Vision Toolkit (Image Processing)

A modular Python tool to perform image processing and computer vision tasks. The project uses a multi-level CLI to easily expand into different fields of computer vision and image processing.

Available Fields and Tools

1. image_processing

General image processing utilities.

• transparency: Converts images between different formats (e.g., PNG to JPG, JPG to PNG). By default, it automatically makes near-white and light-grey backgrounds transparent. If the output format does not support transparency (like JPEG), the script automatically flattens the image onto a white background to prevent errors.

2. computer_vision

(Placeholder for future computer vision features like classification, detection, etc.)

Supported Formats

See the format_support_checklist.md file for a full list of supported formats. Currently implemented:

• PNG
• JPEG / JPG
• ICO

Requirements

• Python >= 3.10
• Pillow
• NumPy

Usage

You can use the new command structure via uv run or directly running main.py.

uv run image-processing <field> <tool_name> [args]

Or using python:

python main.py <field> <tool_name> [args]

Examples

Convert a PNG to a JPG (automatically flattens transparent background to white):

python main.py image_processing transparency input.png -o output.jpg

Convert a JPG to a PNG (with automatic background transparency):

python main.py image_processing transparency input.jpg -o output.png

Convert an image without modifying transparency (keep original background):

python main.py image_processing transparency input.png -o output.ico --no-transparent