BinhQuocNguyen/food-recognition-vit

Food Recognition Model

A Vision Transformer (ViT) fine-tuned for food recognition and classification. This model can identify 10 different types of food from images.

Model Description

This model is based on Google's Vision Transformer (ViT-Base) and has been fine-tuned on a custom food dataset. It can classify images into 10 different food categories with high accuracy.

Food Classes

The model can recognize the following food types:

• apple_pie
• caesar_salad
• chocolate_cake
• cup_cakes
• donuts
• hamburger
• ice_cream
• pancakes
• pizza
• waffles

Model Performance

• Accuracy: 68.0%
• F1 Score: 66.5%
• Precision: 68.2%
• Recall: 68.0%

Usage

Using the Pipeline

from transformers import pipeline

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

# Classify an image
result = classifier("path/to/your/food_image.jpg")
print(result)

Using the Model Directly

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

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

# Load and process image
image = Image.open("path/to/your/food_image.jpg")
inputs = processor(image, return_tensors="pt")

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

# Get top prediction
predicted_class_id = predictions.argmax().item()
predicted_class = model.config.id2label[str(predicted_class_id)]
confidence = predictions[0][predicted_class_id].item()

print(f"Predicted: {predicted_class} ({confidence:.3f})")

Training Details

• Base Model: google/vit-base-patch16-224
• Training Framework: PyTorch with Transformers
• Dataset: Custom food recognition dataset
• Classes: 10 food categories
• Image Size: 224x224 pixels
• Training Time: ~84 minutes

Limitations

• The model is trained on a specific set of food categories and may not generalize well to other food types
• Performance may vary depending on image quality, lighting, and angle
• The model works best with clear, well-lit images of food

Citation

If you use this model in your research, please cite:

@misc{food-recognition-model,
  title={Food Recognition Model},
  author={BinhQuocNguyen},
  year={2025},
  publisher={Hugging Face},
  howpublished={\url{https://huggingface.co/BinhQuocNguyen/food-recognition-vit}}
}

License

This model is released under the MIT License.