CompletedPythonFlaskOpenCV+4 more
Elva
Full-stack AI assistant that supports Alzheimer's patients with real-time facial recognition, object detection, and NLP-driven guidance. HackHive 2025 winning project.
Timeline
HackHive 2025
Role
AI/ML Developer
Team
Hackathon team
Status
CompletedTechnology Stack
Python
Flask
OpenCV
DETR
T5
Docker
Azure
Key Challenges
- Integrating multiple AI models into one unified system
- Real-time face and object recognition accuracy
- Handling sensitive user data with a privacy-focused design
Key Learnings
- Serving transformer models (DETR, T5) in production
- Containerized backend delivery with Flask and Docker
- Designing accessible, privacy-first AI products
Elva: AI Assistant for Alzheimer's Support
Overview
Elva is a full-stack AI application that assists individuals with Alzheimer's and dementia. It integrates facial recognition, object detection, and natural language processing to give patients real-time identification and guidance — helping them recognize the people and objects around them.
🏆 HackHive 2025 Winning Project (Ontario Tech University's annual hackathon).
Features
- Real-time facial recognition to identify family members and caregivers
- Object detection powered by DETR for everyday-item identification
- NLP-driven guidance using T5 for natural, helpful responses
- Containerized Flask backend services deployed with Docker on Azure
- Privacy-focused architecture for secure handling of sensitive user data
Impact
Elva won HackHive 2025, recognized for combining multiple AI models into a unified, privacy-conscious solution to a meaningful real-world healthcare problem.