app.py

import gradio as gr
import torch
from torchvision import transforms
from PIL import Image
import os
from typing import Tuple, List, Dict

def load_model() -> Tuple[torch.nn.Module, List[str]]:
    """
    Loads the model and class labels.
    Returns:
        model: The loaded PyTorch model.
        labels: List of class labels.
    """
    model_path = "skinconvnext_scripted.pt"
    labels_path = "labels.txt"
    
    if not os.path.exists(model_path):
        raise FileNotFoundError(f"Model not found: {model_path}")
    if not os.path.exists(labels_path):
        raise FileNotFoundError("File labels.txt not found.")
    
    device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
    model = torch.jit.load(model_path, map_location=device)
    model.eval()
    
    with open(labels_path, "r") as f:
        labels = [line.strip() for line in f.readlines()]
    
    return model, labels

model, labels = load_model()

# Define image preprocessing steps
preprocess = transforms.Compose([
    transforms.Resize((224, 224)),
    transforms.ToTensor(),
    transforms.Normalize(mean=[0.485, 0.456, 0.406],
                         std=[0.229, 0.224, 0.225]),
])

def predict(image: Image.Image) -> Dict[str, float]:
    """
    Makes a prediction for the given image.
    Args:
        image (PIL.Image): The input image.
    
    Returns:
        Dict[str, float]: A dictionary where keys are class names, and values are probabilities.
    """
    try:
        image = image.convert("RGB")
        image_tensor = preprocess(image).unsqueeze(0)
        device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
        image_tensor = image_tensor.to(device)
        model.to(device)
        
        with torch.no_grad():
            output = model(image_tensor)
            scores = torch.nn.functional.softmax(output[0], dim=0)
        
        predictions = {label: float(score) for label, score in zip(labels, scores)}
        sorted_predictions = dict(sorted(predictions.items(), key=lambda item: item[1], reverse=True))
        
        return sorted_predictions
    except Exception as e:
        return {"error": str(e)}

title = "🔥 Skin-AI"
description = (
    "🔬 **Skin-AI — AI-Powered Skin Disease Classification**\n\n"
    "This project utilizes a deep learning model to classify skin diseases based on an uploaded image.\n\n"
    "### 🚀 How to Use:\n\n"
    "1️⃣ Upload or take a photo of the affected skin area.\n\n"
    "2️⃣ Click the 'Submit' button.\n\n"
    "3️⃣ The app will return the probabilities for possible skin conditions.\n\n"
    "⚠️ **Important!** The results are for informational purposes only and do not constitute a medical diagnosis.\n\n"
    "### 🛠 Technologies Used:\n"
    "- PyTorch (Lightning)\n"
    "- Gradio\n"
    "- Hugging Face Spaces\n\n"
    "🔗 Source Code: [Hugging Face](https://huggingface.co/Eraly-ml/Skin-AI  )"
)

# Adding example images
examples = [
    ["example1.jpg"],
    ["example2.jpg"]
]

def update_submit_state(image):
    return gr.update(interactive=image is not None)

with gr.Blocks() as interface:
    gr.Markdown(title)
    gr.Markdown(description)
    
    with gr.Row():
        image_input = gr.Image(type="pil", label="Upload Image")
    
    submit_button = gr.Button("Submit", interactive=False)
    output_label = gr.Label(num_top_classes=3, label="Prediction")
    
    image_input.change(fn=update_submit_state, inputs=image_input, outputs=submit_button)
    submit_button.click(fn=predict, inputs=image_input, outputs=output_label)
    
    gr.Examples(examples, inputs=image_input)

# Для Docker/HF Spaces — указываем хост и порт явно
if __name__ == "__main__":
    interface.launch(server_name="0.0.0.0", server_port=7860)