swift-ocr-cli

swift-ocr-cli is a Swift command-line tool that leverages Apple's Vision framework (macOS only) to perform Optical Character Recognition (OCR) on images. This tool supports multiple input methods and output formats, making it flexible for various use cases.

$ swift-ocr-cli-arm-mac ./docs/images/test.png ko,en --json

{
  "text" : "[EPA 및 DHA 함유 유지] 혈중 중성지질 개선\n• 혈행 개선에 도움을 줄 수 있음, 기억력 개선\n에 동장에 도수 있 수 건조한 눈을 개선하여\n[비타민트 항산화 작용을 하여 유해산소로부터\n세포를 보호하는데 필요\n1일 섭취량 : 1캡슐(1,200 mg)\n1일 섭취량 당\n함량\n%영양성분기준치\n열량\n10 kcal\n탄수화물\n0 g\n단백질\n지방\n0 g\n1.2 g\n나트륨\n0 mg\nEPA 와 DHA의 합 900 mg\n비타민E\n3.4 mg a-TE\n0 %\n0 %\n2%\n0 %\n31 %\n※ %영양성분기준치 : 1일 영양성분기준치에 대한 비율"
}

$ swift-ocr-cli-arm-mac ./docs/images/test.png ko,en --json --coordinate

• Output

---

Installation with Homebrew

You can easily install swift-ocr-cli using Homebrew:

Install swift-ocr

brew tap az5app/tap
brew install swift-ocr

---

Features

• Multiple Input Forms:
  Accepts either an image file path or a base64-encoded image string.

• Output Formats:

  • Plain Text: Simply outputs the recognized text.
  • JSON Format: Use the --json flag to receive coordinates in JSON.

• Coordinates Mode:
  Use the --coordinate flag to get recognized text along with bounding box coordinates.
  The output includes:

  • Recognized text.
  • Bounding box dimensions (width and height).
  • A default confidence score.
  • Pixel coordinates (x, y) for the top-left corner of the bounding box.

---

Requirements

• macOS: Version 10.15 or later
• Xcode: Version 13 or later (to build Swift tools using Vision)
• Swift: Version 5.5 or later

---

Build & Run

Cloning the Repository

Clone the repository and change to the project directory:

git clone <REPO_URL> swift-ocr-cli
cd swift-ocr-cli

Building the Tool

Build the project using Swift Package Manager:

swift build

Running the Tool

Run the tool with the following syntax:

swift run swift-ocr-cli <imageFilePath or base64Str> [recognitionLanguages] [--coordinate] [--json]

Example Usages

1. Basic Text Recognition (defaults to English):

   swift run swift-ocr-cli /path/to/image.jpg

2. Specifying Custom Recognition Languages:

   swift run swift-ocr-cli /path/to/image.jpg ko-KR,en-US

3. Coordinate Mode with JSON Output:

   swift run swift-ocr-cli /path/to/image.jpg --coordinate --json

4. Using Base64 Input:

   swift run swift-ocr-cli <base64EncodedImageString> --coordinate

---

Release Instructions

To build release versions for different architectures, use the provided scripts.

Running the Release Script

./scripts/release.sh

Manual Build Commands

• For arm64:

  swift build -c release --arch arm64 --build-path .build/arm64

• For x86_64:

  swift build -c release --arch x86_64 --build-path .build/x86_64

---

Testing

A sample test is included which uses a base64-encoded image. Run the tests with:

swift test

---

License

Apache License Version 2.0