Getting started

Getting started with Celantur Container

Overview

Celantur Container is a software application that anonymizes images and videos fully automated.

It is shipped as a scalable Docker container image and can be deployed on-premise, in the cloud and on-the-edge.

Multiple object types can be anonymized:

• Faces

• Persons

• License plates

• Vehicles

Your required object type is not in the list? We can help you. Please ask your Celantur sales representative.

How Celantur Container can be used

Requirements and Installation 🛠

Requirements

Check if your system fulfills the requirements to run Celantur Container

Installation on Linux

How to install Celantur Container on Linux

Installation on Windows (experimental)

How to install Celantur Container on Windows

How to use Celantur Container 💻

Celantur Container can be used in different data input/output modes:

• Batch and Stream mode

• REST API (v1) mode

  • #synchronous-processing
  • #aynchronous-processing
• TCP mode

It's also possible to extract segmentation masks and metadata per processed image.

Customize blurring to optimize visual appearance of the anonymization for your specific requirements.

Enable the CPU mode in case you have no access to GPUs.

Celantur Container can run completely offline

What's new

Celantur Container is getting continuously improved with new features. Read more about past and recent releases in the Release Notes:

Release Notes

Anonymizing Dashcam Videos with Celantur Container

1. Install Celantur Container on your system following the installation guide. Ensure Docker Container is running before proceeding.

2. Check if you video's codec is supported.

3. Make sure you use the object-detection-v2 model: --model object-detection-v2 (Batch/Stream mode) model="object-detection-v2" (API mode)

4. Optimize:

   1. Adapt thresholds (lower = better detections, potentially more false positives; higher = less false positives) --face-threshold 0.2 , --license-plate-threshold 0.2 (Batch/Stream mode) face-threshold="0.2" , license-plate-threshold="0.2" (API mode)

   2. Enable keep-bitrate flag, to make sure file size of the blurred video remains similar to the original one: --keep-bit-rate true (Batch/Stream mode) keep-bit-rate="true" (API mode)