Advanced level · Module 5: Responsible Deep Learning · Lesson 2 of 4
Consent and Privacy for Images, Voice and Text
Collect and use personal media responsibly, with consent, minimisation and protection.
In this lesson you will
- Apply consent and minimisation to media data
- Use privacy-preserving techniques where appropriate
Key concepts
informed consentfaces and voices as biometric datablurring and minimisationon-device processing and federated learning
The lesson
Images, voice recordings and messages often reveal identity. Faces and voices can be biometric data, which deserves strong protection.
Collect only what the task needs, with informed consent; blur faces and number plates; remove metadata such as GPS from photos; store data securely and delete it when no longer needed.
On-device processing keeps data on users' phones. Federated learning trains on devices and shares only model updates, though it is complex and not a complete privacy guarantee.
Step by step
- Write a consent statement for collecting field photos.
- Strip metadata from sample photos.
- Blur faces automatically.
- Plan storage and deletion.
Practice activity
Write a data handling plan for the capstone's data collection.
Expected result: A plan covering consent, minimisation, protection and deletion.
Check your understanding
-
Why remove metadata from photos?
Show answer
It can include GPS location and device details.
-
Does federated learning guarantee privacy?
Show answer
No; it reduces data sharing but is not a complete guarantee.
Taught live by an instructor — online or in person. Free course lessons are for self-study and do not lead to a certificate — see registration options.