Which Tesseract OCR library should I use?

It depends on your language and whether you want a native binding or a CLI wrapper. For Python, pytesseract is the simplest and tesserocr is faster in-process. For JavaScript, tesseract.js runs in the browser and Node via WebAssembly. For Java use Tess4J, for Go use gosseract, and for C++ use the native TessBaseAPI.

What is the difference between a CLI wrapper and a native binding?

A CLI wrapper (like pytesseract) calls the tesseract executable as a subprocess: it is easy to set up but slower per call. A native binding (like tesserocr or Tess4J) links against the C/C++ API in-process: faster and more controllable, but requires the development libraries to be installed.

Tesseract OCR Libraries & Wrappers for Python, JS, Java, Go, .NET

Q: Does tesseract.js support PDFs?

No. tesseract.js is a WebAssembly port that recognises images; it does not support PDF input or modify the recognition model. Convert PDF pages to images first, or use a library such as Scribe.js that adds PDF support.

Q: Do I need to install the Tesseract engine separately?

Yes for most wrappers. CLI wrappers and native bindings rely on the installed Tesseract engine and language data; only tesseract.js bundles a WebAssembly build of the engine.

Which library should I use?

Pick the row for your language, then read the section below.

Language	Recommended	Type	Notes
Python	`pytesseract`	CLI wrapper	Simplest; see the Python guide. Use `tesserocr` for in-process speed.
JavaScript / TypeScript	`tesseract.js`	WASM	Runs in browser and Node; no PDF support.
C++	`TessBaseAPI`	Native	The core API; maximum control.
Java	`Tess4J`	JNA binding	Needs the native libraries on the classpath.
Go	`gosseract`	cgo binding	Requires `libtesseract-dev`.
C# / .NET	`Tesseract` (.NET)	Native wrapper	NuGet package; bundles native binaries.
Rust	`leptess`	Native binding	Links against Leptonica + Tesseract.
PHP	`thiagoalessio/tesseract_ocr`	CLI wrapper	Composer package.

Python

The largest ecosystem: a CLI wrapper plus several native bindings.

pytesseract — the standard CLI wrapper. Easy, but spawns a process per call.
tesserocr — native binding to the C++ API; faster in-process, requires libtesseract-dev.
pyocr — a wrapper supporting multiple engines.
aiopytesseract — async wrapper for concurrent workloads.

python

pip install pytesseract pillow
import pytesseract
from PIL import Image
print(pytesseract.image_to_string(Image.open("image.png")))

Full walkthrough: Tesseract OCR in Python (pytesseract).

JavaScript / TypeScript

tesseract.js brings the engine to the browser and Node through WebAssembly.

javascript

npm install tesseract.js
const Tesseract = require('tesseract.js');
Tesseract.recognize('image.png', 'eng', { logger: m => console.log(m) })
  .then(({ data: { text } }) => console.log(text));

Note: tesseract.js does not support PDF files and does not modify the recognition model. For Node-only use, node-tesseract-ocr wraps the installed CLI instead.

C++

The native TessBaseAPI is the reference integration and what most bindings wrap.

ocr.cpp

#include <tesseract/baseapi.h>
#include <leptonica/allheaders.h>

tesseract::TessBaseAPI* api = new tesseract::TessBaseAPI();
api->Init(NULL, "eng");
Pix* image = pixRead("image.png");
api->SetImage(image);
char* text = api->GetUTF8Text();
printf("%s", text);
api->End(); pixDestroy(&image);

Build against libtesseract-dev and libleptonica-dev (see installation).

Java

Tess4J is the mature JNA wrapper; JavaCPP presets are an alternative.

Ocr.java

// Maven: net.sourceforge.tess4j:tess4j
import net.sourceforge.tess4j.Tesseract;
Tesseract t = new Tesseract();
t.setDatapath("tessdata");
String text = t.doOCR(new java.io.File("image.png"));

Go

gosseract uses cgo to link against the Tesseract C++ library.

main.go

go get github.com/otiai10/gosseract/v2
import "github.com/otiai10/gosseract/v2"
client := gosseract.NewClient()
defer client.Close()
client.SetImage("image.png")
text, _ := client.Text()

Requires libtesseract-dev and libleptonica-dev (Linux) or the Homebrew packages (macOS).

C# / .NET

The Tesseract NuGet package wraps the native library and ships binaries.

Program.cs

// dotnet add package Tesseract
using Tesseract;
using var engine = new TesseractEngine(@"./tessdata", "eng", EngineMode.Default);
using var img = Pix.LoadFromFile("image.png");
using var page = engine.Process(img);
Console.WriteLine(page.GetText());

Other languages

Community bindings exist for most ecosystems.

Rust — leptess (native binding), rusty-tesseract (CLI wrapper).
PHP — thiagoalessio/tesseract_ocr (Composer, CLI wrapper).
Ruby — rtesseract (CLI wrapper).
C — the Tesseract C API (tesseract/capi.h).

CLI wrapper vs native binding

The choice matters more than the language for performance.

Factor	CLI wrapper	Native binding
Setup	Easy — just install the engine	Needs dev headers/libraries
Speed per call	Slower (process per call)	Faster (in-process)
Control	CLI flags and config only	Full API (iterators, boxes, confidence)
Examples	`pytesseract`, `rtesseract`, `node-tesseract-ocr`	`tesserocr`, `Tess4J`, `gosseract`, `TessBaseAPI`

For one-off scripts, a CLI wrapper is usually the fastest path. For high-volume or latency-sensitive services, prefer a native binding or batch through the CLI (see the performance notes).

Libraries FAQ

Which Tesseract library should I use?

Match the row in the table above to your language. Use a native binding for speed and control, a CLI wrapper for simplicity.

Do I need the engine installed separately?

Yes for most wrappers — they rely on the installed Tesseract engine and language data. Only tesseract.js bundles a WebAssembly build of the engine.

Does tesseract.js support PDFs?

No. It recognises images; convert PDF pages to images first, or use a library that adds PDF support.

Which is fastest?

Native in-process bindings such as tesserocr, Tess4J and gosseract avoid per-call process startup. For CLI wrappers, batch pages and use multiprocessing.

Install the engine, then pick your language.