Image Blurring Tool
Workflow Diagram
Task
The AI system blurs images of people in public areas to protect privacy; blurred images are then used for downstream public-service tasks.
Intent
The aims of the AI system are to maintain the privacy of individuals captured on cameras used to support public services and to enable multiple use cases on top of blurred imagery.
AI Workflow
Input
A vehicle captures photos of public areas; originals with people are not saved once sent to the AI system.
Process
Computer vision detects people and applies a blurring overlay.
Output
The blurred image is used as a privacy-preserving photo of a public space for other tasks; missed blurs can be flagged and the system retrained later.
Human Oversight Level
Autonomous AI
Institutional Oversight Examples
- •Ethics impact assessment (organization policy)
- •Bias assessment (organization policy)
- •Internal risk assessment (regulation)
- •Data protection (regulation)
- •Algorithm register for transparency (regulation)
Risk
Some risks of the AI system include potential for bias against certain groups; false positives (blur non-people) or false negatives (miss people).