Update app.py

app.py

@@ -9,20 +9,16 @@ import numpy as np
 from thop import profile
 import io
 
-# Device selection
 device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
 
-# Cache models to avoid repeated downloads
 models_cache = {}
 
-# Preprocess transform for 224x224 input
 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])
 ])
 
-# Class names
 class_names = [
     'Alzheimer Disease',
     'Mild Alzheimer Risk',
@@ -32,19 +28,16 @@ class_names = [
     'Parkinson Disease'
 ]
 
-# Performance metrics calculation outside predict to not block UI
 def calculate_performance(model):
     model.eval()
     dummy = torch.randn(1,3,224,224).to(device)
     flops, params = profile(model, inputs=(dummy,), verbose=False)
     params_m = round(params/1e6,2)
     flops_b = round(flops/1e9,2)
-    # inference timing on CPU
     import time
     start = time.time()
     _ = model(dummy.cpu())
     cpu_ms = round((time.time() - start)*1000,2)
-    # inference timing on GPU if available
     if device.type == 'cuda':
         start_event = torch.cuda.Event(enable_timing=True)
         end_event = torch.cuda.Event(enable_timing=True)
@@ -57,30 +50,24 @@ def calculate_performance(model):
         gpu_ms = None
     return {'params_million':params_m, 'flops_billion':flops_b, 'cpu_ms':cpu_ms, 'gpu_ms':gpu_ms}
 
-# Prediction function
 def predict_and_monitor(version, image):
     try:
-        # load or get cached model
         if version not in models_cache:
             models_cache[version] = load_model(version, device)
         model = models_cache[version]
 
-        # preprocess
         if image is None:
             raise gr.Error("Görsel yüklenmedi.")
         img = image.convert("RGB")
         tensor = transform(img).unsqueeze(0).to(device)
 
-        # inference
         with torch.no_grad():
             logits = model(tensor)
             probs = F.softmax(logits, dim=1)[0]
 
-        # prepare outputs
         pred_dict = {class_names[i]: round(float(probs[i]),4) for i in range(len(class_names))}
         metrics = calculate_performance(model)
 
-        # plot image with top1 label
         top1 = max(pred_dict, key=pred_dict.get)
         buf = io.BytesIO()
         plt.figure(figsize=(3,3))
@@ -90,23 +77,20 @@ def predict_and_monitor(version, image):
         plt.savefig(buf, format='png')
         plt.close()
         buf.seek(0)
-        # Convert buffer to PIL Image for Gradio
         buf_image = Image.open(buf)
         return pred_dict, metrics, buf_image
     except Exception as e:
-        # show exception message
-        raise gr.Error(f"Tahmin hatası: {e}")
+        raise gr.Error(f"Prediction Error: {e}")
 
-# Gradio interface
 with gr.Blocks() as demo:
-    gr.Markdown("# Vbai-DPA Risk Classification & Monitoring")
+    gr.Markdown("Dementia and Parkinson Diagnosis with Vbai-DPA 2.1(f,c,q)")
     with gr.Row():
-        version = gr.Radio(['f','c','q'], value='c', label="Model Version")
-        image_in = gr.Image(type="pil", label="Brain Slice (224x224)")
+        version = gr.Radio(['f | Fastest Model','c | Classic Model','q | Quality Model'], value='c', label="Model Version")
+        image_in = gr.Image(type="pil", label="MRI or fMRI Image")
     with gr.Row():
         preds = gr.JSON(label="Prediction Probabilities")
         stats = gr.JSON(label="Performance Metrics")
-        plot = gr.Image(label="Input & Top1")
+        plot = gr.Image(label="Prediction")
     btn = gr.Button("Run")
     btn.click(fn=predict_and_monitor, inputs=[version, image_in], outputs=[preds, stats, plot])