refined the output by removing unwanted string characters from the generated dict. using a simple python slice method! :/

app.py

@@ -30,7 +30,8 @@ cap = pipeline('image-to-text')
 def predict(image):
     cap = pipeline('image-to-text')
     caption = cap(image)
-    return caption
+    output = str(caption)
+    return output[21:-4]
 
 
 input = gr.inputs.Image(label="Upload your Image and wait for 8-12 seconds!", type = 'pil', optional=False)