AI-powered background removal with local and cloud options

Remove backgrounds from images instantly with AI. Choose between local processing (free) or withoutBG Pro (best quality).

# Install and run in one go
pip install withoutbg

# Remove background from your first image
python -c "from withoutbg import WithoutBG; WithoutBG.opensource().remove_background('your-photo.jpg').save('result.png')"

That's it! Your photo now has a transparent background. ✨

Processing 1-10 images occasionally? → withoutBG Pro (zero setup, best quality)
Processing 100+ images? → Local model (free, reusable)
Need offline processing? → Local model
Want best possible quality? → withoutBG Pro
Building commercial product? → withoutBG Pro (scalable)
Need fastest processing? → withoutBG Pro (optimized infrastructure)

Compare Focus vs Pro →

View Complete Dockerized Web App Documentation →

docker run -p 80:80 withoutbg/app:latest

# Open in browser
open http://localhost

✅ Multi-platform support: Works on Intel/AMD (amd64) and ARM (arm64) architectures

View Complete Python SDK Documentation →

# Install (using uv - recommended)
uv add withoutbg

# Or with pip
pip install withoutbg

Local Processing (Free):

from withoutbg import WithoutBG

# Initialize model once, reuse for multiple images (efficient!)
model = WithoutBG.opensource()
result = model.remove_background("input.jpg")  # Returns PIL Image.Image
result.save("output.png")

# Standard PIL operations work!
result.show()  # View instantly
result.resize((500, 500))  # Resize
result.save("output.webp", quality=95)  # Different format

withoutBG Pro (Best Quality):

See Pro API Results →

from withoutbg import WithoutBG

# Get your API key from withoutbg.com
model = WithoutBG.api(api_key="sk_your_key")
result = model.remove_background("input.jpg")
result.save("output.png")

# Or set environment variable (recommended)
# export WITHOUTBG_API_KEY=sk_your_key
model = WithoutBG.api(api_key="sk_your_key")

CLI:

# Process single image
withoutbg photo.jpg

# Process entire photo album
withoutbg ~/Photos/vacation/ --batch --output-dir ~/Photos/vacation-no-bg/

# Convert to JPEG with white background (for printing)
withoutbg portrait.jpg --format jpg --quality 95

# Use withoutBG Pro for best quality
export WITHOUTBG_API_KEY=sk_your_key
withoutbg wedding-photo.jpg --use-api

# Process and watch progress
withoutbg photo.jpg --verbose

Don't have uv yet? It's a fast, modern Python package installer - get it at astral.sh/uv

See More Focus Model Results →

We are finalizing a new architecture designed to deliver superior accuracy and finer detail. Star the repo to bookmark this upgrade for your future toolkit ⭐.

This is a monorepo containing multiple components:

withoutbg/
├── packages/          # Reusable packages
│   └── python/        # Core Python SDK (published to PyPI)
│
├── apps/              # End-user applications
│   └── web/           # Web application (React + FastAPI)
│
├── integrations/      # Third-party tool integrations
│   └── (future: GIMP, Photoshop, Figma plugins)
│
├── models/            # Shared ML model files
│   └── checkpoints/   # ONNX model files
│
├── docs/              # Documentation
└── scripts/           # Development scripts

Core library for background removal. Published to PyPI.

• Install: uv add withoutbg (or pip install withoutbg)
• Use: Python API + CLI
• Models: Focus v1.0.0 (local), withoutBG Pro

Modern web interface with drag-and-drop UI.

• Stack: React 18 + FastAPI + Nginx
• Deploy: Docker Compose
• Features: Batch processing, live preview

Plugins for popular creative tools.

• GIMP plugin
• Photoshop extension
• Figma plugin
• Blender addon

• ✨ Local Processing: Free, private, offline with Focus v1.0.0
• 🚀 withoutBG Pro: Best quality, scalable, 99.9% uptime
• 📦 Batch Processing: Process multiple images efficiently
• 🌐 Web Interface: Beautiful drag-and-drop UI
• 🔧 CLI Tool: Command-line automation
• 🎨 Integrations: Work in your favorite tools

All methods return PIL Image objects with transparent backgrounds (RGBA mode):

from withoutbg import WithoutBG

model = WithoutBG.opensource()
result = model.remove_background("photo.jpg")  # Returns PIL Image.Image

# Save in different formats
result.save("output.png")    # PNG with transparency
result.save("output.webp")   # WebP with transparency
result.save("output.jpg", quality=95)  # JPEG (no transparency)

JPEG Files Don't Support Transparency:

# ❌ This loses transparency (JPEG doesn't support alpha):
result.save("output.jpg")

# ✅ Use PNG or WebP for transparency:
result.save("output.png")  # Keeps alpha channel
result.save("output.webp")  # Also works!

