Bluree
Statement
Upload the pictures!
Detect human figures
Find the right person
Blur Effect
Add a sticker to it!

Bluree: Automated Privacy Protection for Photos

Inspiration

In today’s world, privacy is increasingly difficult to control. Whether at events, conferences, or public gatherings, people often opt out of photos—yet their images still end up online. Manually editing out individuals is slow and inconsistent, and current AI solutions are often used for surveillance rather than protection.

We wanted to flip the script and build a tool that respects privacy by automatically detecting and processing faces of opt-out individuals, ensuring that their images are either blurred or covered before being shared.

What it does

Bluree automates privacy protection in photos by:
1️⃣ Uploading two images – The photo to process and a reference image of the opt-out person.
2️⃣ Detecting all faces and people – Using ML to find and display bounding boxes around faces.
3️⃣ Verifying the opt-out individual – Matching them against detected faces and confirming accuracy.
4️⃣ Applying privacy protection – Either blurring the face or adding a sticker overlay.
5️⃣ Automating the process at events – Using Raspberry Pi cameras to capture and process photos in real time.

This ensures that privacy violations are prevented at the source, rather than fixed after the fact.

How we built it

🔹 Frontend: React.js for an intuitive and responsive UI.
🔹 Backend: Flask & Python to handle image processing and ML models.
🔹 Face Detection & Matching: OpenCV, face_recognition, and ultralytics for accurate face recognition.
🔹 Processing: OpenCV to blur or overlay stickers.
🔹 Hardware: Raspberry Pi with a camera module for real-time photo processing at events.

Challenges we ran into

🚧 Balancing speed & accuracy – Optimizing the face-matching pipeline to work in real time.
🚧 Protection of Identity – Different ways of processing the picture to protect individual, AI face replacement. 🚧 Hardware integration – Getting the Raspberry Pi to smoothly capture and process images.
🚧 User experience – Designing a simple, easy-to-use interface that makes verification seamless.

Accomplishments that we're proud of

🏆 Successfully built an AI-powered privacy tool that works automatically.
🏆 Integrated real-time processing with Raspberry Pi.
🏆 Achieved high accuracy in face matching.
🏆 Developed a smooth and intuitive UI that makes privacy control effortless.
🏆 Tackled an important ethical challenge in AI and image privacy.

What we learned

🔹 The importance of user validation steps to reduce AI errors.
🔹 Optimizing image processing for low-power hardware like Raspberry Pi.
🔹 Designing a project that balances automation and user control.
🔹 The impact of privacy protection in photography and media ethics.

What's next for Bluree: Automated Privacy Protection for Photos

🚀 Live event deployment – Partnering with conferences, protests, and workplaces to make privacy protection effortless.
🚀 Expanding to video – Applying the same principles to real-time video blurring for live-streaming and security footage.
🚀 API for event photographers – Allowing easy integration with existing event management platforms and hardware.
🚀 Enhancing accuracy – Using deep learning improvements for even more precise face matching, such as in various weather conditions.

Bluree isn’t just a hackathon project—it’s a step towards a future where AI protects privacy instead of violating it.