Current Authors & Contributors:

• Sam Patankar: Campaign Lab Scotland (Github: sampatankar, LinkedIn: sampatankar)

The Project:

• Ideation:

Image database

Build an ethical data management platform that addresses the growing need for permission-based photo usage in political campaigns. This challenge focuses on technical implementation while prioritising privacy, consent, and data ethics. Political campaigns regularly share photos from canvassing activities on social media (e.g., using hashtags like #Labourdoorstep). These posts often contain images of volunteers and members of the public who may not have explicitly consented to their images being stored or shared online. The mission is to create a tool that helps campaigns manage this data ethically.

Phase 1: Image Processing:

• Implement facial recognition capabilities to identify unique individuals across multiple photos
• Create a system to blur faces by default in any public-facing interfaces
• Develop an algorithm to match submitted selfies against the database of scraped images

Phase 2: Security & Ethics:

• Implement robust security measures to protect the facial recognition database
• Create time-limited data retention policies
• Develop transparent documentation about how data is used
• Implement anonymization for analytics
• Technical Stack Suggestions
• Web scraping: Python with BeautifulSoup/Scrapy
• Image processing: OpenCV, TensorFlow for facial recognition
• Location extraction: NLP libraries like spaCy or NLTK
• Backend: Flask/Django/Node.js
• Database: PostgreSQL with proper encryption
• Frontend: React with Material UI components
• Authentication: OAuth 2.0 implementation

Phase 3: Data Collection:

• Develop a scraper to collect tweets/posts containing the target hashtag
• Extract all images and accompanying text from these posts
• Implement text analysis to identify location information mentioned in the posts
• Store this information in a structured database

Phase 4: Permission Management:

• Build a user-friendly frontend where individuals can:
• Upload a selfie to search for their appearance in campaign photos
• View all instances where they appear in the database
• Grant or revoke permission for specific images
• Request image removal from public platforms
• Create an admin interface for campaign staff to:
• Review permission statuses
• Generate reports on consent management
• Automate removal requests for declined photos

These phases are liable to change as we need to think about the MVP - separate document.

This work is subject to the license conditions in license.txt