app.py

import torch
import numpy as np
import gradio as gr
from pathlib import Path
from PIL import Image
from torchvision import transforms
from huggingface_hub import hf_hub_download
from ResNet_for_CC import CC_model

# Define the Clothing1M class labels
CLOTHING1M_CLASSES = [
    "T-Shirt", "Shirt", "Knitwear", "Chiffon", "Sweater",
    "Hoodie", "Windbreaker", "Jacket", "Downcoat",
    "Suit", "Shawl", "Dress", "Vest", "Underwear"
]

# Initialize the model
model = CC_model()
model_path = hf_hub_download(repo_id="mohamdlog/CC", filename="CC_net.pt")
model.load_state_dict(torch.load(model_path, map_location='cpu'))
model.eval()

# Define preprocessing pipeline
def preprocess_image(image):
    if isinstance(image, np.ndarray):
        image = Image.fromarray(image)
    transform = transforms.Compose([
        transforms.Resize((224, 224)),
        transforms.ToTensor(),
    ])
    return transform(image).unsqueeze(0)

# Define classification function
def classify_image(image):
    input_tensor = preprocess_image(image)
    with torch.no_grad():
        output = model(input_tensor)
    
    # Get predicted class and confidence
    probabilities = torch.nn.functional.softmax(output, dim=1)
    predicted_class_idx = output.argmax(dim=1).item()
    predicted_class = CLOTHING1M_CLASSES[predicted_class_idx]
    confidence = probabilities[0][predicted_class_idx].item()
    
    return f"Category: {predicted_class}\nConfidence: {confidence:.2f}"

# Create Gradio interface
interface = gr.Interface(
    fn=classify_image, 
    inputs=gr.Image(label="Uploaded Image"), 
    outputs=gr.Text(label="Predicted Clothing"), 
    title="Clothing Category Classifier",
    description = """
	**Upload an image of clothing, and the model will predict its category.**  
	Try using an image that doesn't belong to any of the available categories, and see how the result differs!  
	**Categories:**  
	| T-Shirt | Shirt | Knitwear | Chiffon | Sweater | Hoodie | Windbreaker |  
	| Jacket | Downcoat | Suit | Shawl | Dress | Vest | Underwear |  
	""",
    examples=[[str(file)] for file in Path("examples").glob("*")],
    flagging_mode="never",
    theme="soft"
)

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