import torch
import torch.nn as nn
import torchvision.transforms as transforms
import gradio as gr
from PIL import Image

class ConvModel(nn.Module):
    def __init__(self):
        super().__init__()
        self.cnn1 = nn.Sequential(
            nn.Conv2d(3, 16, kernel_size=3, padding=1),
            nn.ReLU(),
            nn.MaxPool2d(2)
        )
        self.cnn2 = nn.Sequential(
            nn.Conv2d(16, 32, kernel_size=3, padding=1),
            nn.ReLU(),
            nn.MaxPool2d(2)
        )
        self.fc = nn.Sequential(
            nn.Flatten(),
            nn.Linear(32 * 56 * 56, 2)
        )

    def forward(self, x):
        x = self.cnn1(x)
        x = self.cnn2(x)
        x = self.fc(x)
        return x

model = ConvModel()
model.load_state_dict(torch.load("conv_model.pth", map_location="cpu"))
model.eval()

class_names=['NORMAL', 'PNEUMONIA']

transform = 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(img):
    img = transform(img).unsqueeze(0)
    with torch.inference_mode():
        pred_probs = torch.softmax(model(img), dim=1)

    pred_labels_and_probs = {class_names[i]: float(pred_probs[0][i]) for i in range(len(class_names))}
    return pred_labels_and_probs


title = "Zatürre Bulucu"
description = "Gönderilen fotoğrafa göre Sağlıklı mı yoksa Zatürre mi olduğunu tahmin eder."

demo = gr.Interface(
    fn=predict,
    inputs=gr.Image(type="pil"),
    outputs=[gr.Label(num_top_classes=2, label="Predictions")],
    title=title,
    description=description
)

demo.launch(debug=False, share=True)