Update app.py

app.py

@@ -10,23 +10,26 @@ with open("classlabel.txt", 'r') as file:
     CLASS_LABEL = [x.strip() for x in file.readlines()]
 
 def normalize_image(img):
-    img = tf.cast(img, tf.float32)/255.
+    img = tf.cast(img, tf.float32) / 255.
     img = tf.image.resize(img, (IMG_HEIGHT, IMG_WIDTH), method='bilinear')
     return img
 
-
-def predict_fn(img):
+def predict_top_classes(img, num_top_classes=5):
     img = img.convert('RGB')
     img_data = normalize_image(img)
     x = np.array(img_data)
     x = np.expand_dims(x, axis=0)
     temp = model.predict(x)
-    
-    class_index = np.argmax(temp)
-    return CLASS_LABEL[class_index]
 
+    top_class_indices = np.argpartition(temp, -num_top_classes)[-num_top_classes:]
+    top_class_indices = top_class_indices[np.argsort(temp[0, top_class_indices])[::-1]]
+
+    top_classes = [CLASS_LABEL[i] for i in top_class_indices]
+    top_probabilities = [temp[0, i] for i in top_class_indices]
+
+    return dict(zip(top_classes, top_probabilities))
 
 model = tf.keras.models.load_model("Xception.h5")
 
-interface = gr.Interface(predict_fn, gr.inputs.Image(type='pil'), outputs='label')
-interface.launch()
+interface = gr.Interface(predict_top_classes, gr.inputs.Image(type='pil'), outputs='dictionary', args={'num_top_classes': 5})
+interface.launch()