Image Splitter

Split separate icons, logos, or marks from a single image into individual PNG files.

This repo currently ships:

• a local browser app for previewing bounds and exporting split PNGs
• a CLI for batch-style local runs
• a Python image-processing pipeline built around connected-component detection

Live app

Production:

https://skill-deploy-ms3se8xl4f.vercel.app

Project layout

• prototype/ - the current app, deploy config, tests, and image splitting logic
• prototype/web_app.py - local Python server
• prototype/web/index.html - browser UI
• prototype/split_icons.py - detection and crop pipeline

Quick start

From prototype/:

1. python3 -m pip install -r requirements.txt
2. python3 web_app.py --port 8008
3. Open http://127.0.0.1:8008

You can also run the CLI directly:

python3 split_icons.py /path/to/your/image.png --output-dir out_icons

Features

• image upload with browser preview
• detection bounds overlay before export
• isolated PNG export plus manifest zip
• settings for background sensitivity, grouping, and quality
• advanced controls for fine-tuning cleanup and export behavior
• local automated tests for health, happy path, and invalid uploads

Deploy

The Vercel-linked app lives in prototype/.

Preview deploy:

cd prototype && npx vercel deploy -y

Production deploy:

cd prototype && npx vercel deploy --prod -y

Tests

From prototype/:

python3 -m unittest test_web_app.py

Notes

• works best when objects are visually separated from each other
• touching objects are treated as one component
• textured or noisy backgrounds may still need preprocessing

For implementation details, runtime safeguards, Docker usage, and the full local workflow, see prototype/README.md.