diff --git a/paddleseg/core/export.py b/paddleseg/core/export.py
index 839f22980..586f35e93 100644
--- a/paddleseg/core/export.py
+++ b/paddleseg/core/export.py
@@ -49,8 +49,12 @@ def export(args, model=None, save_dir=None, use_ema=False):
         save_dir = args.save_dir
     os.environ['PADDLESEG_EXPORT_STAGE'] = 'True'
 
-    shape = [None, 3, None, None] if args.input_shape is None \
-        else args.input_shape
+    if args.input_shape is None:
+        assert cfg.model_cfg['type'] not in ['MaskFormer'], \
+            "Please specify `--input_shape` when exporting MaskFormer models!"
+        shape = [None, 3, None, None]
+    else:
+        shape = args.input_shape
     input_spec = [paddle.static.InputSpec(shape=shape, dtype='float32')]
     model.eval()
     model = paddle.jit.to_static(model, input_spec=input_spec)
@@ -79,6 +83,19 @@ def export(args, model=None, save_dir=None, use_ema=False):
     val_dataset_cfg = cfg.val_dataset_cfg
     assert val_dataset_cfg != {}, 'No val_dataset specified in the configuration file.'
     transforms = val_dataset_cfg.get('transforms', None)
+    if cfg.model_cfg["type"] in ["MaskFormer"]:
+        resize_trans = {
+            "type": "Resize",
+            "target_size": shape[-2:],
+            "keep_ratio": False,
+            "interp": "NEAREST"}
+        if transforms is None:
+            transforms = [resize_trans]
+        else:
+            for t in transforms:
+                if t["type"] in ["Resize", "ResizeByShort", "ResizeByLong"]:
+                    t.clear()
+                    t.update(resize_trans)
     output_dtype = 'int32' if args.output_op == 'argmax' else 'float32'
 
     # TODO add test config