Inspiration
The inspiration for Elva came from the challenges faced by Alzheimer's patients and their caregivers. Memory loss, difficulty recognizing loved ones, and navigating daily life create emotional and practical hardships. We aimed to provide a solution that fosters independence, reduces anxiety, and offers peace of mind to caregivers by integrating cutting-edge AI with thoughtful design.
What it does
Elva is a comprehensive app designed to assist Alzheimer’s patients in their daily lives. Its Memory Reconstructor feature uses facial recognition to help users identify loved ones and recall relationships, reducing anxiety during social interactions. The app’s object recognition capability identifies everyday items and provides clear, step-by-step instructions on their usage. Additionally, Elva offers safe navigation features and quick access to emergency contacts to enhance user safety and provide peace of mind for caregivers.
How we built it
We integrated AuraFace Buffalo for facial recognition, leveraging the ResNet-100 architecture for accurate and efficient face identification. For object detection, we used DETR (DEtection TRansformer) to ensure robust recognition of items in real-world environments. The Natural Language Processing (NLP) capabilities were powered by T5, allowing us to generate clear, easy-to-follow instructions for object usage. The backend framework was built using Flask for its lightweight and modular structure, which hosted AI models and handled app logic. The frontend was designed using HTML, CSS, and JavaScript, ensuring accessibility for both Alzheimer’s patients and caregivers. To ensure data privacy, we incorporated encryption and local storage options, minimizing the risk of sensitive information being exposed.
Challenges we ran into
Balancing accuracy and efficiency in AI models was a significant challenge, as we needed to ensure functionality on low-end devices without compromising performance. Ethical data use and transparency also posed difficulties as we processed sensitive user data, requiring us to implement robust security measures. Another challenge was seamlessly integrating features like facial recognition, object detection, and NLP into one cohesive app, which requires detailed planning and teamwork.
Accomplishments that we're proud of
We are proud of creating a comprehensive solution that combines facial recognition, object detection, and NLP into a single app. The lightweight deployment of our app ensures that it can run effectively on lower-end devices, increasing its accessibility. Additionally, we maintained ethical AI practices by prioritizing user privacy, inclusivity, and transparency throughout the development process. The introduction of dynamic relationship mapping and personalized object instructions further showcases the app’s innovative approach to addressing the needs of Alzheimer’s patients.
What we learned
This project emphasized the importance of user-centered design, ensuring the app was both accessible and intuitive. It also taught us the value of ethical AI development, particularly when handling sensitive user data. Collaboration played a key role in integrating multiple advanced AI models into the app, and we learned how to design solutions with scalability in mind for local-based deployments.
What's next for Elva
The next steps for Elva include adding multilingual support using T5, which will make the app more inclusive for a diverse user base. We plan to integrate voice commands for easier interaction, enhancing accessibility further. Expanding the relationship mapping feature to include memory prompts and user-specific histories will provide more personalized assistance. We also aim to build a caregiver interface that allows for remote app management, such as updating emergency contacts or monitoring usage. Finally, we plan to explore compatibility with smartwatches and wearable devices to offer real-time support for users on the go.
Log in or sign up for Devpost to join the conversation.