Image Helper

A Python CLI tool for common image processing operations including resizing, background removal, and vectorization.

Dependencies

• Pillow: Image manipulation
• rembg: Background removal
• opencv-python: Image processing and inpainting
• numpy: Numerical operations
• onnxruntime: AI model inference for advanced watermark removal

Features

• Resize: Resize images to specific dimensions with optional aspect ratio preservation
• Remove Background: Automatically remove backgrounds from images
• Vectorize: Convert raster images to SVG vector format
• Auto-Crop: Automatically detect and remove blank borders from images
• Favicon: Generate multi-size favicon (.ico) files for websites
• Remove Watermark: Remove watermarks from the bottom-right corner using content-aware inpainting

Installation

1. Clone this repository
2. Install Python dependencies:

pip install -r requirements.txt

3. For vectorization, install potrace:

# Ubuntu/Debian
sudo apt-get install potrace

# macOS
brew install potrace

Usage

Resize Image

python main.py resize input.jpg output.jpg --width 800 --height 600
python main.py resize input.jpg output.jpg --width 800 --height 600 --keep-aspect

Remove Background

python main.py remove-bg input.png output.png

Vectorize Image

python main.py vectorize input.jpg output.svg
python main.py vectorize input.jpg output.svg --colors 16

Auto-Crop Image

# Automatically remove blank borders
python main.py auto-crop input.png output.png

# Adjust sensitivity (lower = more aggressive cropping)
python main.py auto-crop input.png output.png --threshold 5

Generate Favicon

# Generate favicon with default sizes (16x16, 32x32, 48x48)
python main.py favicon logo.png favicon.ico

# Generate favicon with custom sizes
python main.py favicon logo.png favicon.ico --sizes 16 32 48 64 128 256

# Generate high-quality favicon for modern browsers
python main.py favicon logo.png favicon.ico --sizes 32 64 128

6. Remove Watermark

Version 1 (OpenCV Inpainting)

Remove watermarks from the bottom-right corner of images using OpenCV's content-aware inpainting:

python main.py remove-watermark-v1 input.jpg output.jpg
python main.py remove-watermark-v1 input.jpg output.jpg --width 25 --height 20 --method ns

Options:

• --width: Watermark width as percentage of image width (default: 30)
• --height: Watermark height as percentage of image height (default: 15)
• --method: Inpainting method - telea (fast) or ns (Navier-Stokes, higher quality)

Version 2 (LaMa AI Model) - Recommended

Advanced watermark removal using the LaMa (Large Mask Inpainting) AI model for superior results.

Note: This implementation is based on the logic from dinoBOLT/Gemini-Watermark-Remover, adapted and transformed to Python.

Setup Instructions:

1. Download the model (lama_fp32.onnx):

   👉 Click here to download from Google Drive

2. Place the file:

   Move the downloaded lama_fp32.onnx file into the assets/ folder inside the project directory.

   Your folder structure must look like this:

   images-helper/
   ├── assets/
   │   └── lama_fp32.onnx  <-- The file goes here (approx. 200MB)
   ├── src/
   │   └── helpers/
   │       └── ...
   ├── main.py
   └── ...
   

Usage:

python main.py remove-watermark-v2 input.jpg output.jpg
python main.py remove-watermark-v2 input.jpg output.jpg --model path/to/custom/model.onnx

Features:

• Uses state-of-the-art AI inpainting model (LaMa)
• Automatically detects and removes watermarks from bottom-right corner (15% × 15%)
• Preserves original image quality outside the watermark region
• Requires lama_fp32.onnx model in assets/ directory

Options:

• --model: Custom path to lama_fp32.onnx model (default: assets/lama_fp32.onnx)

Project Structure

images-helper/
├── main.py                 # CLI entry point
├── src/
│   ├── __init__.py
│   └── helpers/           # Image processing functions
│       ├── __init__.py
│       ├── resize.py
│       ├── remove_background.py
│       ├── vectorize.py
│       ├── auto_crop.py
│       ├── favicon.py
│       └── remove_watermark.py
├── requirements.txt
└── README.md

Adding New Features

To add a new image processing function:

1. Create a new file in src/helpers/ (e.g., src/helpers/my_feature.py)
2. Implement your function with a clear signature:

def my_feature(input_path: str, output_path: str, **kwargs) -> None:
    """
    Description of what this does.
    
    Args:
        input_path: Path to input image
        output_path: Path to save processed image
        **kwargs: Additional parameters
    """
    # Your implementation
    pass

3. Export it in src/helpers/__init__.py:

from .my_feature import my_feature

__all__ = [
    # ... existing exports
    'my_feature',
]

4. Add a command in main.py:

# Import at top
from src.helpers.my_feature import my_feature

# Add parser in main()
feature_parser = subparsers.add_parser('my-feature', help='Description')
feature_parser.add_argument('input', type=str, help='Input image path')
feature_parser.add_argument('output', type=str, help='Output image path')
# Add any additional arguments

# Add handler in the routing section
elif args.command == 'my-feature':
    my_feature(str(input_path), args.output)
    print(f"✓ Feature applied successfully: {args.output}")

License

MIT