app.py

from PIL import Image
import torch
import torchvision.transforms as transforms
from torchvision import models
import gradio as gr
from rembg import remove  # 🆕 Background removal

# 🔧 Set device
device = torch.device("cuda" if torch.cuda.is_available() else "cpu")

# 📦 Load your fine-tuned model
model = models.resnet50(pretrained=False)
model.fc = torch.nn.Linear(model.fc.in_features, 2)  # 2 classes: Edible, Poisonous
model.load_state_dict(torch.load("resnet_mushroom_classifier.pth", map_location=device))
model = model.to(device)
model.eval()

# 🏷️ Class names
class_names = ['Edible', 'Poisonous']

# 🍄 Mapping for more detailed species
mushroom_species = {
    "Edible": "Possible species:\n• Amanita citrina\n• Russula delica\n• Phaeogyroporus portentosus",
    "Poisonous": "Possible species:\n• Amanita phalloides\n• Inocybe rimosa"
}

# 🎨 Image preprocessing (must match training)
transform = transforms.Compose([
    transforms.Resize((224, 224)),
    transforms.ToTensor(),
    transforms.Normalize([0.485, 0.456, 0.406],
                         [0.229, 0.224, 0.225])
])

# 🧠 Prediction function with background removal
def classify_mushroom(image: Image.Image):
    try:
        image = image.convert("RGB")

        # 🆕 Remove background
        image = remove(image)  # returns RGBA with transparency
        image = image.convert("RGB")  # back to RGB

        tensor = transform(image).unsqueeze(0).to(device)

        with torch.no_grad():
            outputs = model(tensor)
            _, predicted = torch.max(outputs, 1)
            label = class_names[predicted.item()]
            score = torch.softmax(outputs, dim=1)[0][predicted.item()].item() * 100

        suggestion = mushroom_species[label]

        return label, "กินได้" if label == "Edible" else "พิษ", f"{score:.2f}%", suggestion

    except Exception as e:
        print(f"❌ Error: {e}")
        return "Error", "ผิดพลาด", "N/A", "N/A"

# 🎛️ Gradio UI
if __name__ == "__main__":
    with gr.Blocks() as demo:
        gr.Markdown("## 🍄 Mushroom Safety Classifier")
        gr.Markdown("Upload a mushroom photo to check if it’s edible or poisonous.\nอัปโหลดรูปเห็ดเพื่อทำนายว่าเห็ดกินได้หรือมีพิษ")

        with gr.Row():
            image_input = gr.Image(type="pil", label="📷 Upload Mushroom Image")
            with gr.Column():
                label_en = gr.Textbox(label="🧠 Prediction (English)")
                label_th = gr.Textbox(label="🗣️ คำทำนาย (ภาษาไทย)")
                confidence = gr.Textbox(label="📶 Confidence Score")
                label_hint = gr.Textbox(label="🏷️ Likely Species (Based on Training Data)")

        classify_btn = gr.Button("🔍 Classify")

        classify_btn.click(
            fn=classify_mushroom,
            inputs=image_input,
            outputs=[label_en, label_th, confidence, label_hint]
        )

    demo.launch()