Update app.py

app.py

@@ -158,7 +158,7 @@ def downsample_video(video_path):
     """
     vidcap = cv2.VideoCapture(video_path)
     total_frames = int(vidcap.get(cv2.CAP_PROP_FRAME_COUNT))
-    fps = vidcap.get(cv2.CAP_PROP_FPS)
+    fps    fps = vidcap.get(cv2.CAP_PROP_FPS)
     frames = []
     frame_indices = np.linspace(0, total_frames - 1, 10, dtype=int)
     for i in frame_indices:
@@ -228,7 +228,9 @@ def generate_image(model_name: str, text: str, image: Image.Image,
         time.sleep(0.01)
         yield buffer, buffer
 
-@spaces.GPU
+@spaces.G “
+
+PU
 def generate_video(model_name: str, text: str, video_path: str,
                    max_new_tokens: int = 1024,
                    temperature: float = 0.6,
@@ -309,6 +311,12 @@ video_examples = [
     ["explain the video in detail.", "videos/2.mp4"]
 ]
 
+# Define examples for object detection
+object_detection_examples = [
+    ["Detect Spider-Man T-shirt.", "images/22.png"],
+    ["Detect Green Car.", "images/11.png"]
+]
+
 # Added CSS to style the output area as a "Canvas"
 css = """
 .submit-btn {
@@ -354,6 +362,10 @@ with gr.Blocks(css=css, theme="bethecloud/storj_theme") as demo:
                             system_prompt = gr.Textbox(label="System Prompt", value=default_system_prompt, visible=False)
                             text_input = gr.Textbox(label="Query Input", value="Detect animal")
                             submit_btn = gr.Button(value="Submit", elem_classes="submit-btn")
+                            gr.Examples(
+                                examples=object_detection_examples,
+                                inputs=[text_input, input_img]
+                            )
                         with gr.Column():
                             model_output_text = gr.Textbox(label="Model Output Text")
                             parsed_boxes = gr.Textbox(label="Parsed Boxes")