Censorship

Overview

Censorship is a Python project designed to automatically detect and censor NSFW (Not Safe For Work) content in images. The project uses a pre-trained ONNX model for nude detection and provides both censored and debug outputs for each processed image.

Features

• Automatically scans images for NSFW content.
• Censors detected nude regions using pixelation.
• Generates debug images with detection overlays for analysis.
• Logs progress and errors during processing.
• Supports batch processing of images.

Requirements

• Python 3.8+
• ONNX Runtime
• OpenCV
• nudenet (>=3.4.2)
• libgl1 (system dependency)

Usage

Install dependencies:

pip install -r requirements.txt

or :

pip install --upgrade "nudenet>=3.4.2";
pip install opencv-python;

sudo apt-get install -y libgl1

Place images to be processed in the files/ directory. Run the main script:

source .venv/bin/activate

python main.py

Command Line Arguments

You can also customize parameters, for example:

python main.py --input my_images --output results --pixel-size 8 --debug --force-process --save-result-json

Argument	Default	Description
-i, --input	files	Input directory containing images to process
-o, --output	output	Output directory to save processed images
-e, --extensions	.jpeg .jpg .png	List of image file extensions to process
-r, --recursive	(flag)	Recursively search for images in subdirectories
-p, --pixel-size	3	Pixelation block size for censorship (higher = more pixelated, range: 1-256)
-d, --debug	(flag)	Enable debug mode to save debug images with detection overlays
--save-result-json	(flag)	Save the results to a JSON file
-l, --labels	FEMALE_GENITALIA and MALE_GENITALIA	List of labels to detect
--force-process	(flag)	Force processing of images even when they have no changes
--logpath	logs	Folder to the log file
--loglevel	INFO	Logging level for the log file (DEBUG, INFO, WARNING, ERROR, CRITICAL)