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

# --- Define the MLP_one CNN architecture ---
class MLP_one(nn.Module):
    def __init__(self):
        super().__init__()
        self.conv1 = nn.Conv2d(3, 6, 5)
        self.pool = nn.MaxPool2d(2, 2)
        self.conv2 = nn.Conv2d(6, 16, 5)
        self.fc1 = nn.Linear(16 * 5 * 5, 120)
        self.fc2 = nn.Linear(120, 84)
        self.fc3 = nn.Linear(84, 10)

    def forward(self, x):
        x = self.pool(F.relu(self.conv1(x)))
        x = self.pool(F.relu(self.conv2(x)))
        x = torch.flatten(x, 1)
        x = F.relu(self.fc1(x))
        x = F.relu(self.fc2(x))
        x = self.fc3(x)
        return x


# --- Load trained model weights ---
model = MLP_one()
model.load_state_dict(torch.load("model.pth", map_location="cpu"))
model.eval()

# --- CIFAR-10 class names ---
classes = [
    "airplane", "automobile", "bird", "cat", "deer",
    "dog", "frog", "horse", "ship", "truck"
]

# --- Transform pipeline ---
transform = transforms.Compose([
    transforms.Resize((32, 32)),  # resize any image to 32x32
    transforms.ToTensor(),
    transforms.Normalize((0.5, 0.5, 0.5), (0.5, 0.5, 0.5))
])

# --- Prediction function ---
def predict(image):
    image = image.convert("RGB")
    x = transform(image).unsqueeze(0)  # (1, 3, 32, 32)
    with torch.no_grad():
        outputs = model(x)              # tensor shape [1, 10]
        probs = torch.nn.functional.softmax(outputs, dim=1)  # apply softmax
        probs = probs[0].cpu().numpy()   # convert to numpy for Gradio
    return {classes[i]: float(probs[i]) for i in range(10)}


# --- Gradio Interface ---
demo = gr.Interface(
    fn=predict,
    inputs=gr.Image(type="pil", label="Upload any image"),
    outputs=gr.Label(num_top_classes=3),
    title="CIFAR-10 Image Classifier (MLP_one)",
    description=(
        "Upload any image (JPG, PNG, etc.) and this model will resize it to 32×32 "
        "and predict the closest CIFAR-10 class."
    )
)

demo.launch()