diff --git a/demo/remote_request.py b/demo/remote_request.py
new file mode 100644
index 0000000..d564096
--- /dev/null
+++ b/demo/remote_request.py
@@ -0,0 +1,69 @@
+# uvicorn test:app --host 0.0.0.0 --port 8000 --reload
+from fastapi import FastAPI
+from pydantic import BaseModel
+
+print('hello')
+app = FastAPI()
+
+class Item(BaseModel):
+    base64_image: str
+    prompt: str
+    # goal: str
+    # idx: int
+    # step_logs: str
+
+@app.post("/send_text/")
+async def send_text(item: Item):
+    print(item.prompt)
+    # print(item.base64_image)
+    return {"text": item.prompt}
+
+
+
+from utils import get_som_labeled_img, check_ocr_box, get_caption_model_processor,  get_dino_model, get_yolo_model
+import torch
+from ultralytics import YOLO
+from PIL import Image
+from typing import Dict, Tuple, List
+import io
+import base64
+
+
+config = {
+    'som_model_path': 'finetuned_icon_detect.pt',
+    'device': 'cpu',
+    'caption_model_path': 'Salesforce/blip2-opt-2.7b',
+    'draw_bbox_config': {
+        'text_scale': 0.8,
+        'text_thickness': 2,
+        'text_padding': 3,
+        'thickness': 3,
+    },
+    'BOX_TRESHOLD': 0.05
+}
+
+
+class Omniparser(object):
+    def __init__(self, config: Dict):
+        self.config = config
+        
+        self.som_model = get_yolo_model(model_path=config['som_model_path'])
+        device='gpu' if torch.cuda.is_available() else 'cpu'
+        self.caption_model_processor = get_caption_model_processor(model_name="florence2", model_name_or_path="weights/icon_caption_florence", device=device)
+
+    def parse(self, image_path: str):
+        print('Parsing image:', image_path)
+        ocr_bbox_rslt, is_goal_filtered = check_ocr_box(image_path, display_img = False, output_bb_format='xyxy', goal_filtering=None, easyocr_args={'paragraph': False, 'text_threshold':0.5}, use_paddleocr=True)
+        text, ocr_bbox = ocr_bbox_rslt
+        dino_labled_img, label_coordinates, parsed_content_list = get_som_labeled_img(image_path, self.som_model, BOX_TRESHOLD = BOX_TRESHOLD, output_coord_in_ratio=True, ocr_bbox=ocr_bbox,draw_bbox_config=draw_bbox_config, caption_model_processor=self.caption_model_processor, ocr_text=text,use_local_semantics=True, iou_threshold=0.7, scale_img=False, batch_size=128)
+
+        image = Image.open(io.BytesIO(base64.b64decode(dino_labled_img)))
+        # formating output
+        return_list = [{'from': 'omniparser', 'shape': {'x':coord[0], 'y':coord[1], 'width':coord[2], 'height':coord[3]},
+                        'text': parsed_content_list[i].split(': ')[1], 'type':'text'} for i, (k, coord) in enumerate(label_coordinates.items()) if i < len(parsed_content_list)]
+        return_list.extend(
+            [{'from': 'omniparser', 'shape': {'x':coord[0], 'y':coord[1], 'width':coord[2], 'height':coord[3]},
+                        'text': 'None', 'type':'icon'} for i, (k, coord) in enumerate(label_coordinates.items()) if i >= len(parsed_content_list)]
+              )
+
+        return [image, return_list]
\ No newline at end of file