Saikatgh/food-calorie-estimation

🍽️ Food Calorie Estimation

An AI-powered food calorie estimation system that combines YOLOv8n-seg for food detection and segmentation with advanced depth estimation to calculate accurate calorie content.

🚀 Features

• —Food Detection & Segmentation: Uses YOLOv8n-seg to detect and segment food items
• —Volume Estimation: Advanced depth estimation for accurate volume calculation
• —Calorie Calculation: Estimates calories based on food type, volume, and nutritional density
• —Flutter API: RESTful endpoints optimized for Flutter mobile apps
• —Web Interface: Interactive Gradio interface for testing

🥘 Supported Food Items

The system currently supports South Asian/Bengali cuisine:

• —Beef curry
• —Biriyani
• —Chicken curry
• —Egg
• —Egg curry
• —Eggplants
• —Fish
• —Khichuri
• —Potato mash
• —Rice

📱 Flutter Integration

API Endpoints

1. Estimate Calories

POST /api/estimate_calories

Request Format:

{
  "image_base64": "data:image/jpeg;base64,/9j/4AAQSkZJRgABAQAAAQ...",
  "confidence": 0.5
}

Response Format:

{
  "success": true,
  "food_items_detected": 2,
  "nutrition_summary": {
    "total_calories": 450.2,
    "total_protein_g": 25.1,
    "total_carbohydrates_g": 45.3,
    "total_fat_g": 18.7,
    "total_weight_g": 320.5,
    "food_count": 2
  },
  "detailed_food_items": [
    {
      "class_name": "Rice",
      "confidence": 0.89,
      "calories": 195.0,
      "volume_ml": 150.0,
      "weight_g": 112.5
    }
  ],
  "api_version": "1.0",
  "timestamp": 1703123456
}

2. Health Check

GET /api/health

3. Supported Foods

GET /api/supported_foods

Flutter Code Example

import 'dart:convert';
import 'dart:io';
import 'package:http/http.dart' as http;

class CalorieEstimationAPI {
  static const String baseUrl = 'https://your-space-name.hf.space';
  
  static Future<Map<String, dynamic>> estimateCalories(File imageFile) async {
    try {
      // Convert image to base64
      List<int> imageBytes = await imageFile.readAsBytes();
      String base64Image = base64Encode(imageBytes);
      
      // Prepare request
      final response = await http.post(
        Uri.parse('$baseUrl/api/estimate_calories'),
        headers: {
          'Content-Type': 'application/json',
        },
        body: jsonEncode({
          'image_base64': 'data:image/jpeg;base64,$base64Image',
          'confidence': 0.5,
        }),
      );
      
      if (response.statusCode == 200) {
        return jsonDecode(response.body);
      } else {
        throw Exception('Failed to estimate calories: ${response.statusCode}');
      }
    } catch (e) {
      throw Exception('Error: $e');
    }
  }
}

🛠️ Deployment Instructions

1. Prepare Files for Hugging Face Spaces

1. 1.Copy these files to your Hugging Face Space:
2. 2.app.py (main application)
3. 3.requirements.txt (dependencies)
4. 4.README.md (this file)
5. 5.All your Python modules
6. 6.labels.txt

2. Create Hugging Face Space

1. 1.Go to Hugging Face Spaces
2. 2.Click "Create new Space"
3. 3.Choose:
4. 4.SDK: Gradio
5. 5.Hardware: CPU Basic (or GPU if needed)
6. 6.Visibility: Public

3. Upload Files

Upload all the files to your space repository.

4. Configure Space

The space will automatically build and deploy. You can access:

• —Web Interface: https://your-username-space-name.hf.space
• —API Endpoints: https://your-username-space-name.hf.space/api/...

🔧 Local Development

# Install dependencies
pip install -r requirements.txt

# Run the app
python app.py

📊 System Architecture

Flutter App → HTTP Request → Hugging Face Space → YOLOv8n-seg → Depth Estimation → Calorie Calculation → JSON Response → Flutter App

🎯 Usage Examples

Web Interface

1. 1.Upload a food image
2. 2.Adjust confidence threshold
3. 3.Click "Analyze Food"
4. 4.View results and visualization

Flutter Integration

1. 1.Capture/select food image
2. 2.Send to API endpoint
3. 3.Parse JSON response
4. 4.Display nutrition information

🤝 Contributing

1. 1.Fork the repository
2. 2.Create a feature branch
3. 3.Make your changes
4. 4.Submit a pull request

📄 License

This project is licensed under the MIT License.

🙏 Acknowledgments

• —Ultralytics YOLOv8 for object detection
• —Gradio for the web interface
• —Hugging Face Spaces for hosting