InstaRizz

• 

  Example of a profile that is generated

• 

  Example RayBans

🌟 Inspiration

InstaRizz emerged from the idea of blending tech with social interactions in a way that’s both fun and innovative. We wanted to push the boundaries of wearable tech and AI to create a tool that offers real-time insights on the go, giving users an edge in social situations and a bit of extra charm.

📲 What it does

InstaRizz streams live video from Ray-Ban’s smart glasses to Instagram Live. As it streams, it captures snapshots of people’s faces in real time and runs facial recognition using OpenCV. Once a person is identified, the app searches for public information through a custom AI pipeline. InstaRizz then creates a short bio and generates three personalized pickup lines, displaying them to the user instantly, giving a boost in any social setting.

🛠️ How we built it

1.  Streaming Setup: We connected the Ray-Ban’s video input to stream directly to Instagram Live.
2.  Snapshot Processing: Using OpenCV, we periodically captured frames and conducted facial recognition.
3.  Reverse Identity Search: We identified individuals with a custom classifier trained on a database of consenting people who wanted the ability to be recognized.
4.  AI-Driven Personalization: With the person’s name, we ran a custom API on Magic Loops to pull relevant data using perplexity search. Claude then generated a short bio and three unique pickup lines.

This project brought together live streaming, facial recognition, machine learning, and natural language processing into a seamless experience.

🚧 Challenges we ran into

• Real-Time Processing: It was a challenge to sync the live video feed with real-time facial recognition.
• Database Training and Perplexity Search: Training an accurate classifier and refining perplexity search for relevant information took a lot of fine-tuning.
• Privacy Concerns: We carefully considered the ethical implications and ensured our process respected privacy.

🏆 Accomplishments that we’re proud of

We’re thrilled that we successfully created a fully functional pipeline that identifies individuals and generates personalized social insights in real time. Integrating diverse technologies and achieving seamless results was a big win for the team.

📚 What we learned

InstaRizz taught us about optimizing real-time data processing, handling ethical considerations in AI, and balancing speed with accuracy. We also honed our skills in facial recognition and language model applications. This was a few of our team member's first time working with computer vision

🚀 What’s next for InstaRizz

Next, we'll write up methods to protect users against potential privacy invasions such as getting scanned by random smart glasses in public. We are going to publish this information publicly on LinkedIn as we believe privacy is an important right in the upcoming dawn of the wearable computing age.

Built With

• magic-loops
• streamlit