How aurora-llin/sea-cuisine-nutrition-estimation works
This Streamlit application turns an uploaded food image into an editable ingredient list and a nutrition estimate. The main workflow classifies the dish, optionally detects ingredients, combines those with recipe ingredients for user correction, and calculates nutrition from Google Sheets data. The sampled code confirms the model, detector, and nutrition-data interfaces; README statements support the user-facing sequence. The app excerpt is partial, so details of omitted UI and calculation code are not inferred.
Main entry points: app.py Streamlit app · database.py Nutrition calculation · model_utils.py Dish classification · efficientnet-b0.pth EfficientNet weights
Components 9
User workflow
- Streamlit app
src/app.py - Ingredient review
src/app.py - Nutrition results
src/app.py
Image inference
- Dish classification
src/model_utils.py - EfficientNet weights
model/efficientnet-b0.pth - Ingredient detection
Nutrition data
- Nutrition calculation
src/database.py - Google Sheets
Other components
- User
Connections 9
- User → Streamlit app: uploads image
Readme.md - Streamlit app → Dish classification: classifies image
src/app.py - Dish classification → EfficientNet weights: loads weights
src/model_utils.py - Streamlit app → Ingredient detection: optionally detects
src/app.py - Streamlit app → Ingredient review: prepares ingredients
src/app.py - User → Ingredient review: confirms amounts
Readme.md - Ingredient review → Nutrition calculation: submits amounts
Readme.md - Nutrition calculation → Google Sheets: reads nutrition data
src/database.py - Nutrition calculation → Nutrition results: returns estimate
Readme.md