Bahlib - Biometric Anonymizer Hub Library

A high-performance, lightweight Python library for 100% local biometric anonymization. Bahlib automatically detects human faces in images and videos, applying smooth Gaussian blur to protect privacy—without ever sending data to third-party servers. Features include multi-scale detection for group photos, tiled processing for small faces, and feathered blending for natural-looking results.

Get Started View on GitHub

pip install bahlib

# Anonymize faces in just 3 lines
from bahlib import Bahlib

with Bahlib() as bh:
    result = bh.anonymize("photo.jpg")
    
# Save your privacy-protected image
cv2.imwrite("anonymous.jpg", result)

Features

Built for Privacy-First Development

Everything you need to protect identities in images and videos

🔒

100% Local Processing

All face detection and blurring happens entirely on your machine. No cloud uploads, no API calls, complete data sovereignty.

🎯

Smooth Blending

Feathered elliptical masks create natural-looking blur that seamlessly blends with surrounding pixels.

👥

Multi-Scale Detection

Advanced tiled detection algorithm finds faces of all sizes, even in crowded group photos with small faces.

🎬

Video Support

Process video files frame-by-frame or run real-time anonymization on live webcam feeds.

📁

Batch Processing

Anonymize entire directories of images with a single command, with progress tracking and recursive support.

💻

CLI & Python API

Use from the command line for quick tasks, or integrate the Python API into your applications.

Installation

Get Started in Seconds

Install the package

pip install bahlib

Import in your code

from bahlib import Bahlib

Anonymize faces

result = Bahlib().anonymize("photo.jpg")

Usage

Three Simple Operations

Master the library with just three concepts

Python API

Integrate face anonymization into your Python applications with a clean, intuitive API.

Context manager for automatic resource cleanup
Accepts file paths or numpy arrays
Configurable blur strength and feathering
Multi-scale and tiled detection for group photos

from bahlib import Bahlib
import cv2

# Basic usage
with Bahlib() as bh:
    result = bh.anonymize("photo.jpg")
    cv2.imwrite("output.jpg", result)

# Group photos with small faces
with Bahlib(min_detection_confidence=0.3) as bh:
    result = bh.anonymize(
        "group.jpg",
        blur_strength=71,
        feather_amount=0.4,
        tiled=True,
        tile_size=320
    )

# Detection only (no blur)
faces = bh.detect_faces("photo.jpg")

Command Line Interface

Use bahlib directly from your terminal for quick anonymization tasks.

Process single images or entire directories
All Python API options available as flags
Progress indicators for batch processing
Preview results with --show flag

# Single image
$ bahlib image photo.jpg -o output.jpg

# Custom blur strength
$ bahlib image photo.jpg -o out.jpg --blur 71

# Group photo with tiled detection
$ bahlib image group.jpg -o out.jpg \
    --tiled --tile-size 320 --confidence 0.3

# Batch process a directory
$ bahlib batch ./photos ./output --recursive

# Preview result
$ bahlib image photo.jpg -o out.jpg --show

Video Processing

Anonymize faces in video files or live webcam feeds with frame-by-frame detection.

Process video files with preserved quality
Real-time webcam anonymization
Configurable detection confidence
Press 'q' to stop webcam mode

# Process video file
$ bahlib video input.mp4 -o output.mp4

# With custom settings
$ bahlib video input.mp4 -o out.mp4 \
    --blur 51 --confidence 0.4

# Real-time webcam
$ bahlib webcam

# Python API
from bahlib.video import anonymize_video

anonymize_video(
    "input.mp4",
    "output.mp4",
    blur_strength=51
)