Update app.py

app.py

@@ -12,6 +12,9 @@ logger = logging.getLogger(__name__)
 model = ViTForImageClassification.from_pretrained("prithivMLmods/Deep-Fake-Detector-Model")
 processor = ViTImageProcessor.from_pretrained("prithivMLmods/Deep-Fake-Detector-Model")
 
+# Log model configuration to verify label mapping
+logger.info(f"Model label mapping: {model.config.id2label}")
+
 def detect(image, confidence_threshold=0.5):
     """Detect deepfake content using prithivMLmods/Deep-Fake-Detector-Model"""
     if image is None:
@@ -34,18 +37,26 @@ def detect(image, confidence_threshold=0.5):
             probabilities = torch.softmax(logits, dim=1)[0]
         
         # Get confidence scores
-        confidence_real = probabilities[0].item() * 100
-        confidence_fake = probabilities[1].item() * 100
+        confidence_real = probabilities[0].item() * 100  # Assuming 0 is Real
+        confidence_fake = probabilities[1].item() * 100  # Assuming 1 is Fake
+        
+        # Verify label mapping from model config
+        id2label = model.config.id2label
+        predicted_class = torch.argmax(logits, dim=1).item()
+        predicted_label = id2label[predicted_class]
         
-        # Determine prediction based on threshold
-        predicted_label = "Fake" if confidence_fake / 100 >= confidence_threshold else "Real"
+        # Adjust prediction based on threshold and label
+        threshold_predicted = "Fake" if confidence_fake / 100 >= confidence_threshold else "Real"
         confidence_score = max(confidence_real, confidence_fake)
         
-        # Log the prediction
-        logger.info(f"Real: {confidence_real:.1f}%, Fake: {confidence_fake:.1f}%, Predicted: {predicted_label}")
+        # Log detailed output
+        logger.info(f"Logits: {logits.tolist()}")
+        logger.info(f"Probabilities - Real: {confidence_real:.1f}%, Fake: {confidence_fake:.1f}%")
+        logger.info(f"Predicted Class: {predicted_class}, Label: {predicted_label}")
+        logger.info(f"Threshold ({confidence_threshold}): {threshold_predicted}")
         
         # Prepare output
-        overall = f"{confidence_score:.1f}% Confidence ({predicted_label})"
+        overall = f"{confidence_score:.1f}% Confidence ({threshold_predicted})"
         aigen = f"{confidence_fake:.1f}% (AI-Generated Content Likelihood)"
         deepfake = f"{confidence_fake:.1f}% (Face Manipulation Likelihood)"
         
@@ -98,7 +109,7 @@ MARKDOWN0 = """
     <h1>DeepFake Detection System</h1>
     <p>Advanced AI-powered analysis for identifying manipulated media<br>
     Powered by prithivMLmods/Deep-Fake-Detector-Model (Updated Jan 2025)<br>
-    Adjustable threshold for tuning detection sensitivity</p>
+    Adjust threshold to tune sensitivity; check logs for detailed output</p>
 </div>
 """