README.md

metadata

title: 'NutriDetect: AI Nutrition Companion'
emoji: 🍽️
colorFrom: red
colorTo: green
sdk: streamlit
sdk_version: 1.44.1
app_file: app.py
pinned: false
short_description: AI-powered Food Detection and Nutritional Analysis

🍽️ NutriDetect – AI Nutrition Companion

NutriDetect is a cutting-edge food detection and nutrition analysis system powered by YOLOv8, designed to assist users in identifying food items from images and retrieving detailed nutritional information. This project has been accepted and presented at an ACM Conference for its innovation in AI-powered dietary monitoring and digital health.

---

🧠 Key Features

• 🔍 Real-time food detection using YOLOv8
• 🌎 Trained on a diverse, culturally adaptive dataset (UECFOOD256, OID, Vietnamese Food)
• 📊 Nutrition analysis for recognized food items via integrated JSON database
• ⚙️ Built with Streamlit – runs locally and on Hugging Face Spaces

---

📂 Dataset Highlights

NutriDetect uses a highly curated dataset combining:

• UECFOOD256 (31k images)
• Open Images Dataset (OID) – 20k+ food-tagged images
• School Lunch & Vietnamese Datasets – 4k+ culturally specific images

Total: ~55,000 annotated images across 345 food classes

---

🧪 Model Performance (YOLOv8)

Metric	Score (%)
Precision	72.4
Recall	69.5
mAP@50	75.9
mAP@50–95	60.1

---

📄 Research Paper (ACM Conference)

This project was accepted at an ACM conference.

---

🚀 Quick Start (for Local Deployment)

git clone https://huggingface.co/spaces/Thirupathi986/Nutridetect
cd Nutridetect
pip install -r requirements.txt
streamlit run app.py