Update app.py

app.py

@@ -502,7 +502,7 @@ def concatenate_images_vertical(image1, image2):
 mask_source_draw = "draw a mask on input image"
 mask_source_segment = "type what to detect below"
 
-def run_anything_task(secret_token, input_image, text_prompt, box_threshold, text_threshold, 
+def run_anything_task(secret_token, input_image_b64, text_prompt, box_threshold, text_threshold, 
             iou_threshold, cleaner_size_limit=1080):
     if secret_token != SECRET_TOKEN:
         raise gr.Error(
@@ -513,28 +513,20 @@ def run_anything_task(secret_token, input_image, text_prompt, box_threshold, tex
     text_prompt = text_prompt.strip()
     if not ((task_type == 'inpainting' or task_type == 'remove') and mask_source_radio == mask_source_draw):
         if text_prompt == '':
-            return [], gr.Gallery.update(label='Detection prompt is not found!😂😂😂😂')
+            return ""
 
-    if input_image is None:
-            return [], gr.Gallery.update(label='Please upload a image!😂😂😂😂')
+    if input_image_b64 is None:
+            return ""
 
     file_temp = int(time.time())
 
     output_images = []
 
     # load image
-
-    if isinstance(input_image, dict):
-        image_pil, image = load_image(input_image['image'].convert("RGB"))
-        input_img = input_image['image']
-        # no need to print back the input image
-        # output_images.append(input_image['image'])
-    else:
-        image_pil, image = load_image(input_image.convert("RGB"))
-        input_img = input_image
-        # no need to print back the input image
-        # output_images.append(input_image)
-
+    input_image = readb64(input_image_b64)
+    
+    image_pil, image = load_image(input_image.convert("RGB"))
+ 
     size = image_pil.size
     
     # run grounding dino model
@@ -565,7 +557,7 @@ def run_anything_task(secret_token, input_image, text_prompt, box_threshold, tex
     # output_images.append(image_with_box)
 
     # now we generate the segmentation
-    image = np.array(input_img)
+    image = np.array(input_image)
     sam_predictor.set_image(image)
 
     H, W = size[1], size[0]
@@ -658,7 +650,7 @@ if __name__ == "__main__":
 
         secret_token = gr.Textbox()
         text_prompt = gr.Textbox()
-        input_image = gr.Textbox()    
+        input_image_b64 = gr.Textbox()    
         text_prompt = gr.Textbox(label="Detection Prompt[To detect multiple objects, seperating each name with '.', like this: cat . dog . chair ]", placeholder="Cannot be empty")                                                
         run_button = gr.Button(label="Run", visible=True)
         with gr.Accordion("Advanced options", open=False) as advanced_options:
@@ -676,7 +668,7 @@ if __name__ == "__main__":
                 fn=run_anything_task,
                 inputs=[
                     secret_token,
-                    input_image,
+                    input_image_b64,
                     text_prompt,
                     box_threshold,
                     text_threshold,