Integration Scenarios

There are three primary options to integrate Celantur

  • Celantur Cloud Service

    • Cloud storage integration
  • Host on Celantur on your system (public/private cloud, on-the-edge or on-premise)

    • Batch and Stream Mode
    • Fetch from cloud storage
    • TCP Mode

Celantur Cloud Service

Cloud Storage Integration

Architecture

  1. Your system transfers images into an AWS S3 bucket (data storage)
  2. The Celantur Cloud Service anonymizes images from the S3 bucket at a specified interval
  3. Once the images have been anonymized, Celantur Cloud Service sends them back into the S3 bucket
  4. Optional: Celantur Cloud Service puts metadata for each image into the S3 bucket as well
  5. Your system can fetch the anonymized images and metadata from the S3 bucket

Components & Assets

  • Your system: deployed in a public/private cloud, on-the-edge or on-premise
  • AWS S3 bucket: Data storage in the Amazon cloud
  • Celantur Cloud Service: anonymizes images and delivers metadata

Considerations

Scalability: The Celantur Cloud Service, where images are anonymized, is scalable and built for large-scale data volumes. Depending on the amount of data to be transferred into the S3 bucket, the available bandwidth for up- and download needs to be considered. If your system is already hosted and stores data in AWS, this typically won’t be an issue.

Data Protection: Anonymized images will be irreversibly deleted by default after downloading them.

Remarks

Other cloud storage services can be integrated as well, please ask our team for more information.

Celantur on your system

Celantur is delivered as a Docker container. This container can either be used in your public/private cloud, on-the-edge or on-premise. Certain scenarios do not require a connection to the internet to run Celantur.

Batch and Stream Mode

Batch and Stream Mode enables the Celantur Container to anonymize images on the local file system.

Batch Mode: All images in the input directory will be anonymized. Celantur Container stops running after all images are anonymized.

Stream Mode: All images in the input directory will be anonymized. Celantur Container keeps running.

Architecture

  1. Your system is copying images into a specified input directory
  2. The Celantur Container is started via a command and anonymizes all the images in the directory
  3. Once the images have been anonymized, Celantur Container writes them into a specified output directory
  4. Optional: Celantur Container puts metadata for each image into the output directory as well
  5. Your system can fetch the anonymized images and metadata from the output directory

Components & Assets

  • Your system: deployed in a public/private cloud, on-the-edge or on-premise
  • Input directory: where you put the original images you want to be anonymized
  • Output directory: where the anonymized images and metadata will be written
  • Celantur Container: anonymizes images and creates metadata on command

Considerations

Scalability: Processing speed depends on the available hardware. For very large amounts of images it should be considered to run multiple instances of Celantur Container on dedicated hardware.

Data Protection: Images will remain fully under your control. No internet connection is required.

Fetch from Cloud Storage

Architecture

  1. Your system is storing images in a cloud storage (e.g. Microsoft Azure Storage) accessible with a REST API using a specific payload
  2. The Celantur Container fetches and returns images via the REST API
  3. Your system can use the anonymized images for further processing

Components & Assets

  • Your system: deployed in a public/private cloud, on-the-edge or on-premise
  • Your system provides a REST API, an endpoint that provides source and target image URIs, see payload in implementation details
  • Celantur Container: anonymizes images and creates metadata on command

Considerations

Scalability: For very large amounts of images it should be considered to run multiple instances of Celantur Container on dedicated hardware.

Data Protection: Images will remain fully under your control.

Implementation details

REST API payload, provided by your system. Celantur fetches the messages, anonymises the images from the input URL and puts them to the output URL.

{
    "messages": [
        {
            "input": "https://url-to-input-image/get-request",
            "output": "https://url-to-output-image/put-request",
            "outputRequiredHeaders": {
           
            },
            "completion": "https://trigger-if-success/put-request",
            "rejection": "https://trigger-if-failur/put-request"
        }
    ]
}

TCP Mode

Server Mode enables the Celantur Container to receive images either as an image or NumPy array via a TCP socket connection.

Architecture

  1. Your system is sending an image or NumPy array to the Celantur Container
  2. The Celantur Container anonymizes the images
  3. Your system is receiving it back as an image or NumPy array
  4. Your system can use the anonymized images for further processing

Components & Assets

  • Your system: deployed in a public/private cloud, on-the-edge or on-premise
  • Image: an image file loaded into memory
  • Numpy array: the NumPy array object representation of the image
  • Celantur Container: anonymizes images and creates metadata on command

Considerations

Scalability: For very large amounts of images it should be considered to run multiple instances of Celantur Container on dedicated hardware. Numpy arrays are bigger than usual images, but as they are used to move around data in applications, it’s faster to use this in-memory object than reading images from disk.

Data Protection: Images will remain fully under your control.

Implementation details