Inspiration
The inspiration for H.A.L.O came from a trip I took to India, where I witnessed firsthand how challenging it can be for visually impaired individuals to identify who is at their front door. After hearing a quote from Srikanth Bolla, a blind entrepreneur who emphasized empowering rather than pitying those with disabilities, I was inspired to take action. With recent break-ins in my own neighborhood and the growing popularity of smart doorbells like Ring, I realized there was a real opportunity to build a more accessible, personalized solution.
What it does
H.A.L.O is a smart security system that uses facial recognition to identify people at the door and sends SMS notifications saying exactly who is there—for example, "Bob is at your front door." If the system doesn’t recognize the person, it simply says "Unknown is at your front door." Users can upload images of known people through a simple app interface, allowing for personalized and accessible security.
How we built it
We built H.A.L.O using a Raspberry Pi paired with a camera module to capture live images. The backend uses Python with OpenCV for face recognition and Flask to serve API endpoints. We integrated Firebase to store image data and user details, and connected the system to Twilio for sending real-time SMS alerts.
On the frontend, we used FlutterFlow to develop a mobile-friendly interface that allows users to upload images directly and view a live log of visitors. All communication between the Pi and the app happens via our REST API.
Challenges we ran into
Facial recognition accuracy: Balancing performance and speed on a Raspberry Pi required careful optimization and testing.
SMS delivery integration: Configuring Twilio with Firebase-authenticated phone numbers was tricky, especially handling permission-denied errors.
User image uploads: We faced issues configuring Firebase Storage in FlutterFlow for image uploads, which required debugging CORS and storage rules.
Hackathon time crunch: Trying to deploy a working system, web dashboard, and app within the hackathon’s limited time was a major challenge—but a rewarding one.
Accomplishments that we're proud of
Built a working face recognition + SMS system end-to-end during the hackathon.
Created an accessible app interface that allows users to easily upload faces for recognition.
Made a system that genuinely addresses a real-world accessibility issue, especially for the visually impaired.
we learned
How to optimize face recognition on low-powered devices like Raspberry Pi.
How to set up and debug Firebase integration with FlutterFlow.
How to design RESTful APIs for real-time communication between hardware and cloud.
The importance of designing with empathy, especially when building for accessibility.
What's next for H.A.L.O
Expand user roles (e.g., admins and family members) within the app.
Train a more robust face recognition model that works across lighting conditions.
Integrate motion detection and allow cloud video storage for better security logging.
Explore deploying the system commercially or open-sourcing it for community use.
Log in or sign up for Devpost to join the conversation.