Food Recognition and Calorie Estimation Model

A comprehensive deep learning system for food recognition, object detection, and calorie estimation using TensorFlow, YOLO, and EfficientNet.

Model Description

This model combines multiple deep learning approaches to provide accurate food recognition and calorie estimation:

• Food Classification: EfficientNet-B0 based multi-label classifier for 101 food categories
• Object Detection: YOLO v8 for detecting multiple food items in images
• Portion Size Estimation: Computer vision techniques for size estimation
• Calorie Calculation: Integration with USDA nutritional database

Model Performance

Usage

Basic Usage

from transformers import pipeline

# Load the model
classifier = pipeline("image-classification", model="BinhQuocNguyen/food-recognition-model")

# Analyze a food image
result = classifier("path/to/food_image.jpg")
print(f"Detected foods: {result}")

Advanced Usage

import torch
from transformers import AutoModel, AutoImageProcessor
from PIL import Image

# Load model and processor
model = AutoModel.from_pretrained("BinhQuocNguyen/food-recognition-model")
processor = AutoImageProcessor.from_pretrained("BinhQuocNguyen/food-recognition-model")

# Process image
image = Image.open("food_image.jpg")
inputs = processor(images=image, return_tensors="pt")

# Get predictions
with torch.no_grad():
    outputs = model(**inputs)
    predictions = torch.nn.functional.softmax(outputs.logits, dim=-1)

Training Data

The model was trained on:

• Food-101 Dataset: 101,000 images across 101 food categories
• Additional Datasets: Food-11, Recipe1M+ (where available)
• Data Augmentation: Rotation, flip, brightness, contrast adjustments

Nutritional Database

The model includes nutritional information for 101 food categories with:

• Calories per 100g
• Protein, carbohydrate, and fat content
• Portion size estimation capabilities

Limitations

• Accuracy may vary with image quality and lighting conditions
• Calorie estimates are approximate and should not replace professional dietary advice
• Model performance depends on food items being within the trained categories
• Portion size estimation is based on visual cues and may not be accurate for all cases

Citation

@misc{food-recognition-model,
  title={Food Recognition and Calorie Estimation Model},
  author={BinhQuocNguyen},
  year={2024},
  publisher={Hugging Face},
  howpublished={\url{https://huggingface.co/BinhQuocNguyen/food-recognition-model}}
}

License

This model is licensed under the MIT License.

Contact

For questions or issues, please contact:

• GitHub: Food Recognition Repository

Acknowledgments

• Food-101 dataset creators
• TensorFlow team
• Hugging Face team
• USDA Food Database
• OpenCV community