AI for efficient data anonymisation
**Client Overview and Business Context**
**Client Profile:** A leading European company that provides extensive collections of infrastructure data, obtained through traditional methods, as well as navigation and aerial surveys.
**Client’s Data:** The only company in the market that effectively integrates and utilizes all available data collection techniques, backed by over 40 years of experience.
**Project:** Development of an AI-based object detection system, focused on automatic number plate and facial recognition, with provisions for the anonymization of private information.
**Challenges:**
Targeting municipalities, utility companies, construction firms, along with industrial and tourism sectors, our client faces the need to gather and anonymize substantial amounts of data, both private and public. As an engineering and surveying firm, and a technological pioneer in mobile road data collection, our client oversees the entire process. Consequently, the primary challenges for this project were related to:
**Results:**
Post-implementation, the algorithm achieved a 90% reduction in overall manual processing efforts. As a result, personnel transitioned from direct execution to review roles. Previously, the human-led anonymization process averaged 15 seconds per image; this time plummeted to just 0.3 seconds post-implementation, representing a 50-fold time reduction that positively affected both delivery timelines and overall processing costs. The client anticipated a minimum accuracy of 80% with a processing time of 1 second. We exceeded these expectations, achieving 92% accuracy within a 0.3-second processing timeframe. This project has set the stage for a future-ready solution, facilitating the potential for new integrations as required. As the project progresses, additional features will be introduced to enhance efficiency, data quality, and to meet a broader array of requirements.