add interface logic

app.py

@@ -82,33 +82,51 @@ def process_image(image, task):
 def main_process(image, task):
     result = process_image(image, task)
     
-    if task in ["Object Detection", "Dense Region Caption", "Region Proposal"]:
-        fig = plot_bbox(image, result[TASK_PROMPTS[task]])
-        output_image = fig_to_pil(fig)
-    elif task == "OCR with Region":
-        output_image = draw_ocr_bboxes(image.copy(), result[TASK_PROMPTS[task]])
+    if task in ["Object Detection", "Dense Region Caption", "Region Proposal", "OCR with Region"]:
+        if task == "OCR with Region":
+            output_image = draw_ocr_bboxes(image.copy(), result[TASK_PROMPTS[task]])
+        else:
+            fig = plot_bbox(image, result[TASK_PROMPTS[task]])
+            output_image = fig_to_pil(fig)
+        return output_image, None
     else:
-        output_image = None
-    
-    return {task: str(result)}, output_image
+        return None, str(result)
+
+def reset_outputs():
+    return None, None
 
 with gr.Blocks(title="Florence-2 Demo") as iface:
     gr.Markdown("# Florence-2 Demo")
     gr.Markdown("Upload an image and select a task to process with Florence-2.")
     
-    with gr.Row():
+    with gr.Column():
         image_input = gr.Image(type="pil", label="Input Image")
         task_dropdown = gr.Dropdown(list(TASK_PROMPTS.keys()), label="Task")
     
-    submit_button = gr.Button("Process")
+    with gr.Row():
+        submit_button = gr.Button("Process")
+        reset_button = gr.Button("Reset")
     
-    output_text = gr.JSON(label="Output")
-    output_image = gr.Image(label="Processed Image")
+    output_image = gr.Image(label="Processed Image", visible=False)
+    output_text = gr.Textbox(label="Output", visible=False)
+    
+    def process_and_show(image, task):
+        img_output, text_output = main_process(image, task)
+        if img_output is not None:
+            return {output_image: gr.update(value=img_output, visible=True), output_text: gr.update(visible=False)}
+        else:
+            return {output_image: gr.update(visible=False), output_text: gr.update(value=text_output, visible=True)}
     
     submit_button.click(
-        fn=main_process,
+        fn=process_and_show,
         inputs=[image_input, task_dropdown],
-        outputs=[output_text, output_image]
+        outputs=[output_image, output_text]
+    )
+    
+    reset_button.click(
+        fn=reset_outputs,
+        inputs=[],
+        outputs=[output_image, output_text]
     )
 
 iface.launch()