Model Downloads Happen on First Run:

# First run: ~5-10 seconds (downloading ~320MB of models from HuggingFace)
model = WithoutBG.opensource()  # Downloads models to cache

# Second run: Instant! (models cached locally)
model = WithoutBG.opensource()  # Uses cache
result = model.remove_background("photo.jpg")  # Now processes in ~2-5s

Output File Naming:

• Single file: photo.jpg → photo-withoutbg.png
• Batch: Creates photo1-withoutbg.png, photo2-withoutbg.png, etc.
• Custom: Use --output or --output-dir to specify

from withoutbg import WithoutBG

# Initialize model once - loaded into memory
model = WithoutBG.opensource()

# Process multiple images - model is reused (much faster!)
images = ["photo1.jpg", "photo2.jpg", "photo3.jpg"]
results = model.remove_background_batch(images, output_dir="results/")
# Returns list of PIL Image objects

def show_progress(progress):
    print(f"Processing: {progress * 100:.0f}%")

model = WithoutBG.opensource()
result = model.remove_background("photo.jpg", progress_callback=show_progress)

from withoutbg import WithoutBG, APIError, WithoutBGError

try:
    model = WithoutBG.api(api_key="sk_your_key")
    result = model.remove_background("photo.jpg")
    result.save("output.png")
except APIError as e:
    print(f"API error: {e}")
except WithoutBGError as e:
    print(f"Processing error: {e}")

Model Files (cached after first download):

• ISNet segmentation: 177 MB
• Depth Anything V2: 99 MB
• Focus Matting: 27 MB
• Focus Refiner: 15 MB
• Total: ~320 MB (one-time download, cached locally)

💡 Pro Tip: For batch processing, initialize the model once and reuse it - this is 10-100x faster than reinitializing for each image!

Model download fails?

• Check your internet connection
• Models are downloaded from HuggingFace on first run (~320MB total)
• Or use local model files (see Configuration)

Out of memory?

• Try processing smaller images or use withoutBG Pro
• Reduce batch size when processing multiple images

Import errors or "module not found"?

# Make sure you're in the right environment
which python  # Check your Python path
pip list | grep withoutbg  # Verify installation

# If using virtual environment
source venv/bin/activate  # Activate first
pip install withoutbg  # Then install

API key invalid?

• Get your API key from withoutbg.com
• Set environment variable: export WITHOUTBG_API_KEY=sk_your_key

Slow processing on first run?

• Normal! Models are being downloaded (~320MB)
• Subsequent runs will be much faster (~2-5s per image)

• Python SDK Docs - Complete API reference and examples
• Python SDK Documentation - Online documentation
• Web App Docs - Deployment and development guide
• Dockerized Web App Documentation - Online documentation
• withoutBG Pro API Results - See example outputs
• Focus Model Results - See example outputs
• Compare Focus vs Pro - Model comparison

cd packages/python

# Install in development mode (using uv - recommended)
uv sync --extra dev

# Or with pip
pip install -e ".[dev]"

# Run tests
pytest

# Format code
black src/ tests/
ruff check src/ tests/

# Development mode (hot-reload)
docker-compose -f apps/web/docker-compose.yml up

# Or run components separately
cd apps/web/backend
uv sync
uvicorn app.main:app --reload

cd apps/web/frontend
npm install
npm run dev

Our most advanced open source model with:

• ✅ Significantly better edge detail - Crisp, clean edges
• ✅ Superior hair/fur handling - Natural-looking fine details
• ✅ Better generalization - Works on diverse image types

See example outputs above and check sample-results/ for more visual comparisons.

Apache License 2.0 - see LICENSE

• Depth Anything V2: Apache 2.0 License (only vits model)
• ISNet: Apache 2.0 License

• segmentation-models-pytorch: MIT License (used to train matting/refiner models)

See THIRD_PARTY_LICENSES.md for complete attribution.

We welcome contributions! See CONTRIBUTING.md for guidelines.

1. Fork the repository
2. Create a feature branch
3. Make your changes
4. Add tests
5. Submit a pull request

• Bug Reports: GitHub Issues
• Feature Requests: GitHub Discussions
• Commercial Support: [email protected]