The Case: MyLED company attempted to tackle the problem of anonymous visual data analysis – how to gather useful data from video feeds without compromising privacy.

The Problem: With the increasing use of visual data, enterprises are looking for ways to gather valuable information about their clients or customers. However, privacy concerns are also on the rise, making it difficult for businesses to gather the information they need without breaking privacy laws or causing harm to their clients.

The Solution: QED Software has helped MyLED with building BlindBox – a technology implementation that allows for irreversible anonymization of visual data, such as images and video streams, while maintaining the ability to analyze it in order to extract information relevant to the user. The project included creating novel homomorphic encryption algorithms, as well as new methods for analysis of distorted images. This, combined, enables enterprises to obtain information about their clients without compromising their privacy.

Key Features:

  • Irreversible anonymization of visual data, making it impossible to interpret for people and standard computer image analysis tools
  • Ability to analyze anonymized data and extract relevant information
  • GDPR compliance.

The Technology: novel homomorphic encryption algorithms optimized for machine learning on video feeds, hardware implementation of the Field-Programmable Gate Array (FPGA) to anonymize the original data.

Conclusions: Thanks to BlindBox, businesses can now extract valuable information without compromising the privacy of their clients. The technology’s implementation in real-life conditions makes it a practical solution for businesses looking to gather information in a privacy-sensitive manner.