A curated list of resources for programmatically redacting personally identifiable information from data.
Often regulatory or ethical considerations frame the need to remove personally identifiable information to protect individual privacy. Modern technologies using machine learning are used to learn generalizable patterns that enable developers to perform this task programmatically.
The distribution of IoT/robotics hardware will exacerbate tensions around security & privacy if we don't treat it as a proper engineering constraint. With this collection of resources, we aim to lower the friction to implementing these techniques.
- NVIDIA DeepStream SDK Redaction
- Coral Dev Board PoseNet Anonymizer
- De-identify Medical Images with AWS
- De-identify Medical Images with GCP
- Deidentifcation by Image Segmentation & Blurring
- Deidentifcation by Image Segmentation & Blurring ROS Node
- Redact PII in Text
- How can we stop smart sensors from spying on us?
- Easy-to-use GDPR Guide for Data Scientist Part 2
- Sensitive Data in ML Datasets
- Custom NLP approaches to data anonymization