Indian-Western-Food-34 is an image classification vision-language encoder model fine-tuned from google/siglip2-base-patch16-224 for a single-label classification task. It is designed to classify food images into various Indian and Western dishes using the SiglipForImageClassification architecture.

Classification Report:
                precision    recall  f1-score   support

  Baked Potato     0.9912    0.9780    0.9846      1500
Crispy Chicken     0.9811    0.9707    0.9759      1500
         Donut     0.9893    0.9893    0.9893      1500
         Fries     0.9742    0.9827    0.9784      1500
       Hot Dog     0.9830    0.9735    0.9783      1548
      Sandwich     0.9898    0.9673    0.9784      1500
          Taco     0.9327    0.9427    0.9377      1500
       Taquito     0.9624    0.9387    0.9504      1500
     Apple Pie     0.9666    0.9540    0.9602      1000
        Burger     0.9114    0.9940    0.9509       331
   Butter Naan     0.9691    0.9186    0.9431       307
          Chai     0.9801    1.0000    0.9899       344
       Chapati     0.9188    0.9694    0.9435       327
    Cheesecake     0.9573    0.9640    0.9606      1000
 Chicken Curry     0.9610    0.9850    0.9728      1000
 Chole Bhature     0.9841    0.9867    0.9854       376
   Dal Makhani     0.9698    0.9797    0.9747       295
        Dhokla     0.9959    0.9959    0.9959       245
    Fried Rice     0.9485    1.0000    0.9736       350
     Ice Cream     0.9569    0.9770    0.9668      1000
          Idli     0.9934    1.0000    0.9967       302
        Jalebi     0.9931    1.0000    0.9965       288
  Kaathi Rolls     0.9640    0.9606    0.9623       279
  Kadai Paneer     0.9848    0.9731    0.9789       334
         Kulfi     0.9810    0.9673    0.9741       214
   Masala Dosa     0.9890    0.9890    0.9890       273
         Momos     0.9908    0.9969    0.9938       323
      Omelette     0.9829    0.9790    0.9810      1000
    Paani Puri     0.9281    0.9861    0.9562       144
        Pakode     0.9738    0.9665    0.9701       269
     Pav Bhaji     0.9901    0.9803    0.9852       305
         Pizza     0.9647    0.9927    0.9785       275
        Samosa     0.9878    0.9959    0.9918       244
         Sushi     0.9969    0.9800    0.9884      1000

      accuracy                         0.9729     23873
     macro avg     0.9719    0.9775    0.9745     23873
  weighted avg     0.9731    0.9729    0.9729     23873

The model categorizes images into 34 food classes:

Western Foods

Class 0: "Baked Potato"
Class 1: "Crispy Chicken"
Class 2: "Donut"
Class 3: "Fries"
Class 4: "Hot Dog"
Class 5: "Sandwich"
Class 6: "Taco"
Class 7: "Taquito"
Class 8: "Apple Pie"
Class 9: "Burger"
Class 13: "Cheesecake"
Class 18: "Fried Rice"
Class 19: "Ice Cream"
Class 27: "Omelette"
Class 31: "Pizza"
Class 33: "Sushi"

Indian Foods

Class 10: "Butter Naan"
Class 11: "Chai"
Class 12: "Chapati"
Class 14: "Chicken Curry"
Class 15: "Chole Bhature"
Class 16: "Dal Makhani"
Class 17: "Dhokla"
Class 20: "Idli"
Class 21: "Jalebi"
Class 22: "Kaathi Rolls"
Class 23: "Kadai Paneer"
Class 24: "Kulfi"
Class 25: "Masala Dosa"
Class 26: "Momos"
Class 28: "Paani Puri"
Class 29: "Pakode"
Class 30: "Pav Bhaji"
Class 32: "Samosa"

Run with Transformers🤗

!pip install -q transformers torch pillow gradio

python

import gradio as gr
from transformers import AutoImageProcessor
from transformers import SiglipForImageClassification
from transformers.image_utils import load_image
from PIL import Image
import torch

# Load model and processor
model_name = "prithivMLmods/Indian-Western-Food-34"
model = SiglipForImageClassification.from_pretrained(model_name)
processor = AutoImageProcessor.from_pretrained(model_name)

def food_classification(image):
    """Predicts the type of food in an image."""
    image = Image.fromarray(image).convert("RGB")
    inputs = processor(images=image, return_tensors="pt")
    
    with torch.no_grad():
        outputs = model(**inputs)
        logits = outputs.logits
        probs = torch.nn.functional.softmax(logits, dim=1).squeeze().tolist()
    
    labels = {
        "0": "Baked Potato", "1": "Crispy Chicken", "2": "Donut", "3": "Fries", 
        "4": "Hot Dog", "5": "Sandwich", "6": "Taco", "7": "Taquito", "8": "Apple Pie", 
        "9": "Burger", "10": "Butter Naan", "11": "Chai", "12": "Chapati", "13": "Cheesecake", 
        "14": "Chicken Curry", "15": "Chole Bhature", "16": "Dal Makhani", "17": "Dhokla", 
        "18": "Fried Rice", "19": "Ice Cream", "20": "Idli", "21": "Jalebi", "22": "Kaathi Rolls", 
        "23": "Kadai Paneer", "24": "Kulfi", "25": "Masala Dosa", "26": "Momos", "27": "Omelette", 
        "28": "Paani Puri", "29": "Pakode", "30": "Pav Bhaji", "31": "Pizza", "32": "Samosa", 
        "33": "Sushi"
    }
    predictions = {labels[str(i)]: round(probs[i], 3) for i in range(len(probs))}
    
    return predictions

# Create Gradio interface
iface = gr.Interface(
    fn=food_classification,
    inputs=gr.Image(type="numpy"),
    outputs=gr.Label(label="Prediction Scores"),
    title="Indian & Western Food Classification",
    description="Upload a food image to classify it into one of the 34 food types."
)

# Launch the app
if __name__ == "__main__":
    iface.launch()

Intended Use:

The Indian-Western-Food-34 model is designed to classify food images into Indian and Western dishes. Potential use cases include:

Restaurant & Food Delivery Apps: Enhancing food recognition for better menu recommendations.
Health & Nutrition Apps: Tracking calorie intake and diet preferences.
Food Blogging & Social Media: Auto-tagging food items in posts.
Educational Purposes: Teaching AI-based food classification.