage | 300 |
Image quality | The quality of generated PNGs. Increasing this helps to OCR smaller text, but increaes cpu usage | 98 |
Additional imagemagick args | Additional commandline arguments passed to imagemagick when converting a PDF to PNG(s) | |
Additional search paths | Additional paths that will be searched when looking for external dependencies. Useful when installing into custom directories | |
OCR Provider | The OCR provider that will be used. See below for a description of providers | NoOp |
Name | Description |
---|---|
NoOp | The NoOp (no operation) provider does, as the name implies, nothing and is only a dummy provider. To get real OCR capability, you have to switch to another provider |
Tesseract | OCR provider using tesseract to OCR documents locally on your computer |
By default, tesseract offers two languages to choose from: eng
and osd
.
- Osd isn't a language, but rather stands for
Orientation and script detection
and is therefore not useful for our use. - Eng stands for english and should be chosen if your documents are written in english.
ℹ After switching your language, only newly indexed documents use the new language. You can reindex your already added
documents by using the Delete all transcripts
command ℹ
Tesseract supports langs
and scripts
for text recognition.
- Langs support a single language
- Scripts support multiple languages with similar characters. Latin, for example, supports all latin-based languages (English, German, Italian, France, ...) except for Vietnamese
To add a custom OCR provider, create a new class that extends OCRProvider
and register it
using OCRProviderManager.registerOCRProviders(new MyCustomProvider())