Update app.py

app.py

@@ -95,6 +95,7 @@ def sepia(input_img):
     fig = draw_plot(pred_img, seg)
 
     # 각 물체에 대한 예측 클래스와 확률 얻기
+
     unique_labels = np.unique(seg.numpy().astype("uint8"))
     class_probabilities = {}
     for label in unique_labels:
@@ -102,12 +103,13 @@ def sepia(input_img):
         class_name = labels_list[label]
         class_prob = tf.nn.softmax(logits.numpy()[0][:, :, label])  # softmax 적용
         class_prob = np.mean(class_prob[mask])
-        class_probabilities[class_name] = class_prob
+        class_probabilities[class_name] = class_prob * 100  # 백분율로 변환
 
     # Gradio Interface에 출력할 문자열 생성
     output_text = "Predicted class probabilities:\n"
     for class_name, prob in class_probabilities.items():
-        output_text += f"{class_name}: {prob:.4f}\n"
+        output_text += f"{class_name}: {prob:.2f}%\n"
+
 
     # 정확성이 가장 높은 물체 정보 출력
     max_prob_class = max(class_probabilities, key=class_probabilities.get)