HabitAware

Inspiration

We all have subconscious habits such as nail-biting, hair-pulling, and face-touching that creep in during moments of stress, anxiety, or boredom. As someone who personally struggled with these behaviors, I wanted to create something that could help others build awareness and take control. That’s how HabitAware was born, an AI-powered assistant that catches these habits in real time and offers support, not shame.

What it does

HabitAware detects stress-linked hand-to-face behaviors (like nail-biting and hair-pulling and more) using computer vision and ML. It tracks stress duration, behavior frequency, and last stressed, then uses OpenAI’s API to generate personalized coaching messages. A clean Streamlit dashboard shows progress, helping users break unconscious cycles and build healthier habits.

How we built it

MediaPipe + OpenCV: For real-time landmark detection of hands and face

Python: Core logic for behavior detection, event tracking, and state updates

Streamlit: UI for stats and habit history

OpenAI API: For generating personalized, supportive messages based on user behavior

Challenges we ran into

Integrating real-time AI responses Deciding where AI adds the most value, we wanted it to coach, not just comment Keeping the UI simple but useful within Streamlit’s limits

Accomplishments that we're proud of

Built a fully working system that detects real behaviors and responds in real-time

Successfully integrated OpenAI to personalize user experience

Created a stress-aware state tracking system that goes beyond basic detection

What we learned

AI is most powerful when it feels personal and purposeful . Learned how to work with computer vision for the first time, integrating it with something we struggle with