FOBG/README.md

🪄 FOBG — AI Background Remover

Remove backgrounds from images and videos using AI, powered by rembg.

Features

• Image BG Removal — Upload any image and get back a transparent PNG
• Video BG Removal — Frame-by-frame processing with true alpha channel output
• Multiple AI Models — u2net, u2netp, u2net_human_seg, isnet-general-use, isnet-anime
• Gradio UI — Clean, interactive web interface with fine-tuning controls
• CLI Mode — Script-friendly command-line interface for batch processing
• Fine-tune Controls — Alpha matting, foreground/background thresholds, erosion mask

Installation

Prerequisites

• Python 3.9 or later
• ffmpeg — required for video processing (Mac: brew install ffmpeg; Debian/Ubuntu: sudo apt install ffmpeg)

Option 1 — Virtual Environment (Recommended)

cd /mnt/games/Coding/Python/AI/FOBG
python -m venv .venv
source .venv/bin/activate
pip install -r requirements.txt

Option 2 — Global Install

pip install -r requirements.txt

Note: The first run will automatically download the selected AI model(s) (~40–170 MB depending on the model).

Usage

Web UI (Gradio)

python main.py

Opens at http://localhost:7860.

Additional options:

python main.py --host 0.0.0.0 --port 7860    # Bind to a specific host/port
python main.py --share                         # Create a public share link
python main.py --no-browser                    # Do not open browser automatically

CLI Mode

python main.py --cli input.jpg                 # Remove image background
python main.py --cli input.jpg -o result.png   # Specify output path
python main.py --cli video.mp4 --video          # Remove background from video
python main.py --cli video.mp4 --video -o out.mov   # .mov output preserves alpha channel

CLI options:

Flag	Description
--model <name>	AI model to use (default: u2net)
--alpha-matting	Enable alpha matting for fine edges
--no-alpha-matting	Disable alpha matting
--video	Treat input as video (frame-by-frame processing)

Available Models

Model	Description	Speed
u2net	Default, balanced quality/speed	⚡ Fast
u2netp	Lightweight, slightly lower quality	🚀 Fastest
u2net_human_seg	Optimized for human segmentation	⚡ Fast
u2net_3b	Heavy, higher accuracy	🐢 Slower
isnet-general-use	General purpose, high accuracy	🐢 Slower
isnet-anime	Optimized for anime illustrations	⚡ Fast

Gradio Controls

Image Tab

• Upload Image — Accepts PNG, JPEG, or WebP
• Model — Choose the AI model
• Alpha Matting — Fine-tune edges (slower but cleaner results)
• Foreground Threshold — How much of the image is considered foreground (1–255)
• Background Threshold — How much is considered background (1–255)
• Erosion Mask Dilation — Morphological operation to refine the mask (0–32)

Video Tab

• Upload Video — Accepts MP4, AVI, WebM, and MOV
• Same model and fine-tuning controls as the image tab

Notes

• Image processing typically takes 1–5 seconds depending on resolution and model
• Video processing is frame-by-frame and can take minutes for longer clips
• Image output is saved as PNG with transparency
• Video output uses .mov with ProRes 4444 alpha when alpha is desired, or MP4 with libx264 (black background where alpha is zero)
• All results are stored in the output/ directory

Project Structure

FOBG/
├── main.py                   # Entry point (Gradio + CLI)
├── requirements.txt          # Python dependencies
├── src/
│   ├── core/
│   │   ├── bg_remove.py      # Image background removal (rembg)
│   │   └── video_bg_remove.py# Video background removal (OpenCV + rembg + ffmpeg)
│   └── ui/
│       └── app.py            # Gradio web interface
├── output/                   # Processed results saved here
├── tests/                    # Unit tests
└── docs/                     # Extended documentation

License

MIT License — see LICENSE for details.