diff --git a/converters/to_onnx.py b/converters/to_onnx.py
index 15b71a0..b8268ba 100644
--- a/converters/to_onnx.py
+++ b/converters/to_onnx.py
@@ -98,10 +98,18 @@ def main() -> None:
         required=True,
     )
 
+    arg(
+        "-i",
+        "--image_path",
+        type=int,
+        help="The path of the input image",
+        required=True,
+    )
+
     arg("-o", "--output_file", type=str, help="Path to save onnx model.", required=True)
     args = parser.parse_args()
 
-    raw_image = cv2.imread("tests/data/13.jpg")
+    raw_image = cv2.imread(args.image_path)
 
     image = prepare_image(raw_image, args.max_size)
 
@@ -134,18 +142,17 @@ def main() -> None:
         if not np.allclose(out_torch[i].numpy(), outputs[i]):
             raise ValueError("torch and onnx models do not match!")
 
-    annotations: List[Dict[str, List[Union[float, List[float]]]]] = []
+    annotations: List[Dict[str, List[Union[float, List[float]]]]] = [
+        {
+            "bbox": box.tolist(),
+            "score": outputs[1][box_id],
+            "landmarks": outputs[2][box_id].reshape(-1, 2).tolist(),
+        }
+        for box_id, box in enumerate(outputs[0])
+    ]
 
-    for box_id, box in enumerate(outputs[0]):
-        annotations += [
-            {
-                "bbox": box.tolist(),
-                "score": outputs[1][box_id],
-                "landmarks": outputs[2][box_id].reshape(-1, 2).tolist(),
-            }
-        ]
 
-    im = albu.Compose([albu.LongestMaxSize(max_size=1280)])(image=raw_image)["image"]
+    im = albu.Compose([albu.LongestMaxSize(max_size=args.max_size)])(image=raw_image)["image"]
     cv2.imwrite("example.jpg", vis_annotations(im, annotations))