Update app.py

app.py

@@ -7,6 +7,7 @@ import PIL.Image
 import spaces
 import torch
 from transformers import AutoProcessor, BlipForConditionalGeneration
+from typing import Union
 
 DESCRIPTION = "# Image Captioning with LongCap"
 
@@ -16,9 +17,18 @@ model_id = "unography/blip-long-cap"
 processor = AutoProcessor.from_pretrained(model_id)
 model = BlipForConditionalGeneration.from_pretrained(model_id).to(device)
 
+torch.hub.download_url_to_file("http://images.cocodataset.org/val2017/000000039769.jpg", "cats.jpg")
+torch.hub.download_url_to_file(
+    "https://huggingface.co/datasets/nielsr/textcaps-sample/resolve/main/stop_sign.png", "stop_sign.png"
+)
+torch.hub.download_url_to_file(
+    "https://cdn.openai.com/dall-e-2/demos/text2im/astronaut/horse/photo/0.jpg", "astronaut.jpg"
+)
 
 @spaces.GPU(duration=30)
-def run(image: PIL.Image.Image) -> str:
+def run(image: Union[str, PIL.Image.Image]) -> str:
+    if isinstance(image, str):
+        image = Image.open(image)
     inputs = processor(images=image, return_tensors="pt").to(device)
     out = model.generate(pixel_values=inputs.pixel_values, num_beams=3, max_length=300)
     generated_caption = processor.decode(out[0], skip_special_tokens=True)
@@ -30,6 +40,17 @@ with gr.Blocks(css="style.css") as demo:
     input_image = gr.Image(type="pil")
     run_button = gr.Button("Caption")
     output = gr.Textbox(label="Result")
+    gr.Examples(
+        examples=[
+            "cats.jpg",
+            "stop_sign.png",
+            "astronaut.jpg",
+        ],
+        inputs=input_image,
+        outputs=output,
+        fn=run,
+        cache_examples=os.getenv("CACHE_EXAMPLES") == "1",
+    )
 
     run_button.click(
         fn=run,