Privacy Mask

Privacy Mask is a pre-processing tool by Inspektlabs that automatically detects and masks sensitive information—such as license plate numbers and human faces—in vehicle inspection images. It ensures privacy compliance and data security before the images are sent to the Inspektlabs API or stored in the cloud. Powered by advanced AI-based computer vision, Inspekt-Mask delivers fast and accurate anonymization without compromising image quality. This tool enables customers to maintain regulatory compliance, protect end-user privacy, and streamline their inspection workflows.

The accompanying Privacy Docker Image provides an end-to-end pipeline for face and license plate detection and anonymization, returning:

• Bounding boxes of detected faces and license plates
• A Base64-encoded image with those regions blurred for privacy

This guarantees sensitive information is anonymized prior to any further processing or storage.

Image Contents

Directory Structure Inside Docker:

/Module/
 ├── main.py
 ├── requirements.txt
 ├── Module/
 │    ├── LicenseModule/
 │    └── FaceModule/
 └── other dependencies...

File Descriptions:

• main.py — Main entry point of the application (runs automatically when the container starts).
• requirements.txt — Lists all Python dependencies (e.g., OpenCV, YOLOv5, etc.).
• Dockerfile — Defines the build configuration for the image.

Environment Details

Base Image

python:3.11-slim

This lightweight Python environment ensures fast container startup and minimal resource usage.

How to Use

1. Load the Docker Image

Load the provided .tar file into Docker:

docker load -i face-license-app.tar

2. Run the Container

Start the application:

docker run --rm -it face-license-app

This command executes main.py automatically inside the container.

Input / Output Format

Input Format

The main.py handler function accepts a dictionary input:

{
    "image_base64": "<base64-encoded-image-string>"
}

Output Format

The output includes detected bounding boxes and the anonymized image:

{
    "face_bbox": [
        [x1, y1, x2, y2],
        [x1, y1, x2, y2]
    ],
    "license_bbox": [x1, y1, x2, y2],
    "base64_image": "<base64-encoded-output-image>"
}

Notes:

• face_bbox — List of detected face bounding boxes
• license_bbox — Coordinates of detected license plate
• base64_image — The blurred, privacy-safe output image

The output is JSON-compatible, ready for use in APIs, logs, or downstream applications.

←Motion Sensor

On this page

• Privacy Mask
• Image Contents
• File Descriptions:
• Environment Details
• How to Use
• Input / Output Format
• Output Format