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

Metric	Value
Classification Accuracy	>85%
Object Detection mAP	>0.75
Calorie Estimation Accuracy	±20%
Inference Speed	<2 seconds/image

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

Downloads last month: 363