NutriVoice: "See with AI, Eat with Confidence"

Inspiration

The inspiration for NutriVoice stemmed from a crucial question:
How can blind individuals easily access nutritional information about their food?

Many existing nutrition apps rely heavily on visual interfaces, making them inaccessible to visually impaired users. We aimed to bridge this gap by leveraging AI-powered speech and vision technologies, ensuring that everyone, regardless of ability, can make informed dietary choices.

What It Does

NutriVoice is an AI-powered nutrition assistant designed to help blind and visually impaired users track their daily calorie intake through voice commands and food recognition.

Installation & Setup

Follow these steps to set up and run the project locally.

Clone the Repository

• git clone https://github.com/birongliu/devfest.git

Navigate to the Project Directory

• Frontend:
  • cd frontend
• Backend:
  • cd inference

Install Dependencies

• Frontend:
  • npm install
• Backend:
  • pip install -r requirements.txt

Run the Project

• Frontend:
  • npm run start
• Backend:
  • python3 app.py

---

Install Python

• Install Python on your machine

Set Up Virtual Environment

• Windows:
  • python3 -m venv venv
• Linux/MacOS:
  • source venv/bin/activate

---

Tech Stack

NutriVoice integrates cutting-edge AI technologies to ensure seamless and accurate nutrition tracking:

The app allows users to:

• Set personal dietary goals based on weight, height, and timeframe.
• Scan food items using their phone's camera.
• Receive instant calorie estimations via AI-powered food recognition.
• Use voice commands for seamless interaction.
• Track daily calorie intake to stay on course with their nutrition goals.

Challenges We Ran Into

Building NutriVoice was not without its hurdles:

• AI Accuracy – Ensuring precise food recognition and calorie estimations required extensive testing and prompt engineering refinements.
• Speech Processing Latency – Optimizing Whisper API’s response time to provide real-time feedback was challenging.
• Data Constraints – Some food items lacked well-documented calorie data, requiring integration with multiple nutrition databases.
• User Accessibility Testing – Designing a voice-first experience for visually impaired users required iterative testing and feedback.

Accomplishments That We're Proud Of

• Developing a fully functional AI-powered nutrition assistant.
• Achieving high accuracy in food recognition and calorie estimation.
• Creating an intuitive voice-first experience that makes nutrition tracking accessible to visually impaired users.
• Successfully integrating speech recognition for seamless hands-free operation.

What We Learned

Throughout the development of NutriVoice, we gained valuable insights:

• Accessibility must be at the core of design – Optimizing all features for voice-based interactions was essential.
• AI models require continuous improvement – Fine-tuning food recognition and calorie estimation models was a major learning curve.
• User experience matters – Refining speech-to-text interactions to minimize errors and enhance usability was crucial.

What's Next for NutriVoice

NutriVoice is just the beginning! Future plans include:

• Personalized meal recommendations based on dietary preferences.
• Multi-language support to make the app accessible worldwide.
• Crowdsourced food database expansion to improve accuracy and include more local cuisines.
• Offline functionality to allow users to track nutrition without an internet connection.

With NutriVoice, we aim to redefine nutrition accessibility and empower individuals to take control of their health through AI-driven innovation.