FruitNutritionDetector

FruitNutritionDetector is a powerful FastAPI-based API designed to detect fruits from images and retrieve their nutritional information using ImageAI and the USDA API. This API utilizes advanced models like Xception for image classification and fetches nutritional data from the USDA FoodData Central database.

Key Features

• Fruit Detection API: Detect fruits from images with high accuracy using the ImageAI framework.
• Nutritional Information Retrieval: Fetch detailed nutritional data from the USDA API for the detected fruit.
• FastAPI Framework: Built with FastAPI, providing high performance and easy deployment.
• YOLO Integration: Potential for advanced object detection with YOLO models.

Getting Started

Installation

Follow these steps to set up the FruitNutritionDetector API:

1. Clone the repository:

   git clone https://github.com/your-username/FruitNutritionDetector.git
   cd FruitNutritionDetector
   

2. Install dependencies:

   Make sure you have Python 3.7+ installed, then install the required packages:

   pip install fastapi requests pyngrok uvicorn
   pip install cython 'pillow>=7.0.0' 'numpy>=1.18.1' 'opencv-python>=4.1.2' 'torch>=1.9.0' --extra-index-url 'https://download.pytorch.org/whl/cpu' 'torchvision>=0.10.0' --extra-index-url     'https://download.pytorch.org/whl/cpu' 'pytest==7.1.3' 'tqdm==4.64.1' 'scipy>=1.7.3' 'matplotlib>=3.4.3' 'mock==4.0.3'
   

3. Download pre-trained models:

   wget https://github.com/OlafenwaMoses/ImageAI/releases/download/3.0.0-pretrained/resnet50-19c8e357.pth/ -O resnet50-19c8e357.pth
   wget https://github.com/vishalbharti1990/deep_feature_clustering/blob/master/Documents/.keras/models/xception_weights_tf_dim_ordering_tf_kernels.h5?raw=true -O xception_weights_tf_dim_ordering_tf_kernels.h5
   

4. Set up USDA API Key:

   Replace {USDA_API_KEY} in the code with your actual USDA API key.

5. Launch the API:

   Start the server with Uvicorn:

   uvicorn main:app --host 0.0.0.0 --port 5000
   The API will be accessible at http://localhost:5000.
   
   

6. API Usage

   Endpoint: POST /fruit-detection This endpoint detects a fruit from an uploaded image and retrieves its nutritional information.

   Request Parameters:

   • file: The image file containing the fruit (Supported formats: .jpg, .png, .jpeg).

     curl -X POST "http://localhost:5000/fruit-detection" \
       -H "accept: application/json" \
       -H "Content-Type: multipart/form-data" \
       -F "file=@/path/to/fruit_image.jpg"
     

   Response:

   • detected_fruit: The name of the detected fruit.
   • probability: The probability score of the detection.
   • nutrition_info: A dictionary containing the nutritional information retrieved from the USDA API.

       {
        "detected_fruit": "Apple",
        "probability": 97.5,
        "nutrition_info": {
          "name": "Apple",
          "nutrients": {
            "Energy": 52,
            "Protein": 0.26,
            "Total lipid (fat)": 0.17,
            "Carbohydrate, by difference": 13.81
          }
        }
      }
     

Contributing

We welcome contributions to FruitNutritionDetector! Feel free to fork the repository, create a branch, and submit a Pull Request with your improvements or new features.