Update app.py

app.py

@@ -3,58 +3,54 @@ import torch
 from torchvision import transforms
 from PIL import Image
 from model import load_model
-import os
 
 device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
 
 models_cache = {}
 
 class_names = [
-    'Alzheimer Disease', 
-    'Mild Alzheimer Risk', 
+    'Alzheimer Disease',
+    'Mild Alzheimer Risk',
     'Moderate Alzheimer Risk',
-    'Very Mild Alzheimer Risk', 
-    'No Risk', 
+    'Very Mild Alzheimer Risk',
+    'No Risk',
     'Parkinson Disease'
 ]
 
 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])
-    ])
+    transforms.Resize((448, 448)),
+    transforms.ToTensor(),
+    transforms.Normalize(mean=[0.485, 0.456, 0.406],
+                         std=[0.229, 0.224, 0.225])
+])
 
-# Kullanıcının seçtiği modele göre tahmin yap
-def predict(model_name, image):
-    if model_name not in models_cache:
-        model = load_model(model_name, device)
-        models_cache[model_name] = model
-    else:
-        model = models_cache[model_name]
+def predict(version: str, image: Image.Image):
+    if version not in models_cache:
+        models_cache[version] = load_model(version, device)
+    model = models_cache[version]
 
-    image = image.convert("RGB")
-    tensor = transform(image).unsqueeze(0).to(device)
+    img = image.convert("RGB")
+    tensor = transform(img).unsqueeze(0).to(device)
 
     with torch.no_grad():
         outputs = model(tensor)
         probs = torch.nn.functional.softmax(outputs, dim=1)[0]
-        predicted = torch.argmax(probs).item()
 
     return {class_names[i]: float(probs[i]) for i in range(len(class_names))}
 
-
 model_options = [
     "Vbai-DPA 2.1f",
     "Vbai-DPA 2.1c",
     "Vbai-DPA 2.1q"
 ]
 
+# Gradio arayüzü
 with gr.Blocks() as demo:
-    gr.Markdown("Dementia and Parkinson Diagnosis 2.1 🧠")
-    gr.Markdown("Select model version and upload an MRI or fMRI image.")
+    gr.Markdown("# Dementia and Parkinson Diagnosis 2.1 🧠")
+    gr.Markdown("Select the model you want to choose (f, c or q) and upload a 2D brain slice.")
 
     with gr.Row():
-        model_selector = gr.Dropdown(choices=model_options, label="Model Version", value="Vbai-DPA 2.1f")
+        version_selector = gr.Radio(choices=model_options, value='f', label="Model Version")
         image_input = gr.Image(type="pil", label="MRI/fMRI Image")
     output_label = gr.Label(num_top_classes=3, label="Top Predictions")