---
license: mit
tags:
- food-recognition
- computer-vision
- calorie-estimation
- efficientnet
- nutrition
- health
pipeline_tag: image-classification
---

# 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

```python
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

```python
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

```bibtex
@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](https://github.com/BinhQuocNguyen/food-recognition-model)

## Acknowledgments

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