🤖 AI Food Recognition
Nutrivio's AI food recognition is powered by advanced machine learning models that can identify over 10,000 different foods, dishes, and ingredients from photos with remarkable accuracy.
How It Works
Our AI uses a combination of computer vision technologies:
- Deep Learning Neural Networks: Trained on millions of food images
- Multi-Scale Recognition: Identifies individual ingredients in complex dishes
- Portion Estimation: Calculates serving sizes using visual cues
- Nutritional Mapping: Links recognized foods to comprehensive nutrition database
- Continuous Learning: Improves accuracy through user feedback
Recognition Accuracy by Food Type
| Food Category | Recognition Accuracy | Examples |
|---|---|---|
| Fruits & Vegetables |
90-95%
|
Apple, broccoli, banana, carrot |
| Packaged Foods |
85-90%
|
Cereal, bread, yogurt, chips |
| Prepared Meals |
75-85%
|
Pasta, salad, sandwich, pizza |
| Beverages |
80-90%
|
Coffee, juice, soda, smoothie |
| Ethnic Cuisine |
70-80%
|
Sushi, curry, tacos, stir-fry |
| Homemade/Complex |
60-75%
|
Casseroles, soups, mixed dishes |
Best Practices for Photo Capture
- Lighting: Use natural light when possible, avoid harsh shadows
- Angle: Take photos from directly above (bird's eye view) or at a 45-degree angle
- Distance: Fill 70-80% of the frame with food
- Focus: Ensure the entire meal is in sharp focus
- Background: Use a simple, contrasting background (white plate on dark table)
- Multiple Items: For complex plates, consider separate photos of individual components
Supported Food Categories
Fruits, vegetables, herbs
Meat, fish, poultry, eggs
Rice, pasta, cereals, baked goods
Milk, cheese, yogurt, butter
Almonds, walnuts, chia, flax
Pizza, burgers, fries, sandwiches
Asian, Mexican, Italian, Indian
Coffee, tea, juice, smoothies
Cakes, cookies, ice cream, candy
Salads, bowls, meal kits
How to Use AI Recognition
- Open the Camera: Tap the camera icon or "Add Meal" button
- Frame Your Food: Position your meal in the camera viewfinder
- Capture: Tap the capture button or volume button
- AI Processing: Wait 2-5 seconds for analysis
- Review Results: Check identified foods and confidence scores
- Confirm or Edit: Verify accuracy and adjust as needed
- Set Portions: Adjust serving sizes using sliders or input fields
- Save to Diary: Add the meal to your food log
Understanding AI Results
Confidence Scores
Each recognized food item comes with a confidence score:
- 90-100%: Very high confidence - likely accurate
- 70-89%: High confidence - usually correct
- 50-69%: Medium confidence - review carefully
- Below 50%: Low confidence - manual verification recommended
Multiple Suggestions
The AI often provides multiple possibilities for unclear items:
Portion Size Estimation
The AI estimates portion sizes using visual reference points:
- Plate size: Standard dinner plate (10-11 inches) as reference
- Hand comparison: Palm, fist, and finger sizes for scaling
- Common objects: Utensils, cups, and napkins for scale
- Food density: Accounts for thickness and layering
Manual Override Options
When AI recognition isn't perfect, you have several options:
- Search Database: Manually search for the correct food
- Update Food Name: Change the name of existing foods will re-run the analysis
- Edit Portions: Adjust serving sizes manually
- Split Recognition: Separate incorrectly combined foods
- Add Missing Items: Include foods the AI missed
Advanced Features
Barcode Scanning
For packaged foods, the app also supports barcode scanning for 100% accuracy.
DB Search
Manually search our extensive food database for accurate matches.
Food Description Analysis
Enter or say a description of your food item and its weight for more accurate recognition.